# coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Testing suite for the PyTorch {{cookiecutter.modelname}} model. """


{% if cookiecutter.is_encoder_decoder_model == "False" -%}
import unittest

from ...test_modeling_common import floats_tensor
from transformers import is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device

from transformers import {{cookiecutter.camelcase_modelname}}Config
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask


if is_torch_available():
    import torch

    from transformers import (
        {{cookiecutter.camelcase_modelname}}ForCausalLM,
        {{cookiecutter.camelcase_modelname}}ForMaskedLM,
        {{cookiecutter.camelcase_modelname}}ForMultipleChoice,
        {{cookiecutter.camelcase_modelname}}ForQuestionAnswering,
        {{cookiecutter.camelcase_modelname}}ForSequenceClassification,
        {{cookiecutter.camelcase_modelname}}ForTokenClassification,
        {{cookiecutter.camelcase_modelname}}Model,
    )
    from transformers.models.{{cookiecutter.lowercase_modelname}}.modeling_{{cookiecutter.lowercase_modelname}} import (
        {{cookiecutter.uppercase_modelname}}_PRETRAINED_MODEL_ARCHIVE_LIST,
    )


class {{cookiecutter.camelcase_modelname}}ModelTester:
    def __init__(
            self,
            parent,
            batch_size=13,
            seq_length=7,
            is_training=True,
            use_input_mask=True,
            use_token_type_ids=True,
            use_labels=True,
            vocab_size=99,
            hidden_size=32,
            num_hidden_layers=5,
            num_attention_heads=4,
            intermediate_size=37,
            hidden_act="gelu",
            hidden_dropout_prob=0.1,
            attention_probs_dropout_prob=0.1,
            max_position_embeddings=512,
            type_vocab_size=16,
            type_sequence_label_size=2,
            initializer_range=0.02,
            num_labels=3,
            num_choices=4,
            scope=None,
    ):
        self.parent = parent
        self.batch_size = batch_size
        self.seq_length = seq_length
        self.is_training = is_training
        self.use_input_mask = use_input_mask
        self.use_token_type_ids = use_token_type_ids
        self.use_labels = use_labels
        self.vocab_size = vocab_size
        self.hidden_size = hidden_size
        self.num_hidden_layers = num_hidden_layers
        self.num_attention_heads = num_attention_heads
        self.intermediate_size = intermediate_size
        self.hidden_act = hidden_act
        self.hidden_dropout_prob = hidden_dropout_prob
        self.attention_probs_dropout_prob = attention_probs_dropout_prob
        self.max_position_embeddings = max_position_embeddings
        self.type_vocab_size = type_vocab_size
        self.type_sequence_label_size = type_sequence_label_size
        self.initializer_range = initializer_range
        self.num_labels = num_labels
        self.num_choices = num_choices
        self.scope = scope

    def prepare_config_and_inputs(self):
        input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)

        input_mask = None
        if self.use_input_mask:
            input_mask = random_attention_mask([self.batch_size, self.seq_length])

        token_type_ids = None
        if self.use_token_type_ids:
            token_type_ids = ids_tensor([self.batch_size, self.seq_length], self.type_vocab_size)

        sequence_labels = None
        token_labels = None
        choice_labels = None
        if self.use_labels:
            sequence_labels = ids_tensor([self.batch_size], self.type_sequence_label_size)
            token_labels = ids_tensor([self.batch_size, self.seq_length], self.num_labels)
            choice_labels = ids_tensor([self.batch_size], self.num_choices)

        config = self.get_config()

        return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels

    def get_config(self):
        return {{cookiecutter.camelcase_modelname}}Config(
            vocab_size=self.vocab_size,
            hidden_size=self.hidden_size,
            num_hidden_layers=self.num_hidden_layers,
            num_attention_heads=self.num_attention_heads,
            intermediate_size=self.intermediate_size,
            hidden_act=self.hidden_act,
            hidden_dropout_prob=self.hidden_dropout_prob,
            attention_probs_dropout_prob=self.attention_probs_dropout_prob,
            max_position_embeddings=self.max_position_embeddings,
            type_vocab_size=self.type_vocab_size,
            is_decoder=False,
            initializer_range=self.initializer_range,
        )

    def prepare_config_and_inputs_for_decoder(self):
        (
            config,
            input_ids,
            token_type_ids,
            input_mask,
            sequence_labels,
            token_labels,
            choice_labels,
        ) = self.prepare_config_and_inputs()

        config.is_decoder = True
        encoder_hidden_states = floats_tensor([self.batch_size, self.seq_length, self.hidden_size])
        encoder_attention_mask = ids_tensor([self.batch_size, self.seq_length], vocab_size=2)

        return (
            config,
            input_ids,
            token_type_ids,
            input_mask,
            sequence_labels,
            token_labels,
            choice_labels,
            encoder_hidden_states,
            encoder_attention_mask,
        )

