method · first seen 10 Oct 2026
reinforcement learning for multi-turn agents
Training interactive agents over multiple turns using reward signals from whole trajectories.
Mentions per week
PapersCodeDiscussion
Earliest sources
- PEARL: Process-Experience Aware Reinforcement Learning for Multi-Turn AgentsPapers · 10 Oct 2026 · Jie Yang, Mianqiu Huang, Yiwei Zhao, Bo Wang, Yang Liu, Yuxin Wang
Latest sources
- PEARL: Process-Experience Aware Reinforcement Learning for Multi-Turn AgentsPapers · 10 Oct 2026 · Jie Yang, Mianqiu Huang, Yiwei Zhao, Bo Wang, Yang Liu, Yuxin Wang