Style

Makefile

@@ -53,7 +53,6 @@ quality:
 	@python -c "from transformers import *" || (echo '🚨 import failed, this means you introduced unprotected imports! 🚨'; exit 1)
 	ruff check $(check_dirs) setup.py conftest.py
 	ruff format --check $(check_dirs) setup.py conftest.py
-	python utils/custom_init_isort.py --check_only
 	python utils/sort_auto_mappings.py --check_only
 	python utils/check_doc_toc.py
 
@@ -61,7 +60,6 @@ quality:
 # Format source code automatically and check is there are any problems left that need manual fixing
 
 extra_style_checks:
-	python utils/custom_init_isort.py
 	python utils/sort_auto_mappings.py
 	python utils/check_doc_toc.py --fix_and_overwrite
 

examples/pytorch/object-detection/run_object_detection.py

@@ -244,9 +244,7 @@ class DataTrainingArguments:
     )
     image_square_size: Optional[int] = field(
         default=600,
-        metadata={
-            "help": "Image longest size will be resized to this value, then image will be padded to square."
-        },
+        metadata={"help": "Image longest size will be resized to this value, then image will be padded to square."},
     )
     max_train_samples: Optional[int] = field(
         default=None,

setup.py

@@ -260,7 +260,15 @@ extras["ja"] = deps_list("fugashi", "ipadic", "unidic_lite", "unidic", "sudachip
 extras["sklearn"] = deps_list("scikit-learn")
 
 extras["tf"] = deps_list("tensorflow", "onnxconverter-common", "tf2onnx", "tensorflow-text", "keras-nlp")
-extras["tf-cpu"] = deps_list("keras", "tensorflow-cpu", "onnxconverter-common", "tf2onnx", "tensorflow-text", "keras-nlp", "tensorflow-probability")
+extras["tf-cpu"] = deps_list(
+    "keras",
+    "tensorflow-cpu",
+    "onnxconverter-common",
+    "tf2onnx",
+    "tensorflow-text",
+    "keras-nlp",
+    "tensorflow-probability",
+)
 
 extras["torch"] = deps_list("torch", "accelerate")
 extras["accelerate"] = deps_list("accelerate")
@@ -380,12 +388,7 @@ extras["dev-tensorflow"] = (
     + extras["tf-speech"]
 )
 extras["dev"] = (
-    extras["all"]
-    + extras["testing"]
-    + extras["quality"]
-    + extras["ja"]
-    + extras["sklearn"]
-    + extras["modelcreation"]
+    extras["all"] + extras["testing"] + extras["quality"] + extras["ja"] + extras["sklearn"] + extras["modelcreation"]
 )
 
 extras["torchhub"] = deps_list(
@@ -470,4 +473,4 @@ extras["tests_examples_torch"] = deps_list()
 extras["tests_examples_tf"] = deps_list()
 extras["tests_custom_tokenizers"] = deps_list()
 extras["tests_exotic_models"] = deps_list()
-extras["consistency"] = deps_list()
+extras["consistency"] = deps_list()

src/transformers/__init__.py

@@ -1168,7 +1168,9 @@ try:
 except OptionalDependencyNotAvailable:
     from .utils import dummy_torch_objects
 
-    _import_structure["utils.dummy_torch_objects"] = [name for name in dir(dummy_torch_objects) if not name.startswith("_")]
+    _import_structure["utils.dummy_torch_objects"] = [
+        name for name in dir(dummy_torch_objects) if not name.startswith("_")
+    ]
 else:
     _import_structure["activations"] = []
     _import_structure["benchmark.benchmark"] = ["PyTorchBenchmark"]

src/transformers/models/albert/configuration_albert.py

@@ -168,7 +168,5 @@ class AlbertOnnxConfig(OnnxConfig):
             ]
         )
 
-__all__ = [
-    "AlbertConfig",
-    "AlbertOnnxConfig"
-]
+
+__all__ = ["AlbertConfig", "AlbertOnnxConfig"]

src/transformers/models/albert/modeling_albert.py

@@ -1393,6 +1393,7 @@ class AlbertForMultipleChoice(AlbertPreTrainedModel):
             attentions=outputs.attentions,
         )
 
+
 __all__ = [
     "load_tf_weights_in_albert",
     "AlbertPreTrainedModel",
@@ -1402,5 +1403,5 @@ __all__ = [
     "AlbertForSequenceClassification",
     "AlbertForTokenClassification",
     "AlbertForQuestionAnswering",
-    "AlbertForMultipleChoice"
+    "AlbertForMultipleChoice",
 ]

src/transformers/models/albert/modeling_flax_albert.py

@@ -1137,5 +1137,5 @@ __all__ = [
     "FlaxAlbertForSequenceClassification",
     "FlaxAlbertForMultipleChoice",
     "FlaxAlbertForTokenClassification",
-    "FlaxAlbertForQuestionAnswering"
+    "FlaxAlbertForQuestionAnswering",
 ]

src/transformers/models/albert/modeling_tf_albert.py

@@ -1569,6 +1569,7 @@ class TFAlbertForMultipleChoice(TFAlbertPreTrainedModel, TFMultipleChoiceLoss):
             with tf.name_scope(self.classifier.name):
                 self.classifier.build([None, None, self.config.hidden_size])
 
+
 __all__ = [
     "TFAlbertPreTrainedModel",
     "TFAlbertModel",
@@ -1577,5 +1578,5 @@ __all__ = [
     "TFAlbertForSequenceClassification",
     "TFAlbertForTokenClassification",
     "TFAlbertForQuestionAnswering",
-    "TFAlbertForMultipleChoice"
+    "TFAlbertForMultipleChoice",
 ]

src/transformers/models/albert/tokenization_albert.py

@@ -347,6 +347,5 @@ class AlbertTokenizer(PreTrainedTokenizer):
 
         return (out_vocab_file,)
 
