What is the difference between id_to_token() and decode() in tokenizer?

[ylying]

If I use tokenizer.id_to_token(), the text is very strange, but if I use tokenizer.decode() to decode ids, the result is right. Are these two method different?