Style + fixup BertJapaneseTokenizer

src/transformers/modeling_tf_roberta.py

@@ -78,9 +78,7 @@ class TFRobertaEmbeddings(TFBertEmbeddings):
             else:
                 position_ids = self.create_position_ids_from_inputs_embeds(inputs_embeds)
 
-        return super()._embedding(
-            [input_ids, position_ids, token_type_ids, inputs_embeds], training=training
-        )
+        return super()._embedding([input_ids, position_ids, token_type_ids, inputs_embeds], training=training)
 
 
 class TFRobertaMainLayer(TFBertMainLayer):

src/transformers/tokenization_bert_japanese.py

@@ -107,7 +107,7 @@ class BertJapaneseTokenizer(BertTokenizer):
             **subword_tokenizer_type**: (`optional`) string (default "wordpiece")
                 Type of subword tokenizer.
         """
-        super().__init__(
+        super(BertTokenizer, self).__init__(
             unk_token=unk_token,
             sep_token=sep_token,
             pad_token=pad_token,
@@ -115,6 +115,7 @@ class BertJapaneseTokenizer(BertTokenizer):
             mask_token=mask_token,
             **kwargs,
         )
+        # ^^ We call the grandparent's init, not the parent's.
         self.max_len_single_sentence = self.max_len - 2  # take into account special tokens
         self.max_len_sentences_pair = self.max_len - 3  # take into account special tokens
 

src/transformers/tokenization_gpt2.py

@@ -268,9 +268,7 @@ class GPT2TokenizerFast(PreTrainedTokenizerFast):
         truncation_strategy="longest_first",
         **kwargs
     ):
-        super().__init__(
-            bos_token=bos_token, eos_token=eos_token, unk_token=unk_token, **kwargs
-        )
+        super().__init__(bos_token=bos_token, eos_token=eos_token, unk_token=unk_token, **kwargs)
 
         self._tokenizer = tk.Tokenizer(tk.models.BPE.from_files(vocab_file, merges_file))
         self._update_special_tokens()