-__all__ = [
-    "AlbertTokenizer"
-]
+
+__all__ = ["AlbertTokenizer"]

src/transformers/models/albert/tokenization_albert_fast.py

@@ -211,6 +211,5 @@ class AlbertTokenizerFast(PreTrainedTokenizerFast):
 
         return (out_vocab_file,)
 
-__all__ = [
-    "AlbertTokenizerFast"
-]
+
+__all__ = ["AlbertTokenizerFast"]

src/transformers/models/align/configuration_align.py

@@ -384,8 +384,5 @@ class AlignConfig(PretrainedConfig):
 
         return cls(text_config=text_config.to_dict(), vision_config=vision_config.to_dict(), **kwargs)
 
-__all__ = [
-    "AlignTextConfig",
-    "AlignVisionConfig",
-    "AlignConfig"
-]
+
+__all__ = ["AlignTextConfig", "AlignVisionConfig", "AlignConfig"]

src/transformers/models/align/modeling_align.py

@@ -1642,9 +1642,5 @@ class AlignModel(AlignPreTrainedModel):
             vision_model_output=vision_outputs,
         )
 
-__all__ = [
-    "AlignPreTrainedModel",
-    "AlignTextModel",
-    "AlignVisionModel",
-    "AlignModel"
-]
+
+__all__ = ["AlignPreTrainedModel", "AlignTextModel", "AlignVisionModel", "AlignModel"]

src/transformers/models/align/processing_align.py

@@ -122,6 +122,5 @@ class AlignProcessor(ProcessorMixin):
         image_processor_input_names = self.image_processor.model_input_names
         return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names))
 
-__all__ = [
-    "AlignProcessor"
-]
+
+__all__ = ["AlignProcessor"]

src/transformers/models/altclip/configuration_altclip.py

@@ -402,8 +402,5 @@ class AltCLIPConfig(PretrainedConfig):
 
         return cls(text_config=text_config.to_dict(), vision_config=vision_config.to_dict(), **kwargs)
 
-__all__ = [
-    "AltCLIPTextConfig",
-    "AltCLIPVisionConfig",
-    "AltCLIPConfig"
-]
+
+__all__ = ["AltCLIPTextConfig", "AltCLIPVisionConfig", "AltCLIPConfig"]

src/transformers/models/altclip/modeling_altclip.py

@@ -1702,9 +1702,5 @@ def create_position_ids_from_input_ids(input_ids, padding_idx, past_key_values_l
     incremental_indices = (torch.cumsum(mask, dim=1).type_as(mask) + past_key_values_length) * mask
     return incremental_indices.long() + padding_idx
 
-__all__ = [
-    "AltCLIPPreTrainedModel",
-    "AltCLIPVisionModel",
-    "AltCLIPTextModel",
-    "AltCLIPModel"
-]
+
+__all__ = ["AltCLIPPreTrainedModel", "AltCLIPVisionModel", "AltCLIPTextModel", "AltCLIPModel"]

src/transformers/models/altclip/processing_altclip.py

@@ -132,6 +132,5 @@ class AltCLIPProcessor(ProcessorMixin):
         image_processor_input_names = self.image_processor.model_input_names
         return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names))
 
-__all__ = [
-    "AltCLIPProcessor"
-]
+
+__all__ = ["AltCLIPProcessor"]

src/transformers/models/audio_spectrogram_transformer/configuration_audio_spectrogram_transformer.py

@@ -122,6 +122,5 @@ class ASTConfig(PretrainedConfig):
         self.max_length = max_length
         self.num_mel_bins = num_mel_bins
 
-__all__ = [
-    "ASTConfig"
-]
+
+__all__ = ["ASTConfig"]

src/transformers/models/audio_spectrogram_transformer/feature_extraction_audio_spectrogram_transformer.py

@@ -237,6 +237,5 @@ class ASTFeatureExtractor(SequenceFeatureExtractor):
 
         return padded_inputs
 
-__all__ = [
-    "ASTFeatureExtractor"
-]
+
+__all__ = ["ASTFeatureExtractor"]

src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py

@@ -613,8 +613,5 @@ class ASTForAudioClassification(ASTPreTrainedModel):
             attentions=outputs.attentions,
         )
 
-__all__ = [
-    "ASTPreTrainedModel",
-    "ASTModel",
-    "ASTForAudioClassification"
-]
+
+__all__ = ["ASTPreTrainedModel", "ASTModel", "ASTForAudioClassification"]

src/transformers/models/autoformer/configuration_autoformer.py

@@ -243,6 +243,5 @@ class AutoformerConfig(PretrainedConfig):
             + self.input_size * 2  # the log1p(abs(loc)) and log(scale) features
         )
 
-__all__ = [
-    "AutoformerConfig"
-]
+
+__all__ = ["AutoformerConfig"]

src/transformers/models/autoformer/modeling_autoformer.py

@@ -2155,8 +2155,5 @@ class AutoformerForPrediction(AutoformerPreTrainedModel):
             )
         )
 
-__all__ = [
-    "AutoformerPreTrainedModel",
-    "AutoformerModel",
-    "AutoformerForPrediction"
-]
+
+__all__ = ["AutoformerPreTrainedModel", "AutoformerModel", "AutoformerForPrediction"]

src/transformers/models/bark/configuration_bark.py

@@ -329,9 +329,5 @@ class BarkConfig(PretrainedConfig):
             **kwargs,
         )
 
