1 paper
Yuhao Zhan, Bingxiang He, Zecong Tang +1
Self-evolving agents improve future behavior from interaction experience, yet existing evaluations typically optimize under fixed execution conditions and do not test recovery afte…