Fixes

src/transformers/models/bart/modeling_tf_bart.py

@@ -1269,6 +1269,7 @@ def build(self, input_shape=None):
  "The bare BART Model outputting raw hidden-states without any specific head on top.",
  BART_START_DOCSTRING,
 )
+@register(backends=("tf",))
 class TFBartModel(TFBartPretrainedModel):
  _requires_load_weight_prefix = True
 
@@ -1382,6 +1383,7 @@ def call(self, x):
  "The BART Model with a language modeling head. Can be used for summarization.",
  BART_START_DOCSTRING,
 )
+@register(backends=("tf",))
 class TFBartForConditionalGeneration(TFBartPretrainedModel, TFCausalLanguageModelingLoss):
  _keys_to_ignore_on_load_missing = [r"final_logits_bias"]
  _requires_load_weight_prefix = True
@@ -1582,6 +1584,7 @@ def build(self, input_shape=None):
  """,
  BART_START_DOCSTRING,
 )
+@register(backends=("tf",))
 class TFBartForSequenceClassification(TFBartPretrainedModel, TFSequenceClassificationLoss):
  def __init__(self, config: BartConfig, load_weight_prefix=None, *inputs, **kwargs):
  super().__init__(config, *inputs, **kwargs)
@@ -1714,5 +1717,8 @@ def build(self, input_shape=None):
  self.classification_head.build(None)
 
 __all__ = [
- "TFBartPretrainedModel"
+ "TFBartPretrainedModel",
+ "TFBartModel",
+ "TFBartForConditionalGeneration",
+ "TFBartForSequenceClassification"
 ]

src/transformers/models/bert_japanese/tokenization_bert_japanese.py

@@ -375,6 +375,7 @@ def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] =
  return (vocab_file,)
 
 
+@register()
 class MecabTokenizer:
  """Runs basic tokenization with MeCab morphological parser."""
 
@@ -647,6 +648,7 @@ def tokenize(self, text, never_split=None, **kwargs):
  return tokens
 
 
+@register()
 class CharacterTokenizer:
  """Runs Character tokenization."""
 
@@ -982,5 +984,7 @@ def tokenize(self, text):
  return new_pieces
 
 __all__ = [
- "BertJapaneseTokenizer"
+ "BertJapaneseTokenizer",
+ "CharacterTokenizer",
+ "MecabTokenizer"
 ]

src/transformers/models/dpr/modeling_dpr.py

@@ -103,6 +103,7 @@ class DPRQuestionEncoderOutput(ModelOutput):
 
 
 @dataclass
+@register(backends=("torch",))
 class DPRReaderOutput(ModelOutput):
  """
  Class for outputs of [`DPRQuestionEncoder`].
@@ -428,6 +429,7 @@ class DPRPretrainedReader(DPRPreTrainedModel):
  "The bare DPRContextEncoder transformer outputting pooler outputs as context representations.",
  DPR_START_DOCSTRING,
 )
+@register(backends=("torch",))
 class DPRContextEncoder(DPRPretrainedContextEncoder):
  def __init__(self, config: DPRConfig):
  super().__init__(config)
@@ -509,6 +511,7 @@ def forward(
  "The bare DPRQuestionEncoder transformer outputting pooler outputs as question representations.",
  DPR_START_DOCSTRING,
 )
+@register(backends=("torch",))
 class DPRQuestionEncoder(DPRPretrainedQuestionEncoder):
  def __init__(self, config: DPRConfig):
  super().__init__(config)
@@ -591,6 +594,7 @@ def forward(
  "The bare DPRReader transformer outputting span predictions.",
  DPR_START_DOCSTRING,
 )
+@register(backends=("torch",))
 class DPRReader(DPRPretrainedReader):
  def __init__(self, config: DPRConfig):
  super().__init__(config)
@@ -666,5 +670,9 @@ def forward(
  "DPRPreTrainedModel",
  "DPRPretrainedContextEncoder",
  "DPRPretrainedQuestionEncoder",
- "DPRPretrainedReader"
+ "DPRPretrainedReader",
+ "DPRContextEncoder",
+ "DPRQuestionEncoder",
+ "DPRReader",
+ "DPRReaderOutput"
 ]

