cuda-vs-cpu 0.9999947547912598
{
  "embeddings.patch_embedding": {
    "max": 1.0,
    "over15": 0.0,
    "weightMax": 0.18357841670513153
  },
  "encoder.layers.0.self_attn.q_proj": {
    "max": 9.544836044311523,
    "over15": 0.0,
    "weightMax": 0.4607301950454712
  },
  "encoder.layers.0.self_attn.k_proj": {
    "max": 9.544836044311523,
    "over15": 0.0,
    "weightMax": 0.7309656739234924
  },
  "encoder.layers.0.self_attn.v_proj": {
    "max": 9.544836044311523,
    "over15": 0.0,
    "weightMax": 0.343400239944458
  },
  "encoder.layers.0.self_attn.out_proj": {
    "max": 2.5156710147857666,
    "over15": 0.0,
    "weightMax": 0.42120611667633057
  },
  "encoder.layers.0.mlp.fc1": {
    "max": 17.17769432067871,
    "over15": 1.4288980310084298e-05,
    "weightMax": 0.54178386926651
  },
  "encoder.layers.0.mlp.fc2": {
    "max": 10.184268951416016,
    "over15": 0.0,
    "weightMax": 0.48006734251976013
  },
  "encoder.layers.1.self_attn.q_proj": {
    "max": 11.57963752746582,
    "over15": 0.0,
    "weightMax": 0.3532828688621521
  },
  "encoder.layers.1.self_attn.k_proj": {
    "max": 11.57963752746582,
    "over15": 0.0,
    "weightMax": 0.2644633650779724
  },
  "encoder.layers.1.self_attn.v_proj": {
    "max": 11.57963752746582,
    "over15": 0.0,
    "weightMax": 0.3203193247318268
  },
  "encoder.layers.1.self_attn.out_proj": {
    "max": 2.9277071952819824,
    "over15": 0.0,
    "weightMax": 0.43071871995925903
  },
  "encoder.layers.1.mlp.fc1": {
    "max": 25.10468101501465,
    "over15": 0.00014288979582488537,
    "weightMax": 0.39755430817604065
  },
  "encoder.layers.1.mlp.fc2": {
    "max": 7.348878860473633,
    "over15": 0.0,
    "weightMax": 0.41356244683265686
  },
  "encoder.layers.2.self_attn.q_proj": {
    "max": 14.80830192565918,
    "over15": 0.0,
    "weightMax": 0.3900470733642578
  },
  "encoder.layers.2.self_attn.k_proj": {
    "max": 14.80830192565918,
    "over15": 0.0,
    "weightMax": 0.2604505717754364
  },
  "encoder.layers.2.self_attn.v_proj": {
    "max": 14.80830192565918,
    "over15": 0.0,
    "weightMax": 0.24337667226791382
  },
  "encoder.layers.2.self_attn.out_proj": {
    "max": 3.036842107772827,
    "over15": 0.0,
    "weightMax": 0.5740410685539246
  },
  "encoder.layers.2.mlp.fc1": {
    "max": 14.494551658630371,
    "over15": 0.0,
    "weightMax": 0.4425497055053711
  },
  "encoder.layers.2.mlp.fc2": {
    "max": 7.965270042419434,
    "over15": 0.0,
    "weightMax": 0.5819864273071289
  },
  "encoder.layers.3.self_attn.q_proj": {
    "max": 18.652353286743164,
    "over15": 1.9051973140449263e-05,
    "weightMax": 0.35118240118026733
  },
  "encoder.layers.3.self_attn.k_proj": {
    "max": 18.652353286743164,
    "over15": 1.9051973140449263e-05,
    "weightMax": 0.3851613998413086
  },
  "encoder.layers.3.self_attn.v_proj": {
    "max": 18.652353286743164,
    "over15": 1.9051973140449263e-05,
    "weightMax": 0.2509399950504303
  },
  "encoder.layers.3.self_attn.out_proj": {
    "max": 2.624849557876587,
    "over15": 0.0,
    "weightMax": 0.3587561249732971
  },
  "encoder.layers.3.mlp.fc1": {
    "max": 16.160676956176758,
    "over15": 2.381496642556158e-06,
    "weightMax": 0.3124311864376068
  },
  "encoder.layers.3.mlp.fc2": {
    "max": 6.110797882080078,
    "over15": 0.0,
    "weightMax": 0.3342714011669159