-__all__ = [
-    "BarkSemanticConfig",
-    "BarkCoarseConfig",
-    "BarkFineConfig",
-    "BarkConfig"
-]
+
+__all__ = ["BarkSemanticConfig", "BarkCoarseConfig", "BarkFineConfig", "BarkConfig"]

src/transformers/models/bark/modeling_bark.py

@@ -1023,7 +1023,7 @@ class BarkSemanticModel(BarkCausalModel):
     language modeling head on top.""",
     BARK_MODEL_START_DOCSTRING.format(config="BarkCoarseConfig"),
 )
-@register(backends=('torch',))
+@register(backends=("torch",))
 class BarkCoarseModel(BarkCausalModel):
     base_model_prefix = "coarse_acoustics"
     config_class = BarkCoarseConfig
@@ -1911,11 +1911,12 @@ class BarkModel(BarkPreTrainedModel):
         config.fine_acoustics_config._attn_implementation = config._attn_implementation
         return config
 
+
 __all__ = [
     "BarkPreTrainedModel",
     "BarkCausalModel",
     "BarkFineModel",
     "BarkCoarseModel",
     "BarkSemanticModel",
-    "BarkModel"
+    "BarkModel",
 ]

src/transformers/models/bark/processing_bark.py

@@ -287,6 +287,5 @@ class BarkProcessor(ProcessorMixin):
 
         return encoded_text
 
-__all__ = [
-    "BarkProcessor"
-]
+
+__all__ = ["BarkProcessor"]

src/transformers/models/bart/configuration_bart.py

@@ -403,7 +403,5 @@ class BartOnnxConfig(OnnxSeq2SeqConfigWithPast):
                 flattened_output, name, idx, t
             )
 
-__all__ = [
-    "BartConfig",
-    "BartOnnxConfig"
-]
+
+__all__ = ["BartConfig", "BartOnnxConfig"]

src/transformers/models/bart/modeling_bart.py

@@ -2334,6 +2334,7 @@ class BartForCausalLM(BartPreTrainedModel):
             )
         return reordered_past
 
+
 __all__ = [
     "BartPreTrainedModel",
     "PretrainedBartModel",
@@ -2343,5 +2344,5 @@ __all__ = [
     "BartForConditionalGeneration",
     "BartForSequenceClassification",
     "BartForQuestionAnswering",
-    "BartForCausalLM"
+    "BartForCausalLM",
 ]

src/transformers/models/bart/modeling_flax_bart.py

@@ -2009,5 +2009,5 @@ __all__ = [
     "FlaxBartForSequenceClassification",
     "FlaxBartForQuestionAnswering",
     "FlaxBartDecoderPreTrainedModel",
-    "FlaxBartForCausalLM"
+    "FlaxBartForCausalLM",
 ]

src/transformers/models/bart/modeling_tf_bart.py

@@ -1716,9 +1716,5 @@ class TFBartForSequenceClassification(TFBartPretrainedModel, TFSequenceClassific
             with tf.name_scope(self.classification_head.name):
                 self.classification_head.build(None)
 
-__all__ = [
-    "TFBartPretrainedModel",
-    "TFBartModel",
-    "TFBartForConditionalGeneration",
-    "TFBartForSequenceClassification"
-]
+
+__all__ = ["TFBartPretrainedModel", "TFBartModel", "TFBartForConditionalGeneration", "TFBartForSequenceClassification"]

src/transformers/models/bart/tokenization_bart.py

@@ -391,6 +391,5 @@ class BartTokenizer(PreTrainedTokenizer):
             text = " " + text
         return (text, kwargs)
 
-__all__ = [
-    "BartTokenizer"
-]
+
+__all__ = ["BartTokenizer"]

src/transformers/models/bart/tokenization_bart_fast.py

@@ -277,6 +277,5 @@ class BartTokenizerFast(PreTrainedTokenizerFast):
             return len(cls + token_ids_0 + sep) * [0]
         return len(cls + token_ids_0 + sep + sep + token_ids_1 + sep) * [0]
 
-__all__ = [
-    "BartTokenizerFast"
-]
+
+__all__ = ["BartTokenizerFast"]

src/transformers/models/barthez/tokenization_barthez.py

@@ -288,6 +288,5 @@ class BarthezTokenizer(PreTrainedTokenizer):
 
         return (out_vocab_file,)
 
-__all__ = [
-    "BarthezTokenizer"
-]
+
+__all__ = ["BarthezTokenizer"]

src/transformers/models/barthez/tokenization_barthez_fast.py

@@ -196,6 +196,5 @@ class BarthezTokenizerFast(PreTrainedTokenizerFast):
 
         return (out_vocab_file,)
 
-__all__ = [
-    "BarthezTokenizerFast"
-]
+
+__all__ = ["BarthezTokenizerFast"]

src/transformers/models/bartpho/tokenization_bartpho.py

@@ -315,6 +315,5 @@ class BartphoTokenizer(PreTrainedTokenizer):
 
         return out_vocab_file, out_monolingual_vocab_file
 
-__all__ = [
-    "BartphoTokenizer"
-]
+
+__all__ = ["BartphoTokenizer"]

src/transformers/models/beit/configuration_beit.py

@@ -230,7 +230,5 @@ class BeitOnnxConfig(OnnxConfig):
     def atol_for_validation(self) -> float:
         return 1e-4
 
-__all__ = [
-    "BeitConfig",
-    "BeitOnnxConfig"
-]
+
+__all__ = ["BeitConfig", "BeitOnnxConfig"]

src/transformers/models/beit/feature_extraction_beit.py

@@ -34,6 +34,5 @@ class BeitFeatureExtractor(BeitImageProcessor):
         )
         super().__init__(*args, **kwargs)
 
-__all__ = [
-    "BeitFeatureExtractor"
-]
+
+__all__ = ["BeitFeatureExtractor"]

src/transformers/models/beit/image_processing_beit.py

@@ -532,6 +532,5 @@ class BeitImageProcessor(BaseImageProcessor):
 
         return semantic_segmentation
 
-__all__ = [
-    "BeitImageProcessor"
-]
+
+__all__ = ["BeitImageProcessor"]

src/transformers/models/beit/modeling_beit.py

@@ -1429,11 +1429,12 @@ class BeitBackbone(BeitPreTrainedModel, BackboneMixin):
             attentions=outputs.attentions,
         )
 
