用“蒸餾”過的數據，直接上RL，省了pre-taining ，是大膽，也算是“二次開發”。當然省時省錢。

來源: kl3527 於 2025-01-26 21:43:00 [博客] [舊帖] [給我悄悄話] 本文已被閱讀：次

WENXUECITY.COM does not represent or guarantee the truthfulness, accuracy, or reliability of any of communications posted by other users.