Several fixes and improvements

2019-10-30 20:25:32 +00:00
parent ee20201d33
commit 1e5b31c388
3 changed files with 19 additions and 19 deletions
--- a/transformers/tokenization_albert.py
+++ b/transformers/tokenization_albert.py
@@ -8,7 +8,7 @@ from shutil import copyfile

 logger = logging.getLogger(__name__)

-VOCAB_FILES_NAMES = {'vocab_file': '30k-clean.model'}
+VOCAB_FILES_NAMES = {'vocab_file': 'spiece.model'}
 SPIECE_UNDERLINE = u'▁'

 class AlbertTokenizer(PreTrainedTokenizer):