+
 __all__ = [
     "BeitPreTrainedModel",
     "BeitModel",
     "BeitForMaskedImageModeling",
     "BeitForImageClassification",
     "BeitForSemanticSegmentation",
-    "BeitBackbone"
+    "BeitBackbone",
 ]

src/transformers/models/beit/modeling_flax_beit.py

@@ -956,5 +956,5 @@ __all__ = [
     "FlaxBeitPreTrainedModel",
     "FlaxBeitModel",
     "FlaxBeitForMaskedImageModeling",
-    "FlaxBeitForImageClassification"
+    "FlaxBeitForImageClassification",
 ]

src/transformers/models/bert/configuration_bert.py

@@ -152,7 +152,5 @@ class BertOnnxConfig(OnnxConfig):
             ]
         )
 
-__all__ = [
-    "BertConfig",
-    "BertOnnxConfig"
-]
+
+__all__ = ["BertConfig", "BertOnnxConfig"]

src/transformers/models/bert/modeling_bert.py

@@ -2030,6 +2030,7 @@ class BertForQuestionAnswering(BertPreTrainedModel):
             attentions=outputs.attentions,
         )
 
+
 __all__ = [
     "load_tf_weights_in_bert",
     "BertPreTrainedModel",
@@ -2041,5 +2042,5 @@ __all__ = [
     "BertForSequenceClassification",
     "BertForMultipleChoice",
     "BertForTokenClassification",
-    "BertForQuestionAnswering"
+    "BertForQuestionAnswering",
 ]

src/transformers/models/bert/modeling_flax_bert.py

@@ -1733,5 +1733,5 @@ __all__ = [
     "FlaxBertForMultipleChoice",
     "FlaxBertForTokenClassification",
     "FlaxBertForQuestionAnswering",
-    "FlaxBertForCausalLM"
+    "FlaxBertForCausalLM",
 ]

src/transformers/models/bert/modeling_tf_bert.py

@@ -2121,6 +2121,7 @@ class TFBertForQuestionAnswering(TFBertPreTrainedModel, TFQuestionAnsweringLoss)
             with tf.name_scope(self.qa_outputs.name):
                 self.qa_outputs.build([None, None, self.config.hidden_size])
 
+
 __all__ = [
     "TFBertPreTrainedModel",
     "TFBertModel",
@@ -2131,5 +2132,5 @@ __all__ = [
     "TFBertForSequenceClassification",
     "TFBertForMultipleChoice",
     "TFBertForTokenClassification",
-    "TFBertForQuestionAnswering"
+    "TFBertForQuestionAnswering",
 ]

src/transformers/models/bert/tokenization_bert.py

@@ -503,8 +503,5 @@ class WordpieceTokenizer(object):
                 output_tokens.extend(sub_tokens)
         return output_tokens
 
-__all__ = [
-    "BertTokenizer",
-    "BasicTokenizer",
-    "WordpieceTokenizer"
-]
+
+__all__ = ["BertTokenizer", "BasicTokenizer", "WordpieceTokenizer"]

src/transformers/models/bert/tokenization_bert_fast.py

@@ -173,6 +173,5 @@ class BertTokenizerFast(PreTrainedTokenizerFast):
         files = self._tokenizer.model.save(save_directory, name=filename_prefix)
         return tuple(files)
 
-__all__ = [
-    "BertTokenizerFast"
-]
+
+__all__ = ["BertTokenizerFast"]

src/transformers/models/bert_generation/configuration_bert_generation.py

@@ -125,6 +125,5 @@ class BertGenerationConfig(PretrainedConfig):
         self.position_embedding_type = position_embedding_type
         self.use_cache = use_cache
 
-__all__ = [
-    "BertGenerationConfig"
-]
+
+__all__ = ["BertGenerationConfig"]

src/transformers/models/bert_generation/modeling_bert_generation.py

@@ -1024,9 +1024,10 @@ class BertGenerationDecoder(BertGenerationPreTrainedModel):
             )
         return reordered_past
 
+
 __all__ = [
     "load_tf_weights_in_bert_generation",
     "BertGenerationPreTrainedModel",
     "BertGenerationEncoder",
-    "BertGenerationDecoder"
+    "BertGenerationDecoder",
 ]

src/transformers/models/bert_generation/tokenization_bert_generation.py

@@ -174,6 +174,5 @@ class BertGenerationTokenizer(PreTrainedTokenizer):
 
         return (out_vocab_file,)
 
-__all__ = [
-    "BertGenerationTokenizer"
-]
+
+__all__ = ["BertGenerationTokenizer"]

src/transformers/models/bert_japanese/tokenization_bert_japanese.py

@@ -983,8 +983,5 @@ class SentencepieceTokenizer(object):
 
         return new_pieces
 
-__all__ = [
-    "BertJapaneseTokenizer",
-    "CharacterTokenizer",
-    "MecabTokenizer"
-]
+
+__all__ = ["BertJapaneseTokenizer", "CharacterTokenizer", "MecabTokenizer"]

src/transformers/models/bertweet/tokenization_bertweet.py

@@ -768,6 +768,4 @@ def casual_tokenize(text, preserve_case=True, reduce_len=False, strip_handles=Fa
 
 ###############################################################################
 
-__all__ = [
-    "BertweetTokenizer"
-]
+__all__ = ["BertweetTokenizer"]

src/transformers/models/big_bird/configuration_big_bird.py

@@ -174,7 +174,5 @@ class BigBirdOnnxConfig(OnnxConfig):
             ]
         )
 
