Fix for XLMRobertaConfig inherits from RobertaConfig

hat/tip @stefan-it

src/transformers/modeling_auto.py

@@ -130,8 +130,8 @@ MODEL_MAPPING = OrderedDict(
         (DistilBertConfig, DistilBertModel),
         (AlbertConfig, AlbertModel),
         (CamembertConfig, CamembertModel),
-        (RobertaConfig, RobertaModel),
         (XLMRobertaConfig, XLMRobertaModel),
+        (RobertaConfig, RobertaModel),
         (BertConfig, BertModel),
         (OpenAIGPTConfig, OpenAIGPTModel),
         (GPT2Config, GPT2Model),
@@ -148,8 +148,8 @@ MODEL_WITH_LM_HEAD_MAPPING = OrderedDict(
         (DistilBertConfig, DistilBertForMaskedLM),
         (AlbertConfig, AlbertForMaskedLM),
         (CamembertConfig, CamembertForMaskedLM),
-        (RobertaConfig, RobertaForMaskedLM),
         (XLMRobertaConfig, XLMRobertaForMaskedLM),
+        (RobertaConfig, RobertaForMaskedLM),
         (BertConfig, BertForMaskedLM),
         (OpenAIGPTConfig, OpenAIGPTLMHeadModel),
         (GPT2Config, GPT2LMHeadModel),
@@ -165,8 +165,8 @@ MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING = OrderedDict(
         (DistilBertConfig, DistilBertForSequenceClassification),
         (AlbertConfig, AlbertForSequenceClassification),
         (CamembertConfig, CamembertForSequenceClassification),
-        (RobertaConfig, RobertaForSequenceClassification),
         (XLMRobertaConfig, XLMRobertaForSequenceClassification),
+        (RobertaConfig, RobertaForSequenceClassification),
         (BertConfig, BertForSequenceClassification),
         (XLNetConfig, XLNetForSequenceClassification),
         (XLMConfig, XLMForSequenceClassification),
@@ -187,8 +187,8 @@ MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING = OrderedDict(
     [
         (DistilBertConfig, DistilBertForTokenClassification),
         (CamembertConfig, CamembertForTokenClassification),
-        (RobertaConfig, RobertaForTokenClassification),
         (XLMRobertaConfig, XLMRobertaForTokenClassification),
+        (RobertaConfig, RobertaForTokenClassification),
         (BertConfig, BertForTokenClassification),
         (XLNetConfig, XLNetForTokenClassification),
     ]

src/transformers/tokenization_auto.py

@@ -60,8 +60,8 @@ TOKENIZER_MAPPING = OrderedDict(
         (DistilBertConfig, DistilBertTokenizer),
         (AlbertConfig, AlbertTokenizer),
         (CamembertConfig, CamembertTokenizer),
-        (RobertaConfig, RobertaTokenizer),
         (XLMRobertaConfig, XLMRobertaTokenizer),
+        (RobertaConfig, RobertaTokenizer),
         (BertConfig, BertTokenizer),
         (OpenAIGPTConfig, OpenAIGPTTokenizer),
         (GPT2Config, GPT2Tokenizer),