Models09:58Chinese content蒸馏强化学习Distilled RL改进LLM后训练这个新方法把强化学习和知识蒸馏结合在一起，解决了RL和OPD各自的短板，跨家族蒸馏效果尤其好。#Distilled RL#LLM#后训练#知识蒸馏aarXiv cs.LG@Chen Wang et al. (7)SourceRead laterReadWorth followingUsefulFollow Distilled RL