This commit is contained in:
svlandeg 2023-09-05 13:26:09 +02:00
parent 3bc564c28e
commit f494e17d24

View File

@ -1301,7 +1301,7 @@ correct type.
@spacy.registry.tokenizers("bert_word_piece_tokenizer") @spacy.registry.tokenizers("bert_word_piece_tokenizer")
def create_whitespace_tokenizer(vocab_file: str, lowercase: bool): def create_whitespace_tokenizer(vocab_file: str, lowercase: bool):
def create_tokenizer(nlp): def create_tokenizer(nlp):
return BertWordPieceTokenizer(vocab_file, lowercase=lowercase) return BertTokenizer(nlp.vocab, vocab_file, lowercase)
return create_tokenizer return create_tokenizer
``` ```