  },
  "encoder.layers.4.self_attn.q_proj": {
    "max": 23.907943725585938,
    "over15": 1.0716735232563224e-05,
    "weightMax": 0.25488975644111633
  },
  "encoder.layers.4.self_attn.k_proj": {
    "max": 23.907943725585938,
    "over15": 1.0716735232563224e-05,
    "weightMax": 0.3293153643608093
  },
  "encoder.layers.4.self_attn.v_proj": {
    "max": 23.907943725585938,
    "over15": 1.0716735232563224e-05,
    "weightMax": 0.29178351163864136
  },
  "encoder.layers.4.self_attn.out_proj": {
    "max": 2.656275510787964,
    "over15": 0.0,
    "weightMax": 0.20805060863494873
  },
  "encoder.layers.4.mlp.fc1": {
    "max": 12.990350723266602,
    "over15": 0.0,
    "weightMax": 0.41922831535339355
  },
  "encoder.layers.4.mlp.fc2": {
    "max": 6.064598083496094,
    "over15": 0.0,
    "weightMax": 0.4641551673412323
  },
  "encoder.layers.5.self_attn.q_proj": {
    "max": 21.269189834594727,
    "over15": 2.6196463295491412e-05,
    "weightMax": 0.19247844815254211
  },
  "encoder.layers.5.self_attn.k_proj": {
    "max": 21.269189834594727,
    "over15": 2.6196463295491412e-05,
    "weightMax": 0.23010361194610596
  },
  "encoder.layers.5.self_attn.v_proj": {
    "max": 21.269189834594727,
    "over15": 2.6196463295491412e-05,
    "weightMax": 0.235695019364357
  },
  "encoder.layers.5.self_attn.out_proj": {
    "max": 2.2667412757873535,
    "over15": 0.0,
    "weightMax": 0.22834673523902893
  },
  "encoder.layers.5.mlp.fc1": {
    "max": 13.730095863342285,
    "over15": 0.0,
    "weightMax": 0.4146582782268524
  },
  "encoder.layers.5.mlp.fc2": {
    "max": 4.032431602478027,
    "over15": 0.0,
    "weightMax": 0.3833647668361664
  },
  "encoder.layers.6.self_attn.q_proj": {
    "max": 25.1704044342041,
    "over15": 1.667047581577208e-05,
    "weightMax": 0.2430504560470581
  },
  "encoder.layers.6.self_attn.k_proj": {
    "max": 25.1704044342041,
    "over15": 1.667047581577208e-05,
    "weightMax": 0.2625119090080261
  },
  "encoder.layers.6.self_attn.v_proj": {
    "max": 25.1704044342041,
    "over15": 1.667047581577208e-05,
    "weightMax": 0.32204073667526245
  },
  "encoder.layers.6.self_attn.out_proj": {
    "max": 4.367775917053223,
    "over15": 0.0,
    "weightMax": 0.6605210900306702
  },
  "encoder.layers.6.mlp.fc1": {
    "max": 14.761872291564941,
    "over15": 0.0,
    "weightMax": 0.35002678632736206
  },
  "encoder.layers.6.mlp.fc2": {
    "max": 4.258421421051025,
    "over15": 0.0,
    "weightMax": 0.5761503577232361
  },
  "encoder.layers.7.self_attn.q_proj": {
    "max": 30.813331604003906,
    "over15": 8.097088721115142e-05,
    "weightMax": 0.29404881596565247
  },
  "encoder.layers.7.self_attn.k_proj": {
    "max": 30.813331604003906,
    "over15": 8.097088721115142e-05,
    "weightMax": 0.28991928696632385
  },
  "encoder.layers.7.self_attn.v_proj": {
    "max": 30.813331604003906,
    "over15": 8.097088721115142e-05,
    "weightMax": 0.16781431436538696
  },
  "encoder.layers.7.self_attn.out_proj": {
    "max": 2.633110284805298,
    "over15": 0.0,
    "weightMax": 0.32742953300476074
  },
  "encoder.layers.7.mlp.fc1": {
    "max": 13.183366775512695,
    "over15": 0.0,
    "weightMax": 0.45574870705604553
  },
  "encoder.layers.7.mlp.fc2": {
    "max": 5.059280872344971,
    "over15": 0.0,
    "weightMax": 0.38412055373191833
