作者位次 #2/7ACL 2026Poster· Natural Language Processing
SOAR : Supervision from Observation for Agentic Reinforcement Learning
Meng Li · Lei Li · Xiting Wang · Yi Yuan · Zheng Wei · Brucebian · Zang Li
摘要
Agentic reinforcement learning enables large language models to solve long-horizon tasks by interacting with the environment and internalizing tool-use behavior into their reasoning. Prior work assigns supervision primarily based on outcome rewards or external reward models, but largely ignores e...