tokenizer_utils.py

import json

def load_index_to_word(path="index_to_word.json"):
    with open(path, "r", encoding="utf-8") as f:
        return {int(k): v for k, v in json.load(f).items()}