  },
  "encoder.layers.8.self_attn.q_proj": {
    "max": 23.06891441345215,
    "over15": 1.667047581577208e-05,
    "weightMax": 0.23720932006835938
  },
  "encoder.layers.8.self_attn.k_proj": {
    "max": 23.06891441345215,
    "over15": 1.667047581577208e-05,
    "weightMax": 0.3083343803882599
  },
  "encoder.layers.8.self_attn.v_proj": {
    "max": 23.06891441345215,
    "over15": 1.667047581577208e-05,
    "weightMax": 0.2643316686153412
  },
  "encoder.layers.8.self_attn.out_proj": {
    "max": 2.812100887298584,
    "over15": 0.0,
    "weightMax": 0.5204288959503174
  },
  "encoder.layers.8.mlp.fc1": {
    "max": 20.975492477416992,
    "over15": 1.786122447811067e-05,
    "weightMax": 0.3390844464302063
  },
  "encoder.layers.8.mlp.fc2": {
    "max": 5.3550920486450195,
    "over15": 0.0,
    "weightMax": 0.4186340570449829
  },
  "encoder.layers.9.self_attn.q_proj": {
    "max": 24.337251663208008,
    "over15": 5.953741492703557e-05,
    "weightMax": 0.29854971170425415
  },
  "encoder.layers.9.self_attn.k_proj": {
    "max": 24.337251663208008,
    "over15": 5.953741492703557e-05,
    "weightMax": 0.3360884487628937
  },
  "encoder.layers.9.self_attn.v_proj": {
    "max": 24.337251663208008,
    "over15": 5.953741492703557e-05,
    "weightMax": 0.13019117712974548
  },
  "encoder.layers.9.self_attn.out_proj": {
    "max": 3.067563056945801,
    "over15": 0.0,
    "weightMax": 0.24109724164009094
  },
  "encoder.layers.9.mlp.fc1": {
    "max": 23.031423568725586,
    "over15": 1.667047581577208e-05,
    "weightMax": 0.30425572395324707
  },
  "encoder.layers.9.mlp.fc2": {
    "max": 6.279391288757324,
    "over15": 0.0,
    "weightMax": 0.35977089405059814
  },
  "encoder.layers.10.self_attn.q_proj": {
    "max": 22.78374671936035,
    "over15": 7.620789256179705e-05,
    "weightMax": 0.257494181394577
  },
  "encoder.layers.10.self_attn.k_proj": {
    "max": 22.78374671936035,
    "over15": 7.620789256179705e-05,
    "weightMax": 0.25740620493888855
  },
  "encoder.layers.10.self_attn.v_proj": {
    "max": 22.78374671936035,
    "over15": 7.620789256179705e-05,
    "weightMax": 0.23113733530044556
  },
  "encoder.layers.10.self_attn.out_proj": {
    "max": 5.999870777130127,
    "over15": 0.0,
    "weightMax": 0.31268030405044556
  },
  "encoder.layers.10.mlp.fc1": {
    "max": 35.75727844238281,
    "over15": 2.6196463295491412e-05,
    "weightMax": 0.35464808344841003
  },
  "encoder.layers.10.mlp.fc2": {
    "max": 7.481848239898682,
    "over15": 0.0,
    "weightMax": 0.3893696069717407
  },
  "encoder.layers.11.self_attn.q_proj": {
    "max": 21.531110763549805,
    "over15": 2.8577960620168597e-05,
    "weightMax": 0.2845437228679657
  },
  "encoder.layers.11.self_attn.k_proj": {
    "max": 21.531110763549805,
    "over15": 2.8577960620168597e-05,
    "weightMax": 0.24490857124328613
  },
  "encoder.layers.11.self_attn.v_proj": {
    "max": 21.531110763549805,
    "over15": 2.8577960620168597e-05,
    "weightMax": 0.20706063508987427
  },
  "encoder.layers.11.self_attn.out_proj": {
    "max": 3.6929852962493896,
    "over15": 0.0,
    "weightMax": 0.26636624336242676
  },
  "encoder.layers.11.mlp.fc1": {
    "max": 45.94890594482422,
    "over15": 2.6196463295491412e-05,
    "weightMax": 0.3664744794368744
  },
  "encoder.layers.11.mlp.fc2": {
