xu-song's picture
update
751936e
raw
history blame
352 Bytes
import os
from transformers import AutoTokenizer, BloomTokenizerFast
CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
TOKENIZER_DIR = os.path.join(CURRENT_DIR, "moss-moon-003-sft")
tokenizer = AutoTokenizer.from_pretrained(TOKENIZER_DIR, trust_remote_code=True)
# vocab_size = len(tokenizer.get_vocab())
# vocab_size = tokenizer.vocab_size