init

transformers/tests/models/mobilebert/test_modeling_mobilebert.py

@@ -0,0 +1,430 @@
+# Copyright 2020 The HuggingFace 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.
+
+
+import unittest
+
+import pytest
+from packaging import version
+
+from transformers import AutoTokenizer, MobileBertConfig, MobileBertForMaskedLM, is_torch_available
+from transformers.models.auto import get_values
+from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
+
+from ...test_configuration_common import ConfigTester
+from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
+from ...test_pipeline_mixin import PipelineTesterMixin
+
+
+if is_torch_available():
+    import torch
+
+    from transformers import (
+        MODEL_FOR_PRETRAINING_MAPPING,
+        MobileBertForMaskedLM,
+        MobileBertForMultipleChoice,
+        MobileBertForNextSentencePrediction,
+        MobileBertForPreTraining,
+        MobileBertForQuestionAnswering,
+        MobileBertForSequenceClassification,
+        MobileBertForTokenClassification,
+        MobileBertModel,
+    )
+
+
+class MobileBertModelTester:
+    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=64,
+        embedding_size=32,
+        num_hidden_layers=2,
+        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.embedding_size = embedding_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 MobileBertConfig(
+            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,
+            embedding_size=self.embedding_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 create_and_check_mobilebert_model(
+        self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        model = MobileBertModel(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))
+        self.parent.assertEqual(result.pooler_output.shape, (self.batch_size, self.hidden_size))
+
+    def create_and_check_mobilebert_for_masked_lm(
+        self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        model = MobileBertForMaskedLM(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_mobilebert_for_next_sequence_prediction(
+        self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        model = MobileBertForNextSentencePrediction(config=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, 2))
+
+    def create_and_check_mobilebert_for_pretraining(
+        self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        model = MobileBertForPreTraining(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,
+            next_sentence_label=sequence_labels,
+        )
+        self.parent.assertEqual(result.prediction_logits.shape, (self.batch_size, self.seq_length, self.vocab_size))
+        self.parent.assertEqual(result.seq_relationship_logits.shape, (self.batch_size, 2))
+
+    def create_and_check_mobilebert_for_question_answering(
+        self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
+    ):
+        model = MobileBertForQuestionAnswering(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_mobilebert_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 = MobileBertForSequenceClassification(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_mobilebert_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 = MobileBertForTokenClassification(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_mobilebert_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 = MobileBertForMultipleChoice(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 MobileBertModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
+    all_model_classes = (
+        (
+            MobileBertModel,
+            MobileBertForMaskedLM,
+            MobileBertForMultipleChoice,
+            MobileBertForNextSentencePrediction,
+            MobileBertForPreTraining,
+            MobileBertForQuestionAnswering,
+            MobileBertForSequenceClassification,
+            MobileBertForTokenClassification,
+        )
+        if is_torch_available()
+        else ()
+    )
+    pipeline_model_mapping = (
+        {
+            "feature-extraction": MobileBertModel,
+            "fill-mask": MobileBertForMaskedLM,
+            "question-answering": MobileBertForQuestionAnswering,
+            "text-classification": MobileBertForSequenceClassification,
+            "token-classification": MobileBertForTokenClassification,
+            "zero-shot": MobileBertForSequenceClassification,
+        }
+        if is_torch_available()
+        else {}
+    )
+    fx_compatible = False  # won't be maintained
+
+    # special case for ForPreTraining model
+    def _prepare_for_class(self, inputs_dict, model_class, return_labels=False):
+        inputs_dict = super()._prepare_for_class(inputs_dict, model_class, return_labels=return_labels)
+
+        if return_labels:
+            if model_class in get_values(MODEL_FOR_PRETRAINING_MAPPING):
+                inputs_dict["labels"] = torch.zeros(
+                    (self.model_tester.batch_size, self.model_tester.seq_length), dtype=torch.long, device=torch_device
+                )
+                inputs_dict["next_sentence_label"] = torch.zeros(
+                    self.model_tester.batch_size, dtype=torch.long, device=torch_device
+                )
+        return inputs_dict
+
+    # TODO (@SunMarc): Fix me
+    @unittest.skip(reason="It's broken.")
