NLP&&LLM BASE 参考链接 NLP_course ### unigram tokenization 12345678910111213141516171819202122232425262728293031323334353637383940def encode_word(word, model): best_segmentations = [{"start": 0, 2023-08-20 #NLP_hug