from glm_chinese import tokenizer

print(tokenizer.decode([20]))