src/transformers/models/dpr/modeling_tf_dpr.py

@@ -539,6 +539,7 @@ class TFDPRPretrainedReader(TFPreTrainedModel):
  "The bare DPRContextEncoder transformer outputting pooler outputs as context representations.",
  TF_DPR_START_DOCSTRING,
 )
+@register(backends=("tf",))
 class TFDPRContextEncoder(TFDPRPretrainedContextEncoder):
  def __init__(self, config: DPRConfig, *args, **kwargs):
  super().__init__(config, *args, **kwargs)
@@ -628,6 +629,7 @@ def build(self, input_shape=None):
  "The bare DPRQuestionEncoder transformer outputting pooler outputs as question representations.",
  TF_DPR_START_DOCSTRING,
 )
+@register(backends=("tf",))
 class TFDPRQuestionEncoder(TFDPRPretrainedQuestionEncoder):
  def __init__(self, config: DPRConfig, *args, **kwargs):
  super().__init__(config, *args, **kwargs)
@@ -716,6 +718,7 @@ def build(self, input_shape=None):
  "The bare DPRReader transformer outputting span predictions.",
  TF_DPR_START_DOCSTRING,
 )
+@register(backends=("tf",))
 class TFDPRReader(TFDPRPretrainedReader):
  def __init__(self, config: DPRConfig, *args, **kwargs):
  super().__init__(config, *args, **kwargs)
@@ -796,5 +799,8 @@ def build(self, input_shape=None):
 __all__ = [
  "TFDPRPretrainedContextEncoder",
  "TFDPRPretrainedQuestionEncoder",
- "TFDPRPretrainedReader"
+ "TFDPRPretrainedReader",
+ "TFDPRContextEncoder",
+ "TFDPRQuestionEncoder",
+ "TFDPRReader"
 ]

src/transformers/models/efficientformer/__init__.py

@@ -20,6 +20,7 @@
 if TYPE_CHECKING:
  from .configuration_efficientformer import *
  from .image_processing_efficientformer import *
+ from .modeling_efficientformer import *
  from .modeling_tf_efficientformer import *
 else:
  import sys

src/transformers/models/efficientformer/modeling_efficientformer.py

@@ -34,7 +34,7 @@
  logging,
 )
 from .configuration_efficientformer import EfficientFormerConfig
-
+from ...utils.import_utils import register
 
 logger = logging.get_logger(__name__)
 
@@ -496,6 +496,7 @@ def forward(
  )
 
 
+@register(backends=('torch',))
 class EfficientFormerPreTrainedModel(PreTrainedModel):
  """
  An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
@@ -548,6 +549,7 @@ def _init_weights(self, module: nn.Module):
  "The bare EfficientFormer Model transformer outputting raw hidden-states without any specific head on top.",
  EFFICIENTFORMER_START_DOCSTRING,
 )
+@register(backends=('torch',))
 class EfficientFormerModel(EfficientFormerPreTrainedModel):
  def __init__(self, config: EfficientFormerConfig):
  super().__init__(config)
@@ -611,6 +613,7 @@ def forward(
  """,
  EFFICIENTFORMER_START_DOCSTRING,
 )
+@register(backends=('torch',))
 class EfficientFormerForImageClassification(EfficientFormerPreTrainedModel):
  def __init__(self, config: EfficientFormerConfig):
  super().__init__(config)
@@ -741,6 +744,7 @@ class token).
  """,
  EFFICIENTFORMER_START_DOCSTRING,
 )
+@register(backends=('torch',))
 class EfficientFormerForImageClassificationWithTeacher(EfficientFormerPreTrainedModel):
  def __init__(self, config: EfficientFormerConfig):
  super().__init__(config)
@@ -799,3 +803,11 @@ def forward(
  hidden_states=outputs.hidden_states,
  attentions=outputs.attentions,
  )
+
+
+__all__ = [
+ "EfficientFormerPreTrainedModel",
+ "EfficientFormerModel",
+ "EfficientFormerForImageClassification",
+ "EfficientFormerForImageClassificationWithTeacher"
+]

src/transformers/models/fastspeech2_conformer/__init__.py

@@ -20,6 +20,7 @@
 if TYPE_CHECKING:
  from .configuration_fastspeech2_conformer import *
  from .modeling_fastspeech2_conformer import *
+ from .tokenization_fastspeech2_conformer import *
 else:
  import sys
 

src/transformers/models/fastspeech2_conformer/tokenization_fastspeech2_conformer.py

@@ -21,13 +21,14 @@
 
 from ...tokenization_utils import PreTrainedTokenizer
 from ...utils import logging, requires_backends
-
+from ...utils.import_utils import register
 
 logger = logging.get_logger(__name__)
 
 VOCAB_FILES_NAMES = {"vocab_file": "vocab.json"}
 
 
+@register()
 class FastSpeech2ConformerTokenizer(PreTrainedTokenizer):
  """
  Construct a FastSpeech2Conformer tokenizer.
@@ -182,3 +183,8 @@ def __setstate__(self, d):
  "You need to install g2p-en to use FastSpeech2ConformerTokenizer. "
  "See https://pypi.org/project/g2p-en/ for installation."
  )
+
+
+__all__ = [
+ "FastSpeech2ConformerTokenizer"
+]

src/transformers/models/fsmt/modeling_fsmt.py

@@ -1031,6 +1031,7 @@ def _get_shape(t):
  "The bare FSMT Model outputting raw hidden-states without any specific head on top.",
  FSMT_START_DOCSTRING,
 )
+@register(backends=("torch",))
 class FSMTModel(PretrainedFSMTModel):
  _tied_weights_keys = ["decoder.embed_tokens.weight", "decoder.output_projection.weight"]
 