    def create_and_check_model(
            self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
    ):
        model = {{cookiecutter.camelcase_modelname}}Model(config=config)
        model.to(torch_device)
        model.eval()
        result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids)
        result = model(input_ids, token_type_ids=token_type_ids)
        result = model(input_ids)
        self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))

    def create_and_check_model_as_decoder(
            self,
            config,
            input_ids,
            token_type_ids,
            input_mask,
            sequence_labels,
            token_labels,
            choice_labels,
            encoder_hidden_states,
            encoder_attention_mask,
    ):
        config.add_cross_attention = True
        model = {{cookiecutter.camelcase_modelname}}Model(config)
        model.to(torch_device)
        model.eval()
        result = model(
            input_ids,
            attention_mask=input_mask,
            token_type_ids=token_type_ids,
            encoder_hidden_states=encoder_hidden_states,
            encoder_attention_mask=encoder_attention_mask,
        )
        result = model(
            input_ids,
            attention_mask=input_mask,
            token_type_ids=token_type_ids,
            encoder_hidden_states=encoder_hidden_states,
        )
        result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids)
        self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))

    def create_and_check_for_causal_lm(
            self,
            config,
            input_ids,
            token_type_ids,
            input_mask,
            sequence_labels,
            token_labels,
            choice_labels,
            encoder_hidden_states,
            encoder_attention_mask,
    ):
        model = {{cookiecutter.camelcase_modelname}}ForCausalLM(config=config)
        model.to(torch_device)
        model.eval()
        result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=token_labels)
        self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.vocab_size))

    def create_and_check_for_masked_lm(
            self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
    ):
        model = {{cookiecutter.camelcase_modelname}}ForMaskedLM(config=config)
        model.to(torch_device)
        model.eval()
        result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=token_labels)
        self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.vocab_size))

    def create_and_check_decoder_model_past_large_inputs(
        self,
        config,
        input_ids,
        token_type_ids,
        input_mask,
        sequence_labels,
        token_labels,
        choice_labels,
        encoder_hidden_states,
        encoder_attention_mask,
    ):
        config.is_decoder = True
        config.add_cross_attention = True
        model = {{cookiecutter.camelcase_modelname}}ForCausalLM(config=config)
        model.to(torch_device)
        model.eval()

        # first forward pass
        outputs = model(
            input_ids,
            attention_mask=input_mask,
            encoder_hidden_states=encoder_hidden_states,
            encoder_attention_mask=encoder_attention_mask,
            use_cache=True,
        )
        past_key_values = outputs.past_key_values

        # create hypothetical multiple next token and extent to next_input_ids
        next_tokens = ids_tensor((self.batch_size, 3), config.vocab_size)
        next_mask = ids_tensor((self.batch_size, 3), vocab_size=2)

        # append to next input_ids and
        next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
        next_attention_mask = torch.cat([input_mask, next_mask], dim=-1)

        output_from_no_past = model(
            next_input_ids,
            attention_mask=next_attention_mask,
            encoder_hidden_states=encoder_hidden_states,
            encoder_attention_mask=encoder_attention_mask,
            output_hidden_states=True,
        )["hidden_states"][0]
        output_from_past = model(
            next_tokens,
            attention_mask=next_attention_mask,
            encoder_hidden_states=encoder_hidden_states,
            encoder_attention_mask=encoder_attention_mask,
            past_key_values=past_key_values,
            output_hidden_states=True,
        )["hidden_states"][0]

        # select random slice
        random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
        output_from_no_past_slice = output_from_no_past[:, -3:, random_slice_idx].detach()
        output_from_past_slice = output_from_past[:, :, random_slice_idx].detach()

        self.parent.assertTrue(output_from_past_slice.shape[1] == next_tokens.shape[1])

        # test that outputs are equal for slice
        self.parent.assertTrue(torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3))

    def create_and_check_for_question_answering(
            self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
    ):
        model = {{cookiecutter.camelcase_modelname}}ForQuestionAnswering(config=config)
        model.to(torch_device)
        model.eval()
        result = model(
            input_ids,
            attention_mask=input_mask,
            token_type_ids=token_type_ids,
            start_positions=sequence_labels,
            end_positions=sequence_labels,
        )
        self.parent.assertEqual(result.start_logits.shape, (self.batch_size, self.seq_length))
        self.parent.assertEqual(result.end_logits.shape, (self.batch_size, self.seq_length))

    def create_and_check_for_sequence_classification(
            self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
    ):
        config.num_labels = self.num_labels
        model = {{cookiecutter.camelcase_modelname}}ForSequenceClassification(config)
        model.to(torch_device)
        model.eval()
        result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=sequence_labels)
        self.parent.assertEqual(result.logits.shape, (self.batch_size, self.num_labels))