+    def test_resize_tokens_embeddings(self):
+        super().test_resize_tokens_embeddings()
+
+    def setUp(self):
+        self.model_tester = MobileBertModelTester(self)
+        self.config_tester = ConfigTester(self, config_class=MobileBertConfig, hidden_size=37)
+
+    def test_config(self):
+        self.config_tester.run_common_tests()
+
+    def test_mobilebert_model(self):
+        config_and_inputs = self.model_tester.prepare_config_and_inputs()
+        self.model_tester.create_and_check_mobilebert_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_mobilebert_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_mobilebert_for_multiple_choice(*config_and_inputs)
+
+    def test_for_next_sequence_prediction(self):
+        config_and_inputs = self.model_tester.prepare_config_and_inputs()
+        self.model_tester.create_and_check_mobilebert_for_next_sequence_prediction(*config_and_inputs)
+
+    def test_for_pretraining(self):
+        config_and_inputs = self.model_tester.prepare_config_and_inputs()
+        self.model_tester.create_and_check_mobilebert_for_pretraining(*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_mobilebert_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_mobilebert_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_mobilebert_for_token_classification(*config_and_inputs)
+
+
+def _long_tensor(tok_lst):
+    return torch.tensor(
+        tok_lst,
+        dtype=torch.long,
+        device=torch_device,
+    )
+
+
+TOLERANCE = 1e-3
+
+
+@require_torch
+@require_sentencepiece
+@require_tokenizers
+class MobileBertModelIntegrationTests(unittest.TestCase):
+    @slow
+    def test_inference_no_head(self):
+        model = MobileBertModel.from_pretrained("google/mobilebert-uncased", attn_implementation="eager").to(
+            torch_device
+        )
+        input_ids = _long_tensor([[101, 7110, 1005, 1056, 2023, 11333, 17413, 1029, 102]])
+        with torch.no_grad():
+            output = model(input_ids)[0]
+        expected_shape = torch.Size((1, 9, 512))
+        self.assertEqual(output.shape, expected_shape)
+        expected_slice = torch.tensor(
+            [
+                [
+                    [-2.4736526e07, 8.2691656e04, 1.6521838e05],
+                    [-5.7541704e-01, 3.9056022e00, 4.4011507e00],
+                    [2.6047359e00, 1.5677652e00, -1.7324188e-01],
+                ]
+            ],
+            device=torch_device,
+        )
+
+        # MobileBERT results range from 10e0 to 10e8. Even a 0.0000001% difference with a value of 10e8 results in a
+        # ~1 difference, it's therefore not a good idea to measure using addition.
+        # Here, we instead divide the expected result with the result in order to obtain ~1. We then check that the
+        # result is held between bounds: 1 - TOLERANCE < expected_result / result < 1 + TOLERANCE
+        lower_bound = torch.all((expected_slice / output[..., :3, :3]) >= 1 - TOLERANCE)
+        upper_bound = torch.all((expected_slice / output[..., :3, :3]) <= 1 + TOLERANCE)
+
+        self.assertTrue(lower_bound and upper_bound)
+
+    @pytest.mark.torch_export_test
+    @slow
+    def test_export(self):
+        if version.parse(torch.__version__) < version.parse("2.4.0"):
+            self.skipTest(reason="This test requires torch >= 2.4 to run.")