-__all__ = [
-    "BigBirdConfig",
-    "BigBirdOnnxConfig"
-]
+
+__all__ = ["BigBirdConfig", "BigBirdOnnxConfig"]

src/transformers/models/big_bird/modeling_big_bird.py

@@ -3162,6 +3162,7 @@ class BigBirdForQuestionAnswering(BigBirdPreTrainedModel):
         mask = torch.where(mask < q_lengths, 1, 0)
         return mask
 
+
 __all__ = [
     "load_tf_weights_in_big_bird",
     "BigBirdPreTrainedModel",
@@ -3172,5 +3173,5 @@ __all__ = [
     "BigBirdForSequenceClassification",
     "BigBirdForMultipleChoice",
     "BigBirdForTokenClassification",
-    "BigBirdForQuestionAnswering"
+    "BigBirdForQuestionAnswering",
 ]

src/transformers/models/big_bird/modeling_flax_big_bird.py

@@ -2653,5 +2653,5 @@ __all__ = [
     "FlaxBigBirdForMultipleChoice",
     "FlaxBigBirdForTokenClassification",
     "FlaxBigBirdForQuestionAnswering",
-    "FlaxBigBirdForCausalLM"
+    "FlaxBigBirdForCausalLM",
 ]

src/transformers/models/big_bird/tokenization_big_bird.py

@@ -323,6 +323,5 @@ class BigBirdTokenizer(PreTrainedTokenizer):
             return len(cls + token_ids_0 + sep) * [0]
         return len(cls + token_ids_0 + sep) * [0] + len(token_ids_1 + sep) * [1]
 
-__all__ = [
-    "BigBirdTokenizer"
-]
+
+__all__ = ["BigBirdTokenizer"]

src/transformers/models/big_bird/tokenization_big_bird_fast.py

@@ -231,6 +231,5 @@ class BigBirdTokenizerFast(PreTrainedTokenizerFast):
 
         return (out_vocab_file,)
 
-__all__ = [
-    "BigBirdTokenizerFast"
-]
+
+__all__ = ["BigBirdTokenizerFast"]

src/transformers/models/bigbird_pegasus/configuration_bigbird_pegasus.py

@@ -411,7 +411,5 @@ class BigBirdPegasusOnnxConfig(OnnxSeq2SeqConfigWithPast):
                 flattened_output, name, idx, t
             )
 
-__all__ = [
-    "BigBirdPegasusConfig",
-    "BigBirdPegasusOnnxConfig"
-]
+
+__all__ = ["BigBirdPegasusConfig", "BigBirdPegasusOnnxConfig"]

src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py

@@ -3093,11 +3093,12 @@ class BigBirdPegasusForCausalLM(BigBirdPegasusPreTrainedModel):
             )
         return reordered_past
 
+
 __all__ = [
     "BigBirdPegasusPreTrainedModel",
     "BigBirdPegasusModel",
     "BigBirdPegasusForConditionalGeneration",
     "BigBirdPegasusForSequenceClassification",
     "BigBirdPegasusForQuestionAnswering",
-    "BigBirdPegasusForCausalLM"
+    "BigBirdPegasusForCausalLM",
 ]

src/transformers/models/biogpt/configuration_biogpt.py

@@ -132,6 +132,5 @@ class BioGptConfig(PretrainedConfig):
         self.activation_dropout = activation_dropout
         super().__init__(pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)
 
-__all__ = [
-    "BioGptConfig"
-]
+
+__all__ = ["BioGptConfig"]

src/transformers/models/biogpt/modeling_biogpt.py

@@ -942,10 +942,11 @@ class BioGptForSequenceClassification(BioGptPreTrainedModel):
     def set_input_embeddings(self, value):
         self.biogpt.embed_tokens = value
 
+
 __all__ = [
     "BioGptPreTrainedModel",
     "BioGptModel",
     "BioGptForCausalLM",
     "BioGptForTokenClassification",
-    "BioGptForSequenceClassification"
+    "BioGptForSequenceClassification",
 ]

src/transformers/models/biogpt/tokenization_biogpt.py

@@ -358,6 +358,5 @@ class BioGptTokenizer(PreTrainedTokenizer):
 
         self.sm = sacremoses
 
-__all__ = [
-    "BioGptTokenizer"
-]
+
+__all__ = ["BioGptTokenizer"]

src/transformers/models/bit/configuration_bit.py

@@ -134,6 +134,5 @@ class BitConfig(BackboneConfigMixin, PretrainedConfig):
             out_features=out_features, out_indices=out_indices, stage_names=self.stage_names
         )
 
-__all__ = [
-    "BitConfig"
-]
+
+__all__ = ["BitConfig"]

src/transformers/models/bit/image_processing_bit.py

@@ -346,6 +346,5 @@ class BitImageProcessor(BaseImageProcessor):
         data = {"pixel_values": images}
         return BatchFeature(data=data, tensor_type=return_tensors)
 
-__all__ = [
-    "BitImageProcessor"
-]
+
+__all__ = ["BitImageProcessor"]

src/transformers/models/bit/modeling_bit.py

@@ -900,9 +900,5 @@ class BitBackbone(BitPreTrainedModel, BackboneMixin):
             attentions=None,
         )
 
-__all__ = [
-    "BitPreTrainedModel",
-    "BitModel",
-    "BitForImageClassification",
-    "BitBackbone"
-]
+
+__all__ = ["BitPreTrainedModel", "BitModel", "BitForImageClassification", "BitBackbone"]

src/transformers/models/blenderbot/configuration_blenderbot.py

@@ -394,7 +394,5 @@ class BlenderbotOnnxConfig(OnnxSeq2SeqConfigWithPast):
             inputs_or_outputs[f"{name}.{i}.encoder.key"] = {0: "batch", 2: encoder_sequence}
             inputs_or_outputs[f"{name}.{i}.encoder.value"] = {0: "batch", 2: encoder_sequence}
 