    def create_and_check_for_token_classification(
            self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
    ):
        config.num_labels = self.num_labels
        model = {{cookiecutter.camelcase_modelname}}ForTokenClassification(config=config)
        model.to(torch_device)
        model.eval()
        result = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, labels=token_labels)
        self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.num_labels))

    def create_and_check_for_multiple_choice(
            self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
    ):
        config.num_choices = self.num_choices
        model = {{cookiecutter.camelcase_modelname}}ForMultipleChoice(config=config)
        model.to(torch_device)
        model.eval()
        multiple_choice_inputs_ids = input_ids.unsqueeze(1).expand(-1, self.num_choices, -1).contiguous()
        multiple_choice_token_type_ids = token_type_ids.unsqueeze(1).expand(-1, self.num_choices, -1).contiguous()
        multiple_choice_input_mask = input_mask.unsqueeze(1).expand(-1, self.num_choices, -1).contiguous()
        result = model(
            multiple_choice_inputs_ids,
            attention_mask=multiple_choice_input_mask,
            token_type_ids=multiple_choice_token_type_ids,
            labels=choice_labels,
        )
        self.parent.assertEqual(result.logits.shape, (self.batch_size, self.num_choices))

    def prepare_config_and_inputs_for_common(self):
        config_and_inputs = self.prepare_config_and_inputs()
        (
            config,
            input_ids,
            token_type_ids,
            input_mask,
            sequence_labels,
            token_labels,
            choice_labels,
        ) = config_and_inputs
        inputs_dict = {"input_ids": input_ids, "token_type_ids": token_type_ids, "attention_mask": input_mask}
        return config, inputs_dict


@require_torch
class {{cookiecutter.camelcase_modelname}}ModelTest(ModelTesterMixin, unittest.TestCase):

    all_model_classes = (
        (
            {{cookiecutter.camelcase_modelname}}Model,
            {{cookiecutter.camelcase_modelname}}ForMaskedLM,
            {{cookiecutter.camelcase_modelname}}ForCausalLM,
            {{cookiecutter.camelcase_modelname}}ForMultipleChoice,
            {{cookiecutter.camelcase_modelname}}ForQuestionAnswering,
            {{cookiecutter.camelcase_modelname}}ForSequenceClassification,
            {{cookiecutter.camelcase_modelname}}ForTokenClassification,
        )
        if is_torch_available()
        else ()
    )
    all_generative_model_classes = ({{cookiecutter.camelcase_modelname}}ForCausalLM,) if is_torch_available() else ()

    def setUp(self):
        self.model_tester = {{cookiecutter.camelcase_modelname}}ModelTester(self)
        self.config_tester = ConfigTester(self, config_class={{cookiecutter.camelcase_modelname}}Config, hidden_size=37)

    def test_config(self):
        self.config_tester.run_common_tests()

    def test_model(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        self.model_tester.create_and_check_model(*config_and_inputs)

    def test_model_various_embeddings(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        for type in ["absolute", "relative_key", "relative_key_query"]:
            config_and_inputs[0].position_embedding_type = type
            self.model_tester.create_and_check_model(*config_and_inputs)

    def test_for_masked_lm(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        self.model_tester.create_and_check_for_masked_lm(*config_and_inputs)

    def test_for_multiple_choice(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        self.model_tester.create_and_check_for_multiple_choice(*config_and_inputs)

    def test_decoder_model_past_with_large_inputs(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs_for_decoder()
        self.model_tester.create_and_check_decoder_model_past_large_inputs(*config_and_inputs)

    def test_for_question_answering(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        self.model_tester.create_and_check_for_question_answering(*config_and_inputs)

    def test_for_sequence_classification(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        self.model_tester.create_and_check_for_sequence_classification(*config_and_inputs)

    def test_for_token_classification(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        self.model_tester.create_and_check_for_token_classification(*config_and_inputs)

    def test_model_as_decoder(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs_for_decoder()
        self.model_tester.create_and_check_model_as_decoder(*config_and_inputs)

    def test_model_as_decoder_with_default_input_mask(self):
        # This regression test was failing with PyTorch < 1.3
        (
            config,
            input_ids,
            token_type_ids,
            input_mask,
            sequence_labels,
            token_labels,
            choice_labels,
            encoder_hidden_states,
            encoder_attention_mask,
        ) = self.model_tester.prepare_config_and_inputs_for_decoder()

        input_mask = None

        self.model_tester.create_and_check_model_as_decoder(
            config,
            input_ids,
            token_type_ids,
            input_mask,
            sequence_labels,
            token_labels,
            choice_labels,
            encoder_hidden_states,
            encoder_attention_mask,
        )

    @slow
    def test_model_from_pretrained(self):
        for model_name in {{cookiecutter.uppercase_modelname}}_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
            model = {{cookiecutter.camelcase_modelname}}Model.from_pretrained(model_name)
            self.assertIsNotNone(model)