+
+        mobilebert_model = "google/mobilebert-uncased"
+        device = "cpu"
+        attn_implementation = "eager"
+        max_length = 512
+
+        tokenizer = AutoTokenizer.from_pretrained(mobilebert_model)
+        inputs = tokenizer(
+            f"the man worked as a {tokenizer.mask_token}.",
+            return_tensors="pt",
+            padding="max_length",
+            max_length=max_length,
+        )
+
+        model = MobileBertForMaskedLM.from_pretrained(
+            mobilebert_model,
+            device_map=device,
+            attn_implementation=attn_implementation,
+        )
+
+        logits = model(**inputs).logits
+        eg_predicted_mask = tokenizer.decode(logits[0, 6].topk(5).indices)
+        self.assertEqual(eg_predicted_mask.split(), ["carpenter", "waiter", "mechanic", "teacher", "clerk"])
+
+        exported_program = torch.export.export(
+            model,
+            args=(inputs["input_ids"],),
+            kwargs={"attention_mask": inputs["attention_mask"]},
+            strict=True,
+        )
+
+        result = exported_program.module().forward(inputs["input_ids"], inputs["attention_mask"])
+        ep_predicted_mask = tokenizer.decode(result.logits[0, 6].topk(5).indices)
+        self.assertEqual(eg_predicted_mask, ep_predicted_mask)

transformers/tests/models/mobilebert/test_tokenization_mobilebert.py

@@ -0,0 +1,362 @@
+# Copyright 2022 Leon Derczynski. 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 MobileBERT tokenizer."""
+
+import os
+import unittest
+
+from transformers import MobileBertTokenizer, MobileBertTokenizerFast
+from transformers.models.bert.tokenization_bert import (
+    VOCAB_FILES_NAMES,
+    BasicTokenizer,
+    WordpieceTokenizer,
+    _is_control,
+    _is_punctuation,
+    _is_whitespace,
+)
+from transformers.testing_utils import require_tokenizers, slow
+
+from ...test_tokenization_common import TokenizerTesterMixin, filter_non_english
+
+
+@require_tokenizers
+class MobileBERTTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
+    from_pretrained_id = "mobilebert-uncased"
+    tokenizer_class = MobileBertTokenizer
+    rust_tokenizer_class = MobileBertTokenizerFast
+    test_rust_tokenizer = True
+    space_between_special_tokens = True
+    from_pretrained_filter = filter_non_english
+    pre_trained_model_path = "google/mobilebert-uncased"
+
+    @classmethod
+    def setUpClass(cls):
+        super().setUpClass()
+
+        vocab_tokens = [
+            "[UNK]",
+            "[CLS]",
+            "[SEP]",
+            "[PAD]",
+            "[MASK]",
+            "want",
+            "##want",
+            "##ed",
+            "wa",
+            "un",
+            "runn",
+            "##ing",
+            ",",
+            "low",
+            "lowest",
+        ]
+        cls.vocab_file = os.path.join(cls.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
+        with open(cls.vocab_file, "w", encoding="utf-8") as vocab_writer:
+            vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
+
+        cls.tokenizers_list = [
+            (tokenizer_def[0], cls.pre_trained_model_path, tokenizer_def[2])  # else the 'google/' prefix is stripped
+            for tokenizer_def in cls.tokenizers_list
+        ]
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.get_input_output_texts
+    def get_input_output_texts(self, tokenizer):
+        input_text = "UNwant\u00e9d,running"
+        output_text = "unwanted, running"
+        return input_text, output_text
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_full_tokenizer
+    def test_full_tokenizer(self):
+        tokenizer = self.tokenizer_class(self.vocab_file)
+
+        tokens = tokenizer.tokenize("UNwant\u00e9d,running")
+        self.assertListEqual(tokens, ["un", "##want", "##ed", ",", "runn", "##ing"])
+        self.assertListEqual(tokenizer.convert_tokens_to_ids(tokens), [9, 6, 7, 12, 10, 11])
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_rust_and_python_full_tokenizers
+    def test_rust_and_python_full_tokenizers(self):
+        if not self.test_rust_tokenizer:
+            self.skipTest(reason="test_rust_tokenizer is set to False")
+
+        tokenizer = self.get_tokenizer()
+        rust_tokenizer = self.get_rust_tokenizer()
+
+        sequence = "UNwant\u00e9d,running"
+
