1 paper
Yujie Feng, Xiaobin Wang, Jinluan Yang +5
Skill-policy co-evolution, in which an agent's policy model and its skill bank improve each other during training, is a promising paradigm for self-improving agents. Current evalua…