@require_torch
class {{cookiecutter.camelcase_modelname}}ModelIntegrationTest(unittest.TestCase):
    @slow
    def test_inference_masked_lm(self):
        model = {{cookiecutter.camelcase_modelname}}ForMaskedLM.from_pretrained("{{cookiecutter.checkpoint_identifier}}")
        input_ids = torch.tensor([[0, 1, 2, 3, 4, 5]])
        output = model(input_ids)[0]

        # TODO Replace vocab size
        vocab_size = 32000

        expected_shape = torch.Size((1, 6, vocab_size))
        self.assertEqual(output.shape, expected_shape)

        # TODO Replace values below with what was printed above.
        expected_slice = torch.tensor(
            [[[-0.0483, 0.1188, -0.0313], [-0.0606, 0.1435, 0.0199], [-0.0235, 0.1519, 0.0175]]]
        )

        self.assertTrue(torch.allclose(output[:, :3, :3], expected_slice, atol=1e-4))


{% else -%}
import copy
import tempfile
import unittest

from transformers import is_torch_available
from transformers.utils import cached_property
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device

from ...test_configuration_common import ConfigTester
from ...generation.test_utils import GenerationTesterMixin
from ...test_modeling_common import ModelTesterMixin, ids_tensor


if is_torch_available():
    import torch

    from transformers import (
        {{cookiecutter.camelcase_modelname}}Config,
        {{cookiecutter.camelcase_modelname}}ForConditionalGeneration,
        {{cookiecutter.camelcase_modelname}}ForQuestionAnswering,
        {{cookiecutter.camelcase_modelname}}ForCausalLM,
        {{cookiecutter.camelcase_modelname}}ForSequenceClassification,
        {{cookiecutter.camelcase_modelname}}Model,
        {{cookiecutter.camelcase_modelname}}Tokenizer,
    )
    from transformers.models.{{cookiecutter.lowercase_modelname}}.modeling_{{cookiecutter.lowercase_modelname}} import (
        {{cookiecutter.camelcase_modelname}}Decoder,
        {{cookiecutter.camelcase_modelname}}Encoder,
    )


def prepare_{{cookiecutter.lowercase_modelname}}_inputs_dict(
    config,
    input_ids,
    decoder_input_ids,
    attention_mask=None,
    decoder_attention_mask=None,
):
    if attention_mask is None:
        attention_mask = input_ids.ne(config.pad_token_id)
    if decoder_attention_mask is None:
        decoder_attention_mask = decoder_input_ids.ne(config.pad_token_id)
    return {
        "input_ids": input_ids,
        "decoder_input_ids": decoder_input_ids,
        "attention_mask": attention_mask,
        "decoder_attention_mask": attention_mask,
    }


@require_torch
class {{cookiecutter.camelcase_modelname}}ModelTester:
    def __init__(
        self,
        parent,
        batch_size=13,
        seq_length=7,
        is_training=True,
        use_labels=False,
        vocab_size=99,
        hidden_size=16,
        num_hidden_layers=2,
        num_attention_heads=4,
        intermediate_size=4,
        hidden_act="gelu",
        hidden_dropout_prob=0.1,
        attention_probs_dropout_prob=0.1,
        max_position_embeddings=20,
        eos_token_id=2,
        pad_token_id=1,
        bos_token_id=0,
    ):
        self.parent = parent
        self.batch_size = batch_size
        self.seq_length = seq_length
        self.is_training = is_training
        self.use_labels = use_labels
        self.vocab_size = vocab_size
        self.hidden_size = hidden_size
        self.num_hidden_layers = num_hidden_layers
        self.num_attention_heads = num_attention_heads
        self.intermediate_size = intermediate_size
        self.hidden_act = hidden_act
        self.hidden_dropout_prob = hidden_dropout_prob
        self.attention_probs_dropout_prob = attention_probs_dropout_prob
        self.max_position_embeddings = max_position_embeddings
        self.eos_token_id = eos_token_id
        self.pad_token_id = pad_token_id
        self.bos_token_id = bos_token_id

    def prepare_config_and_inputs(self):
        input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
        input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size).clamp(
            3,
        )
        input_ids[:, -1] = self.eos_token_id  # Eos Token

        decoder_input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)

        config = {{cookiecutter.camelcase_modelname}}Config(
            vocab_size=self.vocab_size,
            d_model=self.hidden_size,
            encoder_layers=self.num_hidden_layers,
            decoder_layers=self.num_hidden_layers,
            encoder_attention_heads=self.num_attention_heads,
            decoder_attention_heads=self.num_attention_heads,
            encoder_ffn_dim=self.intermediate_size,
            decoder_ffn_dim=self.intermediate_size,
            dropout=self.hidden_dropout_prob,
            attention_dropout=self.attention_probs_dropout_prob,
            max_position_embeddings=self.max_position_embeddings,
            eos_token_id=self.eos_token_id,
            bos_token_id=self.bos_token_id,
            pad_token_id=self.pad_token_id,
        )
        inputs_dict = prepare_{{cookiecutter.lowercase_modelname}}_inputs_dict(config, input_ids, decoder_input_ids)
        return config, inputs_dict

    def prepare_config_and_inputs_for_common(self):
        config, inputs_dict = self.prepare_config_and_inputs()
        return config, inputs_dict

    def create_and_check_decoder_model_past_large_inputs(self, config, inputs_dict):
        model = {{cookiecutter.camelcase_modelname}}Model(config=config).get_decoder().to(torch_device).eval()
        input_ids = inputs_dict["input_ids"]
        attention_mask = inputs_dict["attention_mask"]