    "max": 4.171147346496582,
    "over15": 0.0,
    "weightMax": 0.42437535524368286
  },
  "encoder.layers.12.self_attn.q_proj": {
    "max": 24.808061599731445,
    "over15": 7.739863940514624e-05,
    "weightMax": 0.22305254638195038
  },
  "encoder.layers.12.self_attn.k_proj": {
    "max": 24.808061599731445,
    "over15": 7.739863940514624e-05,
    "weightMax": 0.24658657610416412
  },
  "encoder.layers.12.self_attn.v_proj": {
    "max": 24.808061599731445,
    "over15": 7.739863940514624e-05,
    "weightMax": 0.1596279740333557
  },
  "encoder.layers.12.self_attn.out_proj": {
    "max": 4.1753644943237305,
    "over15": 0.0,
    "weightMax": 0.22394107282161713
  },
  "encoder.layers.12.mlp.fc1": {
    "max": 26.96554183959961,
    "over15": 3.9294693124247715e-05,
    "weightMax": 0.6758708357810974
  },
  "encoder.layers.12.mlp.fc2": {
    "max": 5.359699249267578,
    "over15": 0.0,
    "weightMax": 0.4833727478981018
  },
  "encoder.layers.13.self_attn.q_proj": {
    "max": 24.9854736328125,
    "over15": 7.144489791244268e-05,
    "weightMax": 0.1897425353527069
  },
  "encoder.layers.13.self_attn.k_proj": {
    "max": 24.9854736328125,
    "over15": 7.144489791244268e-05,
    "weightMax": 0.22280113399028778
  },
  "encoder.layers.13.self_attn.v_proj": {
    "max": 24.9854736328125,
    "over15": 7.144489791244268e-05,
    "weightMax": 0.19644315540790558
  },
  "encoder.layers.13.self_attn.out_proj": {
    "max": 6.095090866088867,
    "over15": 0.0,
    "weightMax": 0.2909233272075653
  },
  "encoder.layers.13.mlp.fc1": {
    "max": 78.18529510498047,
    "over15": 0.00017146776372101158,
    "weightMax": 0.4890219569206238
  },
  "encoder.layers.13.mlp.fc2": {
    "max": 4.305015563964844,
    "over15": 0.0,
    "weightMax": 0.4814220368862152
  },
  "encoder.layers.14.self_attn.q_proj": {
    "max": 22.99729347229004,
    "over15": 5.953741492703557e-05,
    "weightMax": 0.2186058759689331
  },
  "encoder.layers.14.self_attn.k_proj": {
    "max": 22.99729347229004,
    "over15": 5.953741492703557e-05,
    "weightMax": 0.22827139496803284
  },
  "encoder.layers.14.self_attn.v_proj": {
    "max": 22.99729347229004,
    "over15": 5.953741492703557e-05,
    "weightMax": 0.19762998819351196
  },
  "encoder.layers.14.self_attn.out_proj": {
    "max": 7.9565253257751465,
    "over15": 0.0,
    "weightMax": 0.30530238151550293
  },
  "encoder.layers.14.mlp.fc1": {
    "max": 72.69683837890625,
    "over15": 3.572244895622134e-05,
    "weightMax": 0.30597978830337524
  },
  "encoder.layers.14.mlp.fc2": {
    "max": 4.649754524230957,
    "over15": 0.0,
    "weightMax": 0.5815067887306213
  },
  "encoder.layers.15.self_attn.q_proj": {
    "max": 24.377376556396484,
    "over15": 6.549115641973913e-05,
    "weightMax": 0.21377991139888763
  },
  "encoder.layers.15.self_attn.k_proj": {
    "max": 24.377376556396484,
    "over15": 6.549115641973913e-05,
    "weightMax": 0.23913982510566711
  },
  "encoder.layers.15.self_attn.v_proj": {
    "max": 24.377376556396484,
    "over15": 6.549115641973913e-05,
    "weightMax": 0.18743984401226044
  },
  "encoder.layers.15.self_attn.out_proj": {
    "max": 3.439500331878662,
    "over15": 0.0,
    "weightMax": 0.21996630728244781
  },
  "encoder.layers.15.mlp.fc1": {
    "max": 33.50436019897461,
    "over15": 1.547972897242289e-05,