+        tokens = tokenizer.tokenize(sequence)
+        rust_tokens = rust_tokenizer.tokenize(sequence)
+        self.assertListEqual(tokens, rust_tokens)
+
+        ids = tokenizer.encode(sequence, add_special_tokens=False)
+        rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
+        self.assertListEqual(ids, rust_ids)
+
+        rust_tokenizer = self.get_rust_tokenizer()
+        ids = tokenizer.encode(sequence)
+        rust_ids = rust_tokenizer.encode(sequence)
+        self.assertListEqual(ids, rust_ids)
+
+        # With lower casing
+        tokenizer = self.get_tokenizer(do_lower_case=True)
+        rust_tokenizer = self.get_rust_tokenizer(do_lower_case=True)
+
+        sequence = "UNwant\u00e9d,running"
+
+        tokens = tokenizer.tokenize(sequence)
+        rust_tokens = rust_tokenizer.tokenize(sequence)
+        self.assertListEqual(tokens, rust_tokens)
+
+        ids = tokenizer.encode(sequence, add_special_tokens=False)
+        rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
+        self.assertListEqual(ids, rust_ids)
+
+        rust_tokenizer = self.get_rust_tokenizer()
+        ids = tokenizer.encode(sequence)
+        rust_ids = rust_tokenizer.encode(sequence)
+        self.assertListEqual(ids, rust_ids)
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_chinese
+    def test_chinese(self):
+        tokenizer = BasicTokenizer()
+
+        self.assertListEqual(tokenizer.tokenize("ah\u535a\u63a8zz"), ["ah", "\u535a", "\u63a8", "zz"])
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_basic_tokenizer_lower
+    def test_basic_tokenizer_lower(self):
+        tokenizer = BasicTokenizer(do_lower_case=True)
+
+        self.assertListEqual(
+            tokenizer.tokenize(" \tHeLLo!how  \n Are yoU?  "), ["hello", "!", "how", "are", "you", "?"]
+        )
+        self.assertListEqual(tokenizer.tokenize("H\u00e9llo"), ["hello"])
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_basic_tokenizer_lower_strip_accents_false
+    def test_basic_tokenizer_lower_strip_accents_false(self):
+        tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=False)
+
+        self.assertListEqual(
+            tokenizer.tokenize(" \tHäLLo!how  \n Are yoU?  "), ["hällo", "!", "how", "are", "you", "?"]
+        )
+        self.assertListEqual(tokenizer.tokenize("H\u00e9llo"), ["h\u00e9llo"])
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_basic_tokenizer_lower_strip_accents_true
+    def test_basic_tokenizer_lower_strip_accents_true(self):
+        tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=True)
+
+        self.assertListEqual(
+            tokenizer.tokenize(" \tHäLLo!how  \n Are yoU?  "), ["hallo", "!", "how", "are", "you", "?"]
+        )
+        self.assertListEqual(tokenizer.tokenize("H\u00e9llo"), ["hello"])
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_basic_tokenizer_lower_strip_accents_default
+    def test_basic_tokenizer_lower_strip_accents_default(self):
+        tokenizer = BasicTokenizer(do_lower_case=True)
+
+        self.assertListEqual(
+            tokenizer.tokenize(" \tHäLLo!how  \n Are yoU?  "), ["hallo", "!", "how", "are", "you", "?"]
+        )
+        self.assertListEqual(tokenizer.tokenize("H\u00e9llo"), ["hello"])
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_basic_tokenizer_no_lower
+    def test_basic_tokenizer_no_lower(self):
+        tokenizer = BasicTokenizer(do_lower_case=False)
+
+        self.assertListEqual(
+            tokenizer.tokenize(" \tHeLLo!how  \n Are yoU?  "), ["HeLLo", "!", "how", "Are", "yoU", "?"]
+        )
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_basic_tokenizer_no_lower_strip_accents_false
+    def test_basic_tokenizer_no_lower_strip_accents_false(self):
+        tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=False)
+
+        self.assertListEqual(
+            tokenizer.tokenize(" \tHäLLo!how  \n Are yoU?  "), ["HäLLo", "!", "how", "Are", "yoU", "?"]
+        )
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_basic_tokenizer_no_lower_strip_accents_true
+    def test_basic_tokenizer_no_lower_strip_accents_true(self):
+        tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=True)
+
+        self.assertListEqual(
+            tokenizer.tokenize(" \tHäLLo!how  \n Are yoU?  "), ["HaLLo", "!", "how", "Are", "yoU", "?"]