        # first forward pass
        outputs = model(input_ids, attention_mask=attention_mask, use_cache=True)

        output, past_key_values = outputs.to_tuple()

        # create hypothetical multiple next token and extent to next_input_ids
        next_tokens = ids_tensor((self.batch_size, 3), config.vocab_size)
        next_attn_mask = ids_tensor((self.batch_size, 3), 2)

        # append to next input_ids and
        next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
        next_attention_mask = torch.cat([attention_mask, next_attn_mask], dim=-1)

        output_from_no_past = model(next_input_ids, attention_mask=next_attention_mask)["last_hidden_state"]
        output_from_past = model(next_tokens, attention_mask=next_attention_mask, past_key_values=past_key_values)["last_hidden_state"]

        # select random slice
        random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
        output_from_no_past_slice = output_from_no_past[:, -3:, random_slice_idx].detach()
        output_from_past_slice = output_from_past[:, :, random_slice_idx].detach()

        self.parent.assertTrue(output_from_past_slice.shape[1] == next_tokens.shape[1])

        # test that outputs are equal for slice
        self.parent.assertTrue(torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-2))

    def check_encoder_decoder_model_standalone(self, config, inputs_dict):
        model = {{cookiecutter.camelcase_modelname}}Model(config=config).to(torch_device).eval()
        outputs = model(**inputs_dict)

        encoder_last_hidden_state = outputs.encoder_last_hidden_state
        last_hidden_state = outputs.last_hidden_state

        with tempfile.TemporaryDirectory() as tmpdirname:
            encoder = model.get_encoder()
            encoder.save_pretrained(tmpdirname)
            encoder = {{cookiecutter.camelcase_modelname}}Encoder.from_pretrained(tmpdirname).to(torch_device)

        encoder_last_hidden_state_2 = encoder(inputs_dict["input_ids"], attention_mask=inputs_dict["attention_mask"])[
            0
        ]

        self.parent.assertTrue((encoder_last_hidden_state_2 - encoder_last_hidden_state).abs().max().item() < 1e-3)

        with tempfile.TemporaryDirectory() as tmpdirname:
            decoder = model.get_decoder()
            decoder.save_pretrained(tmpdirname)
            decoder = {{cookiecutter.camelcase_modelname}}Decoder.from_pretrained(tmpdirname).to(torch_device)

        last_hidden_state_2 = decoder(
            input_ids=inputs_dict["decoder_input_ids"],
            attention_mask=inputs_dict["decoder_attention_mask"],
            encoder_hidden_states=encoder_last_hidden_state,
            encoder_attention_mask=inputs_dict["attention_mask"],
        )[0]

        self.parent.assertTrue((last_hidden_state_2 - last_hidden_state).abs().max().item() < 1e-3)


@require_torch
class {{cookiecutter.camelcase_modelname}}ModelTest(ModelTesterMixin, GenerationTesterMixin, unittest.TestCase):
    all_model_classes = (
        ({{cookiecutter.camelcase_modelname}}Model, {{cookiecutter.camelcase_modelname}}ForConditionalGeneration, {{cookiecutter.camelcase_modelname}}ForSequenceClassification, {{cookiecutter.camelcase_modelname}}ForQuestionAnswering)
        if is_torch_available()
        else ()
    )
    all_generative_model_classes = ({{cookiecutter.camelcase_modelname}}ForConditionalGeneration,) if is_torch_available() else ()
    is_encoder_decoder = True
    test_pruning = False
    test_head_masking = False
    test_missing_keys = False

    def setUp(self):
        self.model_tester = {{cookiecutter.camelcase_modelname}}ModelTester(self)
        self.config_tester = ConfigTester(self, config_class={{cookiecutter.camelcase_modelname}}Config)

    def test_config(self):
        self.config_tester.run_common_tests()

    def test_save_load_strict(self):
        config, inputs_dict = self.model_tester.prepare_config_and_inputs()
        for model_class in self.all_model_classes:
            model = model_class(config)

            with tempfile.TemporaryDirectory() as tmpdirname:
                model.save_pretrained(tmpdirname)
                model2, info = model_class.from_pretrained(tmpdirname, output_loading_info=True)
            self.assertEqual(info["missing_keys"], [])

    def test_decoder_model_past_with_large_inputs(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        self.model_tester.create_and_check_decoder_model_past_large_inputs(*config_and_inputs)

    def test_encoder_decoder_model_standalone(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs_for_common()
        self.model_tester.check_encoder_decoder_model_standalone(*config_and_inputs)

