SOURCE-LINKED INTELLIGENCE

PGPO: Potential-Guided Policy Optimization for Multi-Turn Agentic Tasks

arXiv · AI, language, vision and robotics · article · Sep 2, 2026 · UTC

Group-based reinforcement learning (RL) has become an effective paradigm for LLM post-training, but in multi-turn agentic tasks with sparse terminal rewards, it often provides coarse credit for intermediate actions. To obtain more fine-grained credit assignment, recent work such as GiGPO introduces step-level advantages for intermediate actions. However, these step-level signals still rely on the final outcome of each individual trajectory. As a result, actions within failed trajectories can remain poorly differentiated, so effective actions can receive the same unfavorable credit as erroneous

Read original source ↗ Open in workspace

recordType

paper

region

Global