TF Longformer

docs/source/model_doc/longformer.rst

@@ -102,3 +102,25 @@ LongformerForQuestionAnswering
 
 .. autoclass:: transformers.LongformerForQuestionAnswering
     :members:
+
+
+TFLongformerModel
+~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.TFLongformerModel
+    :members:
+
+
+TFLongformerForMaskedLM
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.TFLongformerForMaskedLM
+    :members:
+
+
+TFLongformerForQuestionAnswering
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: transformers.TFLongformerForQuestionAnswering
+    :members:
+

src/transformers/__init__.py

@@ -399,6 +399,7 @@
         LongformerForMultipleChoice,
         LongformerForTokenClassification,
         LongformerForQuestionAnswering,
+        LongformerSelfAttention,
         LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
     )
 
@@ -568,6 +569,14 @@
         TFGPT2PreTrainedModel,
     )
 
+    from .modeling_tf_longformer import (
+        TF_LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
+        TFLongformerModel,
+        TFLongformerForMaskedLM,
+        TFLongformerForQuestionAnswering,
+        TFLongformerSelfAttention,
+    )
+
     from .modeling_tf_mobilebert import (
         TF_MOBILEBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
         TFMobileBertModel,

src/transformers/modeling_tf_auto.py

@@ -29,6 +29,7 @@
     ElectraConfig,
     FlaubertConfig,
     GPT2Config,
+    LongformerConfig,
     MobileBertConfig,
     OpenAIGPTConfig,
     RobertaConfig,
@@ -93,6 +94,7 @@
     TFFlaubertWithLMHeadModel,
 )
 from .modeling_tf_gpt2 import TFGPT2LMHeadModel, TFGPT2Model
+from .modeling_tf_longformer import TFLongformerForMaskedLM, TFLongformerForQuestionAnswering, TFLongformerModel
 from .modeling_tf_mobilebert import (
     TFMobileBertForMaskedLM,
     TFMobileBertForMultipleChoice,
@@ -149,6 +151,7 @@
         (AlbertConfig, TFAlbertModel),
         (CamembertConfig, TFCamembertModel),
         (XLMRobertaConfig, TFXLMRobertaModel),
+        (LongformerConfig, TFLongformerModel),
         (RobertaConfig, TFRobertaModel),
         (BertConfig, TFBertModel),
         (OpenAIGPTConfig, TFOpenAIGPTModel),
@@ -191,6 +194,7 @@
         (AlbertConfig, TFAlbertForMaskedLM),
         (CamembertConfig, TFCamembertForMaskedLM),
         (XLMRobertaConfig, TFXLMRobertaForMaskedLM),
+        (LongformerConfig, TFLongformerForMaskedLM),
         (RobertaConfig, TFRobertaForMaskedLM),
         (BertConfig, TFBertForMaskedLM),
         (OpenAIGPTConfig, TFOpenAIGPTLMHeadModel),
@@ -226,6 +230,7 @@
         (AlbertConfig, TFAlbertForMaskedLM),
         (CamembertConfig, TFCamembertForMaskedLM),
         (XLMRobertaConfig, TFXLMRobertaForMaskedLM),
+        (LongformerConfig, TFLongformerForMaskedLM),
         (RobertaConfig, TFRobertaForMaskedLM),
         (BertConfig, TFBertForMaskedLM),
         (MobileBertConfig, TFMobileBertForMaskedLM),
@@ -259,6 +264,7 @@
         (AlbertConfig, TFAlbertForQuestionAnswering),
         (CamembertConfig, TFCamembertForQuestionAnswering),
         (XLMRobertaConfig, TFXLMRobertaForQuestionAnswering),
+        (LongformerConfig, TFLongformerForQuestionAnswering),
         (RobertaConfig, TFRobertaForQuestionAnswering),
         (BertConfig, TFBertForQuestionAnswering),
         (XLNetConfig, TFXLNetForQuestionAnsweringSimple),