    # {{cookiecutter.camelcase_modelname}}ForSequenceClassification does not support inputs_embeds
    def test_inputs_embeds(self):
        config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()

        for model_class in ({{cookiecutter.camelcase_modelname}}Model, {{cookiecutter.camelcase_modelname}}ForConditionalGeneration, {{cookiecutter.camelcase_modelname}}ForQuestionAnswering):
            model = model_class(config)
            model.to(torch_device)
            model.eval()

            inputs = copy.deepcopy(self._prepare_for_class(inputs_dict, model_class))

            if not self.is_encoder_decoder:
                input_ids = inputs["input_ids"]
                del inputs["input_ids"]
            else:
                encoder_input_ids = inputs["input_ids"]
                decoder_input_ids = inputs.get("decoder_input_ids", encoder_input_ids)
                del inputs["input_ids"]
                inputs.pop("decoder_input_ids", None)

            wte = model.get_input_embeddings()
            if not self.is_encoder_decoder:
                inputs["inputs_embeds"] = wte(input_ids)
            else:
                inputs["inputs_embeds"] = wte(encoder_input_ids)
                inputs["decoder_inputs_embeds"] = wte(decoder_input_ids)

            with torch.no_grad():
                model(**inputs)[0]

    def test_generate_fp16(self):
        config, input_dict = self.model_tester.prepare_config_and_inputs()
        input_ids = input_dict["input_ids"]
        attention_mask = input_ids.ne(1).to(torch_device)
        model = {{cookiecutter.camelcase_modelname}}ForConditionalGeneration(config).eval().to(torch_device)
        if torch_device == "cuda":
            model.half()
        model.generate(input_ids, attention_mask=attention_mask)
        model.generate(num_beams=4, do_sample=True, early_stopping=False, num_return_sequences=3)


def assert_tensors_close(a, b, atol=1e-12, prefix=""):
    """If tensors have different shapes, different values or a and b are not both tensors, raise a nice Assertion error."""
    if a is None and b is None:
        return True
    try:
        if torch.allclose(a, b, atol=atol):
            return True
        raise
    except Exception:
        pct_different = (torch.gt((a - b).abs(), atol)).float().mean().item()
        if a.numel() > 100:
            msg = f"tensor values are {pct_different:.1%} percent different."
        else:
            msg = f"{a} != {b}"
        if prefix:
            msg = prefix + ": " + msg
        raise AssertionError(msg)


def _long_tensor(tok_lst):
    return torch.tensor(tok_lst, dtype=torch.long, device=torch_device)


TOLERANCE = 1e-4


@require_torch
@require_sentencepiece
@require_tokenizers
@slow
class {{cookiecutter.camelcase_modelname}}ModelIntegrationTests(unittest.TestCase):
    @cached_property
    def default_tokenizer(self):
        return {{cookiecutter.camelcase_modelname}}Tokenizer.from_pretrained('{{cookiecutter.checkpoint_identifier}}')

    def test_inference_no_head(self):
        model = {{cookiecutter.camelcase_modelname}}Model.from_pretrained('{{cookiecutter.checkpoint_identifier}}').to(torch_device)
        input_ids = _long_tensor([[0, 31414, 232, 328, 740, 1140, 12695, 69, 46078, 1588, 2]])
        decoder_input_ids = _long_tensor([[2, 0, 31414, 232, 328, 740, 1140, 12695, 69, 46078, 1588]])
        inputs_dict = prepare_{{cookiecutter.lowercase_modelname}}_inputs_dict(model.config, input_ids, decoder_input_ids)
        with torch.no_grad():
            output = model(**inputs_dict)[0]
        expected_shape = torch.Size((1, 11, 1024))
        self.assertEqual(output.shape, expected_shape)
        # change to expected output here
        expected_slice = torch.tensor(
            [[0.7144, 0.8143, -1.2813], [0.7144, 0.8143, -1.2813], [-0.0467, 2.5911, -2.1845]], device=torch_device
        )
        self.assertTrue(torch.allclose(output[:, :3, :3], expected_slice, atol=TOLERANCE))

    def test_inference_head(self):
        model = {{cookiecutter.camelcase_modelname}}ForConditionalGeneration.from_pretrained('{{cookiecutter.checkpoint_identifier}}').to(torch_device)