+        )
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_basic_tokenizer_respects_never_split_tokens
+    def test_basic_tokenizer_respects_never_split_tokens(self):
+        tokenizer = BasicTokenizer(do_lower_case=False, never_split=["[UNK]"])
+
+        self.assertListEqual(
+            tokenizer.tokenize(" \tHeLLo!how  \n Are yoU? [UNK]"), ["HeLLo", "!", "how", "Are", "yoU", "?", "[UNK]"]
+        )
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_wordpiece_tokenizer
+    def test_wordpiece_tokenizer(self):
+        vocab_tokens = ["[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing"]
+
+        vocab = {}
+        for i, token in enumerate(vocab_tokens):
+            vocab[token] = i
+        tokenizer = WordpieceTokenizer(vocab=vocab, unk_token="[UNK]")
+
+        self.assertListEqual(tokenizer.tokenize(""), [])
+
+        self.assertListEqual(tokenizer.tokenize("unwanted running"), ["un", "##want", "##ed", "runn", "##ing"])
+
+        self.assertListEqual(tokenizer.tokenize("unwantedX running"), ["[UNK]", "runn", "##ing"])
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_is_whitespace
+    def test_is_whitespace(self):
+        self.assertTrue(_is_whitespace(" "))
+        self.assertTrue(_is_whitespace("\t"))
+        self.assertTrue(_is_whitespace("\r"))
+        self.assertTrue(_is_whitespace("\n"))
+        self.assertTrue(_is_whitespace("\u00a0"))
+
+        self.assertFalse(_is_whitespace("A"))
+        self.assertFalse(_is_whitespace("-"))
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_is_control
+    def test_is_control(self):
+        self.assertTrue(_is_control("\u0005"))
+
+        self.assertFalse(_is_control("A"))
+        self.assertFalse(_is_control(" "))
+        self.assertFalse(_is_control("\t"))
+        self.assertFalse(_is_control("\r"))
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_is_punctuation
+    def test_is_punctuation(self):
+        self.assertTrue(_is_punctuation("-"))
+        self.assertTrue(_is_punctuation("$"))
+        self.assertTrue(_is_punctuation("`"))
+        self.assertTrue(_is_punctuation("."))
+
+        self.assertFalse(_is_punctuation("A"))
+        self.assertFalse(_is_punctuation(" "))
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_clean_text
+    def test_clean_text(self):
+        tokenizer = self.get_tokenizer()
+        rust_tokenizer = self.get_rust_tokenizer()
+
+        # Example taken from the issue https://github.com/huggingface/tokenizers/issues/340
+        self.assertListEqual([tokenizer.tokenize(t) for t in ["Test", "\xad", "test"]], [["[UNK]"], [], ["[UNK]"]])
+
+        self.assertListEqual(
+            [rust_tokenizer.tokenize(t) for t in ["Test", "\xad", "test"]], [["[UNK]"], [], ["[UNK]"]]
+        )
+
+    @slow
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_sequence_builders with google-bert/bert-base-uncased->google/mobilebert-uncased
+    def test_sequence_builders(self):
+        tokenizer = self.tokenizer_class.from_pretrained("google/mobilebert-uncased")
+
+        text = tokenizer.encode("sequence builders", add_special_tokens=False)
+        text_2 = tokenizer.encode("multi-sequence build", add_special_tokens=False)
+
+        encoded_sentence = tokenizer.build_inputs_with_special_tokens(text)
+        encoded_pair = tokenizer.build_inputs_with_special_tokens(text, text_2)
+
+        assert encoded_sentence == [101] + text + [102]
+        assert encoded_pair == [101] + text + [102] + text_2 + [102]
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_offsets_with_special_characters
+    def test_offsets_with_special_characters(self):
+        for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
+            with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
+                tokenizer_r = self.get_rust_tokenizer(pretrained_name, **kwargs)
+
+                sentence = f"A, naïve {tokenizer_r.mask_token} AllenNLP sentence."