    "weightMax": 0.5088617205619812
  },
  "encoder.layers.15.mlp.fc2": {
    "max": 7.529631614685059,
    "over15": 0.0,
    "weightMax": 0.8114994764328003
  },
  "encoder.layers.16.self_attn.q_proj": {
    "max": 27.468360900878906,
    "over15": 6.310966273304075e-05,
    "weightMax": 0.23381294310092926
  },
  "encoder.layers.16.self_attn.k_proj": {
    "max": 27.468360900878906,
    "over15": 6.310966273304075e-05,
    "weightMax": 0.2359100580215454
  },
  "encoder.layers.16.self_attn.v_proj": {
    "max": 27.468360900878906,
    "over15": 6.310966273304075e-05,
    "weightMax": 0.18034087121486664
  },
  "encoder.layers.16.self_attn.out_proj": {
    "max": 3.2610812187194824,
    "over15": 0.0,
    "weightMax": 0.17228947579860687
  },
  "encoder.layers.16.mlp.fc1": {
    "max": 48.673606872558594,
    "over15": 1.547972897242289e-05,
    "weightMax": 0.37315821647644043
  },
  "encoder.layers.16.mlp.fc2": {
    "max": 22.491432189941406,
    "over15": 9.561399565427564e-07,
    "weightMax": 0.5947832465171814
  },
  "encoder.layers.17.self_attn.q_proj": {
    "max": 30.747997283935547,
    "over15": 0.0001262193254660815,
    "weightMax": 0.1854780614376068
  },
  "encoder.layers.17.self_attn.k_proj": {
    "max": 30.747997283935547,
    "over15": 0.0001262193254660815,
    "weightMax": 0.26729005575180054
  },
  "encoder.layers.17.self_attn.v_proj": {
    "max": 30.747997283935547,
    "over15": 0.0001262193254660815,
    "weightMax": 0.1460886150598526
  },
  "encoder.layers.17.self_attn.out_proj": {
    "max": 4.202603816986084,
    "over15": 0.0,
    "weightMax": 0.13997997343540192
  },
  "encoder.layers.17.mlp.fc1": {
    "max": 57.3701286315918,
    "over15": 0.0029625818133354187,
    "weightMax": 0.762123703956604
  },
  "encoder.layers.17.mlp.fc2": {
    "max": 57.30663299560547,
    "over15": 3.1871331884758547e-07,
    "weightMax": 0.5548782348632812
  },
  "encoder.layers.18.self_attn.q_proj": {
    "max": 33.207801818847656,
    "over15": 8.573388186050579e-05,
    "weightMax": 0.20967818796634674
  },
  "encoder.layers.18.self_attn.k_proj": {
    "max": 33.207801818847656,
    "over15": 8.573388186050579e-05,
    "weightMax": 0.16187238693237305
  },
  "encoder.layers.18.self_attn.v_proj": {
    "max": 33.207801818847656,
    "over15": 8.573388186050579e-05,
    "weightMax": 0.1668742597103119
  },
  "encoder.layers.18.self_attn.out_proj": {
    "max": 4.395178318023682,
    "over15": 0.0,
    "weightMax": 0.2865675091743469
  },
  "encoder.layers.18.mlp.fc1": {
    "max": 52.14497375488281,
    "over15": 0.004883259069174528,
    "weightMax": 0.41489166021347046
  },
  "encoder.layers.18.mlp.fc2": {
    "max": 14.401910781860352,
    "over15": 0.0,
    "weightMax": 0.34858039021492004
  },
  "encoder.layers.19.self_attn.q_proj": {
    "max": 31.560243606567383,
    "over15": 9.64506107266061e-05,
    "weightMax": 0.19383127987384796
  },
  "encoder.layers.19.self_attn.k_proj": {
    "max": 31.560243606567383,
    "over15": 9.64506107266061e-05,
    "weightMax": 0.19668807089328766
  },
  "encoder.layers.19.self_attn.v_proj": {
    "max": 31.560243606567383,
    "over15": 9.64506107266061e-05,
    "weightMax": 0.132775217294693
  },
  "encoder.layers.19.self_attn.out_proj": {
    "max": 4.554391384124756,
    "over15": 0.0,
    "weightMax": 0.2345474660396576