        # change to intended input
        input_ids = _long_tensor([[0, 31414, 232, 328, 740, 1140, 12695, 69, 46078, 1588, 2]])
        decoder_input_ids = _long_tensor([[0, 31414, 232, 328, 740, 1140, 12695, 69, 46078, 1588, 2]])
        inputs_dict = prepare_{{cookiecutter.lowercase_modelname}}_inputs_dict(model.config, input_ids, decoder_input_ids)
        with torch.no_grad():
            output = model(**inputs_dict)[0]
        expected_shape = torch.Size((1, 11, model.config.vocab_size))
        self.assertEqual(output.shape, expected_shape)
        # change to expected output here
        expected_slice = torch.tensor(
            [[0.7144, 0.8143, -1.2813], [0.7144, 0.8143, -1.2813], [-0.0467, 2.5911, -2.1845]], device=torch_device
        )
        self.assertTrue(torch.allclose(output[:, :3, :3], expected_slice, atol=TOLERANCE))

    def test_seq_to_seq_generation(self):
        hf = {{cookiecutter.camelcase_modelname}}ForConditionalGeneration.from_pretrained('{{cookiecutter.checkpoint_identifier}}').to(torch_device)
        tok = {{cookiecutter.camelcase_modelname}}Tokenizer.from_pretrained('{{cookiecutter.checkpoint_identifier}}')

        batch_input = [
            # string 1,
            # string 2,
            # string 3,
            # string 4,
        ]

        # The below article tests that we don't add any hypotheses outside of the top n_beams
        dct = tok.batch_encode_plus(
            batch_input,
            max_length=512,
            padding="max_length",
            truncation_strategy="only_first",
            truncation=True,
            return_tensors="pt",
        )

        hypotheses_batch = hf.generate(
            input_ids=dct["input_ids"].to(torch_device),
            attention_mask=dct["attention_mask"].to(torch_device),
            num_beams=2,
        )

        EXPECTED = [
            # here expected 1,
            # here expected 2,
            # here expected 3,
            # here expected 4,
        ]

        generated = tok.batch_decode(
            hypotheses_batch.tolist(), clean_up_tokenization_spaces=True, skip_special_tokens=True
        )
        assert generated == EXPECTED


class {{cookiecutter.camelcase_modelname}}StandaloneDecoderModelTester:
    def __init__(
        self,
        parent,
        vocab_size=99,
        batch_size=13,
        d_model=16,
        decoder_seq_length=7,
        is_training=True,
        is_decoder=True,
        use_attention_mask=True,
        use_cache=False,
        use_labels=True,
        decoder_start_token_id=2,
        decoder_ffn_dim=32,
        decoder_layers=4,
        encoder_attention_heads=4,
        decoder_attention_heads=4,
        max_position_embeddings=30,
        is_encoder_decoder=False,
        pad_token_id=0,
        bos_token_id=1,
        eos_token_id=2,
        scope=None,
    ):
        self.parent = parent
        self.batch_size = batch_size
        self.decoder_seq_length = decoder_seq_length
        # For common tests
        self.seq_length = self.decoder_seq_length
        self.is_training = is_training
        self.use_attention_mask = use_attention_mask
        self.use_labels = use_labels

        self.vocab_size = vocab_size
        self.d_model = d_model
        self.hidden_size = d_model
        self.num_hidden_layers = decoder_layers
        self.decoder_layers = decoder_layers
        self.decoder_ffn_dim = decoder_ffn_dim
        self.encoder_attention_heads = encoder_attention_heads
        self.decoder_attention_heads = decoder_attention_heads
        self.num_attention_heads = decoder_attention_heads
        self.eos_token_id = eos_token_id
        self.bos_token_id = bos_token_id
        self.pad_token_id = pad_token_id
        self.decoder_start_token_id = decoder_start_token_id
        self.use_cache = use_cache
        self.max_position_embeddings = max_position_embeddings
        self.is_encoder_decoder = is_encoder_decoder

        self.scope = None
        self.decoder_key_length = decoder_seq_length
        self.base_model_out_len = 2
        self.decoder_attention_idx = 1

    def prepare_config_and_inputs(self):
        input_ids = ids_tensor([self.batch_size, self.decoder_seq_length], self.vocab_size)

        attention_mask = None
        if self.use_attention_mask:
            attention_mask = ids_tensor([self.batch_size, self.decoder_seq_length], vocab_size=2)

        lm_labels = None
        if self.use_labels:
            lm_labels = ids_tensor([self.batch_size, self.decoder_seq_length], self.vocab_size)

        config = {{cookiecutter.camelcase_modelname}}Config(
            vocab_size=self.vocab_size,
            d_model=self.d_model,
            decoder_layers=self.decoder_layers,
            decoder_ffn_dim=self.decoder_ffn_dim,
            encoder_attention_heads=self.encoder_attention_heads,
            decoder_attention_heads=self.decoder_attention_heads,
            eos_token_id=self.eos_token_id,
            bos_token_id=self.bos_token_id,
            use_cache=self.use_cache,
            pad_token_id=self.pad_token_id,
            decoder_start_token_id=self.decoder_start_token_id,
            max_position_embeddings=self.max_position_embeddings,
            is_encoder_decoder=self.is_encoder_decoder,
        )