-__all__ = [
-    "BlenderbotConfig",
-    "BlenderbotOnnxConfig"
-]
+
+__all__ = ["BlenderbotConfig", "BlenderbotOnnxConfig"]

src/transformers/models/blenderbot/modeling_blenderbot.py

@@ -1616,9 +1616,10 @@ class BlenderbotForCausalLM(BlenderbotPreTrainedModel):
             )
         return reordered_past
 
+
 __all__ = [
     "BlenderbotPreTrainedModel",
     "BlenderbotModel",
     "BlenderbotForConditionalGeneration",
-    "BlenderbotForCausalLM"
+    "BlenderbotForCausalLM",
 ]

src/transformers/models/blenderbot/modeling_flax_blenderbot.py

@@ -1508,8 +1508,4 @@ append_replace_return_docstrings(
     FlaxBlenderbotForConditionalGeneration, output_type=FlaxSeq2SeqLMOutput, config_class=_CONFIG_FOR_DOC
 )
 
-__all__ = [
-    "FlaxBlenderbotPreTrainedModel",
-    "FlaxBlenderbotModel",
-    "FlaxBlenderbotForConditionalGeneration"
-]
+__all__ = ["FlaxBlenderbotPreTrainedModel", "FlaxBlenderbotModel", "FlaxBlenderbotForConditionalGeneration"]

src/transformers/models/blenderbot/modeling_tf_blenderbot.py

@@ -1559,8 +1559,5 @@ class TFBlenderbotForConditionalGeneration(TFBlenderbotPreTrainedModel, TFCausal
             with tf.name_scope(self.bias_layer.name):
                 self.bias_layer.build(None)
 
-__all__ = [
-    "TFBlenderbotPreTrainedModel",
-    "TFBlenderbotModel",
-    "TFBlenderbotForConditionalGeneration"
-]
+
+__all__ = ["TFBlenderbotPreTrainedModel", "TFBlenderbotModel", "TFBlenderbotForConditionalGeneration"]

src/transformers/models/blenderbot/tokenization_blenderbot.py

@@ -422,6 +422,5 @@ class BlenderbotTokenizer(PreTrainedTokenizer):
             "{{ eos_token }}"
         )
 
-__all__ = [
-    "BlenderbotTokenizer"
-]
+
+__all__ = ["BlenderbotTokenizer"]

src/transformers/models/blenderbot/tokenization_blenderbot_fast.py

@@ -304,6 +304,5 @@ class BlenderbotTokenizerFast(PreTrainedTokenizerFast):
             "{{ eos_token }}"
         )
 
-__all__ = [
-    "BlenderbotTokenizerFast"
-]
+
+__all__ = ["BlenderbotTokenizerFast"]

src/transformers/models/blenderbot_small/configuration_blenderbot_small.py

@@ -389,7 +389,5 @@ class BlenderbotSmallOnnxConfig(OnnxSeq2SeqConfigWithPast):
                 flattened_output, name, idx, t
             )
 
-__all__ = [
-    "BlenderbotSmallConfig",
-    "BlenderbotSmallOnnxConfig"
-]
+
+__all__ = ["BlenderbotSmallConfig", "BlenderbotSmallOnnxConfig"]

src/transformers/models/blenderbot_small/modeling_blenderbot_small.py

@@ -1568,9 +1568,10 @@ class BlenderbotSmallForCausalLM(BlenderbotSmallPreTrainedModel):
             )
         return reordered_past
 
+
 __all__ = [
     "BlenderbotSmallPreTrainedModel",
     "BlenderbotSmallModel",
     "BlenderbotSmallForConditionalGeneration",
-    "BlenderbotSmallForCausalLM"
+    "BlenderbotSmallForCausalLM",
 ]

src/transformers/models/blenderbot_small/modeling_flax_blenderbot_small.py

@@ -1528,5 +1528,5 @@ append_replace_return_docstrings(
 __all__ = [
     "FlaxBlenderbotSmallPreTrainedModel",
     "FlaxBlenderbotSmallModel",
-    "FlaxBlenderbotSmallForConditionalGeneration"
+    "FlaxBlenderbotSmallForConditionalGeneration",
 ]

src/transformers/models/blenderbot_small/modeling_tf_blenderbot_small.py

@@ -1529,8 +1529,5 @@ class TFBlenderbotSmallForConditionalGeneration(TFBlenderbotSmallPreTrainedModel
             with tf.name_scope(self.bias_layer.name):
                 self.bias_layer.build(None)
 
-__all__ = [
-    "TFBlenderbotSmallPreTrainedModel",
-    "TFBlenderbotSmallModel",
-    "TFBlenderbotSmallForConditionalGeneration"
-]
+
+__all__ = ["TFBlenderbotSmallPreTrainedModel", "TFBlenderbotSmallModel", "TFBlenderbotSmallForConditionalGeneration"]

src/transformers/models/blenderbot_small/tokenization_blenderbot_small.py

@@ -235,6 +235,5 @@ class BlenderbotSmallTokenizer(PreTrainedTokenizer):
             "{{ eos_token }}"
         )
 
-__all__ = [
-    "BlenderbotSmallTokenizer"
-]
+
+__all__ = ["BlenderbotSmallTokenizer"]

src/transformers/models/blenderbot_small/tokenization_blenderbot_small_fast.py

@@ -115,6 +115,5 @@ class BlenderbotSmallTokenizerFast(PreTrainedTokenizerFast):
             "{{ eos_token }}"
         )
 
-__all__ = [
-    "BlenderbotSmallTokenizerFast"
-]
+
+__all__ = ["BlenderbotSmallTokenizerFast"]

src/transformers/models/blip/configuration_blip.py

@@ -365,8 +365,5 @@ class BlipConfig(PretrainedConfig):
 
         return cls(text_config=text_config.to_dict(), vision_config=vision_config.to_dict(), **kwargs)
 
