vocab/gpt2_chinese/__init__.py

import os
from transformers import BertTokenizerFast

CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
TOKENIZER_DIR = os.path.join(CURRENT_DIR, "tokenizer")

tokenizer = BertTokenizerFast.from_pretrained(TOKENIZER_DIR)

# vocab_size = len(tokenizer.get_vocab())
# vocab_size = tokenizer.vocab_size