@@ -1172,6 +1173,7 @@ def set_output_embeddings(self, value):
 @add_start_docstrings(
  "The FSMT Model with a language modeling head. Can be used for summarization.", FSMT_START_DOCSTRING
 )
+@register(backends=("torch",))
 class FSMTForConditionalGeneration(PretrainedFSMTModel):
  base_model_prefix = "model"
  _tied_weights_keys = ["decoder.embed_tokens.weight", "decoder.output_projection.weight"]
@@ -1388,5 +1390,7 @@ def forward(
  return super().forward(positions)
 
 __all__ = [
- "PretrainedFSMTModel"
+ "PretrainedFSMTModel",
+ "FSMTModel",
+ "FSMTForConditionalGeneration"
 ]

src/transformers/models/mt5/tokenization_mt5.py

@@ -0,0 +1,23 @@
+# coding=utf-8
+# Copyright 2020, The T5 Authors and HuggingFace Inc.
+#
+# 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.
+""" mT5 tokenization file"""
+from transformers import T5Tokenizer
+from transformers.utils.import_utils import register
+
+
+@register(backends=("sentencepiece",))
+class MT5Tokenizer(T5Tokenizer):
+ pass
+

src/transformers/models/mt5/tokenization_mt5_fast.py

@@ -0,0 +1,23 @@
+# coding=utf-8
+# Copyright 2020, The T5 Authors and HuggingFace Inc.
+#
+# 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.
+""" mT5 fast tokenization file"""
+from transformers import T5TokenizerFast
+from transformers.utils.import_utils import register
+
+
+@register(backends=("tokenizers",))
+class MT5TokenizerFast(T5TokenizerFast):
+ pass
+

src/transformers/models/nllb_moe/modeling_nllb_moe.py

@@ -211,6 +211,7 @@ def create_position_ids_from_inputs_embeds(self, inputs_embeds, past_key_values_
  return position_ids.unsqueeze(0).expand(input_shape).contiguous() + past_key_values_length
 
 
+@register(backends=("torch",))
 class NllbMoeTop2Router(nn.Module):
  """
  Router using tokens choose top-2 experts assignment.
@@ -388,6 +389,7 @@ def forward(self, hidden_states):
  return hidden_states
 
 
+@register(backends=("torch",))
 class NllbMoeSparseMLP(nn.Module):
  r"""
  Implementation of the NLLB-MoE sparse MLP module.
@@ -1796,5 +1798,7 @@ def _reorder_cache(past_key_values, beam_idx):
 __all__ = [
  "NllbMoePreTrainedModel",
  "NllbMoeModel",
- "NllbMoeForConditionalGeneration"
+ "NllbMoeForConditionalGeneration",
+ "NllbMoeSparseMLP",
+ "NllbMoeTop2Router",
 ]

src/transformers/models/pvt_v2/modeling_pvt_v2.py

@@ -631,6 +631,7 @@ def forward(
  """,
  PVT_V2_START_DOCSTRING,
 )
+@register(backends=("torch",))
 class PvtV2Backbone(PvtV2Model, BackboneMixin):
  def __init__(self, config: PvtV2Config):
  super().__init__(config)
@@ -706,5 +707,6 @@ def forward(
 __all__ = [
  "PvtV2PreTrainedModel",
  "PvtV2Model",
- "PvtV2ForImageClassification"
+ "PvtV2ForImageClassification",
+ "PvtV2Backbone"
 ]

src/transformers/models/rag/__init__.py

@@ -22,6 +22,7 @@
  from .modeling_rag import *
  from .modeling_tf_rag import *
  from .tokenization_rag import *
+ from .retrieval_rag import *
 else:
  import sys
 

src/transformers/models/rag/retrieval_rag.py

@@ -26,7 +26,7 @@
 from ...utils import cached_file, is_datasets_available, is_faiss_available, logging, requires_backends, strtobool
 from .configuration_rag import RagConfig
 from .tokenization_rag import RagTokenizer
-
+from ...utils.import_utils import register
 
 if is_datasets_available():
  from datasets import Dataset, load_dataset, load_from_disk
@@ -341,6 +341,7 @@ def init_index(self):
  self._index_initialized = True
 
 
+@register()
 class RagRetriever:
  """
  Retriever used to get documents from vector queries. It retrieves the documents embeddings as well as the documents
@@ -672,3 +673,8 @@ def __call__(
  },
  tensor_type=return_tensors,
  )
+
+
+__all__ = [
+ "RagRetriever"
+]

src/transformers/models/realm/__init__.py

@@ -22,6 +22,7 @@
  from .modeling_realm import *
  from .tokenization_realm import *
  from .tokenization_realm_fast import *
+ from .retrieval_realm import *
 else:
  import sys