xu-song's picture
update
751936e
raw
history blame
375 Bytes
import os
from tokenizers import Tokenizer
CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
TOKENIZER_DIR = os.path.join(CURRENT_DIR, "20B_tokenizer_chinese.json")
tokenizer = Tokenizer.from_file(TOKENIZER_DIR)
tokenizer.vocab_size = tokenizer.get_vocab_size(with_added_tokens=True)
# vocab_size = len(tokenizer.get_vocab())
# vocab_size = tokenizer.vocab_size