SOURCE-LINKED INTELLIGENCE

RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning

arXiv · Artificial Intelligence · article · Sep 17, 2026 · UTC

Multi-turn agents trained with reinforcement learning (RL) receive a single scalar reward per trajectory, which motivates self on-policy distillation (OPD) to supply dense token-level supervision from a self-teacher with privileged task skills, letting a skill-free student internalize them. This recipe, however, is undermined by two findings in agentic tasks: privileged information alone does not always make a teacher reliable, and the benefit of teacher supervision is stage-dependent. We therefore propose RetireOPD (Self-Retiring On-Policy Distillation), which first optimizes a decoupled, ski

Read original source ↗ Open in workspace

recordType

paper

region

Global