  },
  "encoder.layers.19.mlp.fc1": {
    "max": 62.2261848449707,
    "over15": 0.024686593562364578,
    "weightMax": 0.3576207756996155
  },
  "encoder.layers.19.mlp.fc2": {
    "max": 15.347545623779297,
    "over15": 3.1871331884758547e-07,
    "weightMax": 0.3363354206085205
  },
  "encoder.layers.20.self_attn.q_proj": {
    "max": 28.272241592407227,
    "over15": 0.00010240435221930966,
    "weightMax": 0.1938808560371399
  },
  "encoder.layers.20.self_attn.k_proj": {
    "max": 28.272241592407227,
    "over15": 0.00010240435221930966,
    "weightMax": 0.2253892719745636
  },
  "encoder.layers.20.self_attn.v_proj": {
    "max": 28.272241592407227,
    "over15": 0.00010240435221930966,
    "weightMax": 0.17759199440479279
  },
  "encoder.layers.20.self_attn.out_proj": {
    "max": 4.364089488983154,
    "over15": 0.0,
    "weightMax": 0.20314835011959076
  },
  "encoder.layers.20.mlp.fc1": {
    "max": 63.6938362121582,
    "over15": 0.04221322014927864,
    "weightMax": 0.31069669127464294
  },
  "encoder.layers.20.mlp.fc2": {
    "max": 19.755334854125977,
    "over15": 1.912279913085513e-06,
    "weightMax": 0.4173778295516968
  },
  "encoder.layers.21.self_attn.q_proj": {
    "max": 29.798311233520508,
    "over15": 9.406911703990772e-05,
    "weightMax": 0.25799158215522766
  },
  "encoder.layers.21.self_attn.k_proj": {
    "max": 29.798311233520508,
    "over15": 9.406911703990772e-05,
    "weightMax": 0.34685996174812317
  },
  "encoder.layers.21.self_attn.v_proj": {
    "max": 29.798311233520508,
    "over15": 9.406911703990772e-05,
    "weightMax": 0.22604598104953766
  },
  "encoder.layers.21.self_attn.out_proj": {
    "max": 5.886954307556152,
    "over15": 0.0,
    "weightMax": 0.18786026537418365
  },
  "encoder.layers.21.mlp.fc1": {
    "max": 170.61080932617188,
    "over15": 0.10434646904468536,
    "weightMax": 0.367507666349411
  },
  "encoder.layers.21.mlp.fc2": {
    "max": 25.098163604736328,
    "over15": 1.1154966159665491e-05,
    "weightMax": 0.2833153009414673
  },
  "encoder.layers.22.self_attn.q_proj": {
    "max": 30.1865291595459,
    "over15": 0.00014765279775019735,
    "weightMax": 0.20620223879814148
  },
  "encoder.layers.22.self_attn.k_proj": {
    "max": 30.1865291595459,
    "over15": 0.00014765279775019735,
    "weightMax": 0.2046823501586914
  },
  "encoder.layers.22.self_attn.v_proj": {
    "max": 30.1865291595459,
    "over15": 0.00014765279775019735,
    "weightMax": 0.17910943925380707
  },
  "encoder.layers.22.self_attn.out_proj": {
    "max": 3.361057758331299,
    "over15": 0.0,
    "weightMax": 0.20936480164527893
  },
  "encoder.layers.22.mlp.fc1": {
    "max": 137.3541259765625,
    "over15": 0.1520954668521881,
    "weightMax": 0.3065267503261566
  },
  "encoder.layers.22.mlp.fc2": {
    "max": 23.519023895263672,
    "over15": 1.97602257685503e-05,
    "weightMax": 0.16916124522686005
  },
  "encoder.layers.23.self_attn.q_proj": {
    "max": 32.97484588623047,
    "over15": 0.00017146776372101158,
    "weightMax": 0.3434046506881714
  },
  "encoder.layers.23.self_attn.k_proj": {
    "max": 32.97484588623047,
    "over15": 0.00017146776372101158,
    "weightMax": 0.34081748127937317
  },
  "encoder.layers.23.self_attn.v_proj": {
    "max": 32.97484588623047,
    "over15": 0.00017146776372101158,
    "weightMax": 0.19471871852874756