-__all__ = [
-    "BlipTextConfig",
-    "BlipVisionConfig",
-    "BlipConfig"
-]
+
+__all__ = ["BlipTextConfig", "BlipVisionConfig", "BlipConfig"]

src/transformers/models/blip/image_processing_blip.py

@@ -313,6 +313,5 @@ class BlipImageProcessor(BaseImageProcessor):
 
         return encoded_outputs
 
-__all__ = [
-    "BlipImageProcessor"
-]
+
+__all__ = ["BlipImageProcessor"]

src/transformers/models/blip/modeling_blip.py

@@ -1489,11 +1489,12 @@ class BlipForImageTextRetrieval(BlipPreTrainedModel):
             question_embeds=question_embeds,
         )
 
+
 __all__ = [
     "BlipPreTrainedModel",
     "BlipVisionModel",
     "BlipModel",
     "BlipForConditionalGeneration",
     "BlipForQuestionAnswering",
-    "BlipForImageTextRetrieval"
+    "BlipForImageTextRetrieval",
 ]

src/transformers/models/blip/modeling_blip_text.py

@@ -951,6 +951,5 @@ class BlipTextLMHeadModel(BlipTextPreTrainedModel):
             )
         return reordered_past
 
-__all__ = [
-    "BlipTextModel"
-]
+
+__all__ = ["BlipTextModel"]

src/transformers/models/blip/modeling_tf_blip.py

@@ -1704,11 +1704,12 @@ class TFBlipForImageTextRetrieval(TFBlipPreTrainedModel):
             with tf.name_scope(self.itm_head.name):
                 self.itm_head.build([None, None, self.config.text_config.hidden_size])
 
+
 __all__ = [
     "TFBlipPreTrainedModel",
     "TFBlipVisionModel",
     "TFBlipModel",
     "TFBlipForConditionalGeneration",
     "TFBlipForQuestionAnswering",
-    "TFBlipForImageTextRetrieval"
+    "TFBlipForImageTextRetrieval",
 ]

src/transformers/models/blip/modeling_tf_blip_text.py

@@ -1123,6 +1123,5 @@ class TFBlipTextLMHeadModel(TFBlipTextPreTrainedModel):
             with tf.name_scope(self.cls.name):
                 self.cls.build(None)
 
-__all__ = [
-    "TFBlipTextModel"
-]
+
+__all__ = ["TFBlipTextModel"]

src/transformers/models/blip/processing_blip.py

@@ -151,6 +151,5 @@ class BlipProcessor(ProcessorMixin):
         image_processor_input_names = self.image_processor.model_input_names
         return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names))
 
-__all__ = [
-    "BlipProcessor"
-]
+
+__all__ = ["BlipProcessor"]

src/transformers/models/blip_2/configuration_blip_2.py

@@ -355,8 +355,5 @@ class Blip2Config(PretrainedConfig):
             **kwargs,
         )
 
-__all__ = [
-    "Blip2VisionConfig",
-    "Blip2QFormerConfig",
-    "Blip2Config"
-]
+
+__all__ = ["Blip2VisionConfig", "Blip2QFormerConfig", "Blip2Config"]

src/transformers/models/blip_2/modeling_blip_2.py

@@ -1863,10 +1863,11 @@ class Blip2ForConditionalGeneration(Blip2PreTrainedModel):
                 outputs = torch.cat([bos_tokens, outputs], dim=-1)
         return outputs
 
+
 __all__ = [
     "Blip2PreTrainedModel",
     "Blip2VisionModel",
     "Blip2QFormerModel",
     "Blip2Model",
-    "Blip2ForConditionalGeneration"
+    "Blip2ForConditionalGeneration",
 ]

src/transformers/models/blip_2/processing_blip_2.py

@@ -156,6 +156,5 @@ class Blip2Processor(ProcessorMixin):
         image_processor_input_names = self.image_processor.model_input_names
         return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names))
 
-__all__ = [
-    "Blip2Processor"
-]
+
+__all__ = ["Blip2Processor"]

src/transformers/models/bloom/configuration_bloom.py

@@ -236,7 +236,5 @@ class BloomOnnxConfig(OnnxConfigWithPast):
     def default_onnx_opset(self) -> int:
         return 13
 
-__all__ = [
-    "BloomConfig",
-    "BloomOnnxConfig"
-]
+
+__all__ = ["BloomConfig", "BloomOnnxConfig"]

src/transformers/models/bloom/modeling_bloom.py

@@ -1246,11 +1246,12 @@ class BloomForQuestionAnswering(BloomPreTrainedModel):
             attentions=outputs.attentions,
         )
 
+
 __all__ = [
     "BloomPreTrainedModel",
     "BloomModel",
     "BloomForCausalLM",
     "BloomForSequenceClassification",
     "BloomForTokenClassification",
-    "BloomForQuestionAnswering"
+    "BloomForQuestionAnswering",
 ]

src/transformers/models/bloom/modeling_flax_bloom.py

@@ -737,8 +737,4 @@ class FlaxBloomForCausalLM(FlaxBloomPreTrainedModel):
 
 append_call_sample_docstring(FlaxBloomForCausalLM, _CHECKPOINT_FOR_DOC, FlaxCausalLMOutput, _CONFIG_FOR_DOC)
 
-__all__ = [
-    "FlaxBloomPreTrainedModel",
-    "FlaxBloomModel",
-    "FlaxBloomForCausalLM"
-]
+__all__ = ["FlaxBloomPreTrainedModel", "FlaxBloomModel", "FlaxBloomForCausalLM"]

src/transformers/models/bloom/tokenization_bloom_fast.py

@@ -159,6 +159,5 @@ class BloomTokenizerFast(PreTrainedTokenizerFast):
         """
         return "{% for message in messages %}" "{{ message.content }}{{ eos_token }}" "{% endfor %}"
 
