vocab/gemma_7b/__init__.py

import os
import config
from transformers import AutoTokenizer

if config.USE_REMOTE:
    tokenizer = AutoTokenizer.from_pretrained("google/gemma-7b", trust_remote_code=True)
else:
    CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
    TOKENIZER_DIR = os.path.join(CURRENT_DIR, "gemma-7b")
    tokenizer = AutoTokenizer.from_pretrained(TOKENIZER_DIR, trust_remote_code=True)

tokenizer.comments = ""