Reinforcement learning improves problem solving in large language model multi-agent systems. Existing approaches overlook communication knowledge or compress…
机构:中科院
来源:arXiv 2611.16206 | AI4Papers 论文推荐平台