-__all__ = [
-    "BloomTokenizerFast"
-]
+
+__all__ = ["BloomTokenizerFast"]

src/transformers/models/bridgetower/configuration_bridgetower.py

@@ -349,8 +349,5 @@ class BridgeTowerConfig(PretrainedConfig):
 
         return cls(text_config=text_config.to_dict(), vision_config=vision_config.to_dict(), **kwargs)
 
-__all__ = [
-    "BridgeTowerVisionConfig",
-    "BridgeTowerTextConfig",
-    "BridgeTowerConfig"
-]
+
+__all__ = ["BridgeTowerVisionConfig", "BridgeTowerTextConfig", "BridgeTowerConfig"]

src/transformers/models/bridgetower/image_processing_bridgetower.py

@@ -562,6 +562,5 @@ class BridgeTowerImageProcessor(BaseImageProcessor):
 
         return encoded_outputs
 
-__all__ = [
-    "BridgeTowerImageProcessor"
-]
+
+__all__ = ["BridgeTowerImageProcessor"]

src/transformers/models/bridgetower/modeling_bridgetower.py

@@ -1912,10 +1912,11 @@ class BridgeTowerForContrastiveLearning(BridgeTowerPreTrainedModel):
             attentions=outputs.attentions,
         )
 
+
 __all__ = [
     "BridgeTowerPreTrainedModel",
     "BridgeTowerModel",
     "BridgeTowerForMaskedLM",
     "BridgeTowerForImageAndTextRetrieval",
-    "BridgeTowerForContrastiveLearning"
+    "BridgeTowerForContrastiveLearning",
 ]

src/transformers/models/bridgetower/processing_bridgetower.py

@@ -120,6 +120,5 @@ class BridgeTowerProcessor(ProcessorMixin):
         image_processor_input_names = self.image_processor.model_input_names
         return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names))
 
-__all__ = [
-    "BridgeTowerProcessor"
-]
+
+__all__ = ["BridgeTowerProcessor"]

src/transformers/models/bros/configuration_bros.py

@@ -136,6 +136,5 @@ class BrosConfig(PretrainedConfig):
         self.dim_bbox_projection = self.hidden_size // self.num_attention_heads
         self.classifier_dropout_prob = classifier_dropout_prob
 
-__all__ = [
-    "BrosConfig"
-]
+
+__all__ = ["BrosConfig"]

src/transformers/models/bros/modeling_bros.py

@@ -1320,10 +1320,11 @@ class BrosSpadeELForTokenClassification(BrosPreTrainedModel):
             attentions=outputs.attentions,
         )
 
+
 __all__ = [
     "BrosPreTrainedModel",
     "BrosModel",
     "BrosForTokenClassification",
     "BrosSpadeEEForTokenClassification",
-    "BrosSpadeELForTokenClassification"
+    "BrosSpadeELForTokenClassification",
 ]

src/transformers/models/bros/processing_bros.py

@@ -110,6 +110,5 @@ class BrosProcessor(ProcessorMixin):
         tokenizer_input_names = self.tokenizer.model_input_names
         return list(dict.fromkeys(tokenizer_input_names))
 
-__all__ = [
-    "BrosProcessor"
-]
+
+__all__ = ["BrosProcessor"]

src/transformers/models/byt5/tokenization_byt5.py

@@ -235,6 +235,5 @@ class ByT5Tokenizer(PreTrainedTokenizer):
     def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
         return ()
 
-__all__ = [
-    "ByT5Tokenizer"
-]
+
+__all__ = ["ByT5Tokenizer"]

src/transformers/models/camembert/configuration_camembert.py

@@ -154,7 +154,5 @@ class CamembertOnnxConfig(OnnxConfig):
             ]
         )
 
-__all__ = [
-    "CamembertConfig",
-    "CamembertOnnxConfig"
-]
+
+__all__ = ["CamembertConfig", "CamembertOnnxConfig"]

src/transformers/models/camembert/modeling_camembert.py

@@ -1583,6 +1583,7 @@ def create_position_ids_from_input_ids(input_ids, padding_idx, past_key_values_l
     incremental_indices = (torch.cumsum(mask, dim=1).type_as(mask) + past_key_values_length) * mask
     return incremental_indices.long() + padding_idx
 
+
 __all__ = [
     "CamembertPreTrainedModel",
     "CamembertModel",
@@ -1591,5 +1592,5 @@ __all__ = [
     "CamembertForMultipleChoice",
     "CamembertForTokenClassification",
     "CamembertForQuestionAnswering",
-    "CamembertForCausalLM"
+    "CamembertForCausalLM",
 ]

src/transformers/models/camembert/modeling_tf_camembert.py

@@ -1798,6 +1798,7 @@ class TFCamembertForCausalLM(TFCamembertPreTrainedModel, TFCausalLanguageModelin
             with tf.name_scope(self.lm_head.name):
                 self.lm_head.build(None)
 
+
 __all__ = [
     "TFCamembertPreTrainedModel",
     "TFCamembertModel",
@@ -1806,5 +1807,5 @@ __all__ = [
     "TFCamembertForTokenClassification",
     "TFCamembertForMultipleChoice",
     "TFCamembertForQuestionAnswering",
-    "TFCamembertForCausalLM"
+    "TFCamembertForCausalLM",
 ]

src/transformers/models/camembert/tokenization_camembert.py

@@ -320,6 +320,5 @@ class CamembertTokenizer(PreTrainedTokenizer):
             return len(cls + token_ids_0 + sep) * [0]
         return len(cls + token_ids_0 + sep + sep + token_ids_1 + sep) * [0]
 
-__all__ = [
-    "CamembertTokenizer"
-]
+
+__all__ = ["CamembertTokenizer"]