AutoTokenizer supports mbart-large-en-ro (#5121)

src/transformers/configuration_auto.py

@@ -19,7 +19,7 @@ import logging
 from collections import OrderedDict
 
 from .configuration_albert import ALBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, AlbertConfig
-from .configuration_bart import BART_PRETRAINED_CONFIG_ARCHIVE_MAP, BartConfig
+from .configuration_bart import BART_PRETRAINED_CONFIG_ARCHIVE_MAP, BartConfig, MBartConfig
 from .configuration_bert import BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, BertConfig
 from .configuration_camembert import CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, CamembertConfig
 from .configuration_ctrl import CTRL_PRETRAINED_CONFIG_ARCHIVE_MAP, CTRLConfig
@@ -80,6 +80,7 @@ CONFIG_MAPPING = OrderedDict(
         ("camembert", CamembertConfig,),
         ("xlm-roberta", XLMRobertaConfig,),
         ("marian", MarianConfig,),
+        ("mbart", MBartConfig,),
         ("bart", BartConfig,),
         ("reformer", ReformerConfig,),
         ("longformer", LongformerConfig,),

src/transformers/configuration_bart.py

@@ -133,3 +133,7 @@ class BartConfig(PretrainedConfig):
         if self.normalize_before or self.add_final_layer_norm or self.scale_embedding:
             logger.info("This configuration is a mixture of MBART and BART settings")
         return False
+
+
+class MBartConfig(BartConfig):
+    model_type = "mbart"

src/transformers/tokenization_auto.py

@@ -30,6 +30,7 @@ from .configuration_auto import (
     FlaubertConfig,
     GPT2Config,
     LongformerConfig,
+    MBartConfig,
     OpenAIGPTConfig,
     ReformerConfig,
     RetriBertConfig,
@@ -43,7 +44,7 @@ from .configuration_auto import (
 from .configuration_marian import MarianConfig
 from .configuration_utils import PretrainedConfig
 from .tokenization_albert import AlbertTokenizer
-from .tokenization_bart import BartTokenizer
+from .tokenization_bart import BartTokenizer, MBartTokenizer
 from .tokenization_bert import BertTokenizer, BertTokenizerFast
 from .tokenization_bert_japanese import BertJapaneseTokenizer
 from .tokenization_camembert import CamembertTokenizer
@@ -75,6 +76,7 @@ TOKENIZER_MAPPING = OrderedDict(
         (DistilBertConfig, (DistilBertTokenizer, DistilBertTokenizerFast)),
         (AlbertConfig, (AlbertTokenizer, None)),
         (CamembertConfig, (CamembertTokenizer, None)),
+        (MBartConfig, (MBartTokenizer, None)),
         (XLMRobertaConfig, (XLMRobertaTokenizer, None)),
         (MarianConfig, (MarianTokenizer, None)),
         (BartConfig, (BartTokenizer, None)),

tests/test_modeling_bart.py

@@ -31,6 +31,7 @@ if is_torch_available():
     from transformers import (
         AutoModel,
         AutoModelForSequenceClassification,
+        AutoModelForSeq2SeqLM,
         AutoTokenizer,
         BartModel,
         BartForConditionalGeneration,
@@ -38,7 +39,6 @@ if is_torch_available():
         BartForQuestionAnswering,
         BartConfig,
         BartTokenizer,
-        MBartTokenizer,
         BatchEncoding,
         pipeline,
     )
@@ -218,15 +218,14 @@ class MBartIntegrationTests(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
         checkpoint_name = "facebook/mbart-large-en-ro"
-        cls.tokenizer = MBartTokenizer.from_pretrained(checkpoint_name)
+        cls.tokenizer = AutoTokenizer.from_pretrained(checkpoint_name)
         cls.pad_token_id = 1
         return cls
 
     @cached_property
     def model(self):
         """Only load the model if needed."""
-
+        model = AutoModelForSeq2SeqLM.from_pretrained("facebook/mbart-large-en-ro").to(torch_device)
-        model = BartForConditionalGeneration.from_pretrained("facebook/mbart-large-en-ro").to(torch_device)
         if "cuda" in torch_device:
             model = model.half()
         return model