        return (
            config,
            input_ids,
            attention_mask,
            lm_labels,
        )

    def create_and_check_decoder_model_past(
        self,
        config,
        input_ids,
        attention_mask,
        lm_labels,
    ):
        config.use_cache = True
        model = {{cookiecutter.camelcase_modelname}}Decoder(config=config).to(torch_device).eval()
        # first forward pass
        outputs = model(input_ids, use_cache=True)
        outputs_use_cache_conf = model(input_ids)
        outputs_no_past = model(input_ids, use_cache=False)

        self.parent.assertTrue(len(outputs) == len(outputs_use_cache_conf))
        self.parent.assertTrue(len(outputs) == len(outputs_no_past) + 1)

        past_key_values = outputs["past_key_values"]

        # create hypothetical next token and extent to next_input_ids
        next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size)

        # append to next input_ids and
        next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)

        output_from_no_past = model(next_input_ids)["last_hidden_state"]
        output_from_past = model(next_tokens, past_key_values=past_key_values)["last_hidden_state"]

        # select random slice
        random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
        output_from_no_past_slice = output_from_no_past[:, next_input_ids.shape[-1] - 1, random_slice_idx].detach()
        output_from_past_slice = output_from_past[:, 0, random_slice_idx].detach()

        # test that outputs are equal for slice
        assert torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3)

    def create_and_check_decoder_model_attention_mask_past(
        self,
        config,
        input_ids,
        attention_mask,
        lm_labels,
    ):
        model = {{cookiecutter.camelcase_modelname}}Decoder(config=config).to(torch_device).eval()

        # create attention mask
        attn_mask = torch.ones(input_ids.shape, dtype=torch.long, device=torch_device)

        half_seq_length = input_ids.shape[-1] // 2
        attn_mask[:, half_seq_length:] = 0

        # first forward pass
        past_key_values = model(input_ids, attention_mask=attn_mask, use_cache=True)["past_key_values"]

        # create hypothetical next token and extent to next_input_ids
        next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size)

        # change a random masked slice from input_ids
        random_seq_idx_to_change = ids_tensor((1,), half_seq_length).item() + 1
        random_other_next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size).squeeze(-1)
        input_ids[:, -random_seq_idx_to_change] = random_other_next_tokens

        # append to next input_ids and attn_mask
        next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
        attn_mask = torch.cat(
            [attn_mask, torch.ones((attn_mask.shape[0], 1), dtype=torch.long, device=torch_device)],
            dim=1,
        )

        # get two different outputs
        output_from_no_past = model(next_input_ids)["last_hidden_state"]
        output_from_past = model(next_tokens, past_key_values=past_key_values)["last_hidden_state"]

        # select random slice
        random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
        output_from_no_past_slice = output_from_no_past[:, next_input_ids.shape[-1] - 1, random_slice_idx].detach()
        output_from_past_slice = output_from_past[:, 0, random_slice_idx].detach()

        # test that outputs are equal for slice
        assert torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-2)

    def prepare_config_and_inputs_for_common(self):
        config_and_inputs = self.prepare_config_and_inputs()
        (
            config,
            input_ids,
            attention_mask,
            lm_labels,
        ) = config_and_inputs

        inputs_dict = {
            "input_ids": input_ids,
            "attention_mask": attention_mask,
        }
        return config, inputs_dict


@require_torch
class {{cookiecutter.camelcase_modelname}}StandaloneDecoderModelTest(ModelTesterMixin, GenerationTesterMixin, unittest.TestCase):
    all_model_classes = ({{cookiecutter.camelcase_modelname}}Decoder, {{cookiecutter.camelcase_modelname}}ForCausalLM) if is_torch_available() else ()
    all_generative_model_classes = ({{cookiecutter.camelcase_modelname}}ForCausalLM,) if is_torch_available() else ()
    test_pruning = False
    is_encoder_decoder = False

    def setUp(
        self,
    ):
        self.model_tester = {{cookiecutter.camelcase_modelname}}StandaloneDecoderModelTester(self, is_training=False)
        self.config_tester = ConfigTester(self, config_class={{cookiecutter.camelcase_modelname}}Config)

    def test_config(self):
        self.config_tester.run_common_tests()

    def test_decoder_model_past(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        self.model_tester.create_and_check_decoder_model_past(*config_and_inputs)

    def test_decoder_model_attn_mask_past(self):
        config_and_inputs = self.model_tester.prepare_config_and_inputs()
        self.model_tester.create_and_check_decoder_model_attention_mask_past(*config_and_inputs)

    def test_retain_grad_hidden_states_attentions(self):
        # decoder cannot keep gradients
        return
{% endif -%}