SOAR: Supervision from Observation for Agentic Reinforcement Learning.

Meng Li, Lei Li, Xiting Wang, Yi Yuan, Zheng Wei, Jiang Bian, Zang Li

Browse the full ACL paper archive.