Longformer (#4352)

* first commit

* bug fixes

* better examples

* undo padding

* remove wrong VOCAB_FILES_NAMES

* License

* make style

* make isort happy

* unit tests

* integration test

* make `black` happy by undoing `isort` changes!!

* lint

* no need for the padding value

* batch_size not bsz

* remove unused type casting

* seqlen not seq_len

* staticmethod

* `bert` selfattention instead of `n2`

* uint8 instead of bool + lints

* pad inputs_embeds using embeddings not a constant

* black

* unit test with padding

* fix unit tests

* remove redundant unit test

* upload model weights

* resolve todo

* simpler _mask_invalid_locations without lru_cache + backward compatible masked_fill_

* increase unittest coverage

src/transformers/tokenization_auto.py

@@ -29,6 +29,7 @@ from .configuration_auto import (
     ElectraConfig,
     FlaubertConfig,
     GPT2Config,
+    LongformerConfig,
     OpenAIGPTConfig,
     ReformerConfig,
     RobertaConfig,
@@ -50,6 +51,7 @@ from .tokenization_distilbert import DistilBertTokenizer, DistilBertTokenizerFas
 from .tokenization_electra import ElectraTokenizer, ElectraTokenizerFast
 from .tokenization_flaubert import FlaubertTokenizer
 from .tokenization_gpt2 import GPT2Tokenizer, GPT2TokenizerFast
+from .tokenization_longformer import LongformerTokenizer
 from .tokenization_marian import MarianTokenizer
 from .tokenization_openai import OpenAIGPTTokenizer, OpenAIGPTTokenizerFast
 from .tokenization_reformer import ReformerTokenizer
@@ -73,6 +75,7 @@ TOKENIZER_MAPPING = OrderedDict(
         (XLMRobertaConfig, (XLMRobertaTokenizer, None)),
         (MarianConfig, (MarianTokenizer, None)),
         (BartConfig, (BartTokenizer, None)),
+        (LongformerConfig, (LongformerTokenizer, None)),
         (RobertaConfig, (RobertaTokenizer, RobertaTokenizerFast)),
         (ReformerConfig, (ReformerTokenizer, None)),
         (ElectraConfig, (ElectraTokenizer, ElectraTokenizerFast)),
@@ -105,6 +108,7 @@ class AutoTokenizer:
             - contains `albert`: AlbertTokenizer (ALBERT model)
             - contains `camembert`: CamembertTokenizer (CamemBERT model)
             - contains `xlm-roberta`: XLMRobertaTokenizer (XLM-RoBERTa model)
+            - contains `longformer`: LongformerTokenizer (AllenAI Longformer model)
             - contains `roberta`: RobertaTokenizer (RoBERTa model)
             - contains `bert`: BertTokenizer (Bert model)
             - contains `openai-gpt`: OpenAIGPTTokenizer (OpenAI GPT model)
@@ -136,6 +140,7 @@ class AutoTokenizer:
             - contains `albert`: AlbertTokenizer (ALBERT model)
             - contains `camembert`: CamembertTokenizer (CamemBERT model)
             - contains `xlm-roberta`: XLMRobertaTokenizer (XLM-RoBERTa model)
+            - contains `longformer`: LongformerTokenizer (AllenAI Longformer model)
             - contains `roberta`: RobertaTokenizer (RoBERTa model)
             - contains `bert-base-japanese`: BertJapaneseTokenizer (Bert model)
             - contains `bert`: BertTokenizer (Bert model)