+                tokens = tokenizer_r.encode_plus(
+                    sentence,
+                    return_attention_mask=False,
+                    return_token_type_ids=False,
+                    return_offsets_mapping=True,
+                    add_special_tokens=True,
+                )
+
+                do_lower_case = tokenizer_r.do_lower_case if hasattr(tokenizer_r, "do_lower_case") else False
+                expected_results = (
+                    [
+                        ((0, 0), tokenizer_r.cls_token),
+                        ((0, 1), "A"),
+                        ((1, 2), ","),
+                        ((3, 5), "na"),
+                        ((5, 6), "##ï"),
+                        ((6, 8), "##ve"),
+                        ((9, 15), tokenizer_r.mask_token),
+                        ((16, 21), "Allen"),
+                        ((21, 23), "##NL"),
+                        ((23, 24), "##P"),
+                        ((25, 33), "sentence"),
+                        ((33, 34), "."),
+                        ((0, 0), tokenizer_r.sep_token),
+                    ]
+                    if not do_lower_case
+                    else [
+                        ((0, 0), tokenizer_r.cls_token),
+                        ((0, 1), "a"),
+                        ((1, 2), ","),
+                        ((3, 8), "naive"),
+                        ((9, 15), tokenizer_r.mask_token),
+                        ((16, 21), "allen"),
+                        ((21, 23), "##nl"),
+                        ((23, 24), "##p"),
+                        ((25, 33), "sentence"),
+                        ((33, 34), "."),
+                        ((0, 0), tokenizer_r.sep_token),
+                    ]
+                )
+
+                self.assertEqual(
+                    [e[1] for e in expected_results], tokenizer_r.convert_ids_to_tokens(tokens["input_ids"])
+                )
+                self.assertEqual([e[0] for e in expected_results], tokens["offset_mapping"])
+
+    # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_change_tokenize_chinese_chars
+    def test_change_tokenize_chinese_chars(self):
+        list_of_common_chinese_char = ["的", "人", "有"]
+        text_with_chinese_char = "".join(list_of_common_chinese_char)
+        for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
+            with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
+                kwargs["tokenize_chinese_chars"] = True
+                tokenizer_p = self.get_tokenizer(pretrained_name, **kwargs)
+                tokenizer_r = self.get_rust_tokenizer(pretrained_name, **kwargs)
+
+                ids_without_spe_char_p = tokenizer_p.encode(text_with_chinese_char, add_special_tokens=False)
+                ids_without_spe_char_r = tokenizer_r.encode(text_with_chinese_char, add_special_tokens=False)
+
+                tokens_without_spe_char_r = tokenizer_r.convert_ids_to_tokens(ids_without_spe_char_r)
+                tokens_without_spe_char_p = tokenizer_p.convert_ids_to_tokens(ids_without_spe_char_p)
+
+                # it is expected that each Chinese character is not preceded by "##"
+                self.assertListEqual(tokens_without_spe_char_p, list_of_common_chinese_char)
+                self.assertListEqual(tokens_without_spe_char_r, list_of_common_chinese_char)
+
+                kwargs["tokenize_chinese_chars"] = False
+                tokenizer_r = self.get_rust_tokenizer(pretrained_name, **kwargs)
+                tokenizer_p = self.get_tokenizer(pretrained_name, **kwargs)
+
+                ids_without_spe_char_r = tokenizer_r.encode(text_with_chinese_char, add_special_tokens=False)
+                ids_without_spe_char_p = tokenizer_p.encode(text_with_chinese_char, add_special_tokens=False)
+
+                tokens_without_spe_char_r = tokenizer_r.convert_ids_to_tokens(ids_without_spe_char_r)
+                tokens_without_spe_char_p = tokenizer_p.convert_ids_to_tokens(ids_without_spe_char_p)
+
+                # it is expected that only the first Chinese character is not preceded by "##".
+                expected_tokens = [
+                    f"##{token}" if idx != 0 else token for idx, token in enumerate(list_of_common_chinese_char)
+                ]
+                self.assertListEqual(tokens_without_spe_char_p, expected_tokens)
+                self.assertListEqual(tokens_without_spe_char_r, expected_tokens)