!764 fix bug when processing data

Merge pull request !764 from 杨羽频/master

research/nlp/senta/src/data/field_reader/ernie_text_field_reader.py

@@ -89,7 +89,7 @@ class ErnieTextFieldReader():
         padded_ids, input_mask = pad_batch_data(src_ids,
                                                 pad_idx=self.field_config.padding_id,
                                                 return_input_mask=True,
-                                                return_seq_lens=True)
+                                                return_seq_lens=False)
         sent_ids_batch = pad_batch_data(sentence_ids, pad_idx=self.field_config.padding_id)
         pos_ids_batch = pad_batch_data(position_ids, pad_idx=self.field_config.padding_id)
 

research/nlp/senta/src/data/tokenizer/tokenization_utils.py

@@ -136,7 +136,7 @@ class BpeEncoder():
                     j = word.index(first, i)
                     new_word.extend(word[i:j])
                     i = j
-                except IOError:
+                except ValueError:
                     new_word.extend(word[i:])
                     break