  },
  "encoder.layers.23.self_attn.out_proj": {
    "max": 5.905884742736816,
    "over15": 0.0,
    "weightMax": 0.17035984992980957
  },
  "encoder.layers.23.mlp.fc1": {
    "max": 287.2550354003906,
    "over15": 0.20161035656929016,
    "weightMax": 0.22452372312545776
  },
  "encoder.layers.23.mlp.fc2": {
    "max": 26.14187240600586,
    "over15": 3.187133188475855e-05,
    "weightMax": 0.26613184809684753
  },
  "encoder.layers.24.self_attn.q_proj": {
    "max": 34.598392486572266,
    "over15": 0.00021909769566264004,
    "weightMax": 0.23561057448387146
  },
  "encoder.layers.24.self_attn.k_proj": {
    "max": 34.598392486572266,
    "over15": 0.00021909769566264004,
    "weightMax": 0.32973816990852356
  },
  "encoder.layers.24.self_attn.v_proj": {
    "max": 34.598392486572266,
    "over15": 0.00021909769566264004,
    "weightMax": 0.16817793250083923
  },
  "encoder.layers.24.self_attn.out_proj": {
    "max": 5.965823173522949,
    "over15": 0.0,
    "weightMax": 0.3040755093097687
  },
  "encoder.layers.24.mlp.fc1": {
    "max": 311.1141357421875,
    "over15": 0.22575992345809937,
    "weightMax": 0.3162248730659485
  },
  "encoder.layers.24.mlp.fc2": {
    "max": 29.746540069580078,
    "over15": 8.541517308913171e-05,
    "weightMax": 0.17755484580993652
  },
  "encoder.layers.25.self_attn.q_proj": {
    "max": 41.20786666870117,
    "over15": 0.00016075102030299604,
    "weightMax": 0.3333044648170471
  },
  "encoder.layers.25.self_attn.k_proj": {
    "max": 41.20786666870117,
    "over15": 0.00016075102030299604,
    "weightMax": 0.32847651839256287
  },
  "encoder.layers.25.self_attn.v_proj": {
    "max": 41.20786666870117,
    "over15": 0.00016075102030299604,
    "weightMax": 0.19008082151412964
  },
  "encoder.layers.25.self_attn.out_proj": {
    "max": 5.2321391105651855,
    "over15": 0.0,
    "weightMax": 0.3009927570819855
  },
  "encoder.layers.25.mlp.fc1": {
    "max": 252.7682342529297,
    "over15": 0.1282281130552292,
    "weightMax": 0.5487827062606812
  },
  "encoder.layers.25.mlp.fc2": {
    "max": 52.68931198120117,
    "over15": 0.00031457003206014633,
    "weightMax": 0.3679836094379425
  },
  "encoder.layers.26.self_attn.q_proj": {
    "max": 44.30027389526367,
    "over15": 6.191891588969156e-05,
    "weightMax": 0.2666754722595215
  },
  "encoder.layers.26.self_attn.k_proj": {
    "max": 44.30027389526367,
    "over15": 6.191891588969156e-05,
    "weightMax": 0.3050522804260254
  },
  "encoder.layers.26.self_attn.v_proj": {
    "max": 44.30027389526367,
    "over15": 6.191891588969156e-05,
    "weightMax": 0.18168525397777557
  },
  "encoder.layers.26.self_attn.out_proj": {
    "max": 4.220678806304932,
    "over15": 0.0,
    "weightMax": 0.3198651969432831
  },
  "encoder.layers.26.mlp.fc1": {
    "max": 221.23170471191406,
    "over15": 0.03260745108127594,
    "weightMax": 0.2978965640068054
  },
  "encoder.layers.26.mlp.fc2": {
    "max": 36.184288024902344,
    "over15": 0.00022819873993285,
    "weightMax": 0.3672327995300293
  },
  "head.mlp.fc1": {
    "max": 4.635818004608154,
    "over15": 0.0,
    "weightMax": 0.4165218472480774
  },
  "head.mlp.fc2": {
    "max": 2.2876768112182617,
    "over15": 0.0,
    "weightMax": 0.35299229621887207
  }
}
FP8 operands fixed -4, FP32 accumulation cosine 0.90521240234375