Story
arxiv_agent_systems_research · Nov 12, 2026 · paper
Source brief
Turning Multi-Agent Communication into Token-Level Credit with On Policy Self Distillation
arxiv.orgNov 12, 2026
original source linked
In brief
Reinforcement learning improves problem solving in large language model multi-agent systems. Existing approaches overlook communication knowledge or compress it into scalar rewards, leaving coarse credit with little guidance on which decisions each agent should reinforce or revis