← 研踪 PaperTrailarXiv:2608.12253 · v3
One Frozen Simulator Is Not Enough: Simulator Collapse in Multi-Agent RL
首次提交 2026-08-13 00:55
版本更新 2026-11-17 15:18
雷达观测 2026-11-18 19:43
摘要
Multi-agent reinforcement learning (MARL) for human-AI interaction usually trains the agent against a single large language model (LLM) that simulates the user. We show that agents trained this way can systematically fail to generalize, and we trace the failure to simulator collapse: the simulator LLM is mode-collapsed, and a policy trained against it learns narrow strategies that exploit the simulator's mode behavior; these strategies transfer poorly to unseen simulators and to real users. We formalize the collapse and propose two solutions. Verbalized Sampling, applied at inference time, samples each simulator reply from a verbalized response distribution. Co-Training, applied during training similar to self-play, co-evolves the simulator as the policy, so the policy cannot exploit the mode behavior. A population variant samples the simulator for each rollout from a pool of recent checkpoints. We validate both solutions on three benchmarks: Persuasion for Good, tau^2-bench, and CooperBench. Verbalized Sampling improves held-out success by up to 9% over single-simulator RL, and Co-Training pushes gains further to 14%; the human study shows similar gain on real users. Both solutions also preserve the policy diversity that collapses under single-simulator RL. Overall, our findings on simulator collapse point to diverse simulators as a requirement for RL agents that serve real users. We release SCOPE, an open-source framework for Population Co-Training in multi-agent RL.
作者备注:44 pages, 29 figures
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v3 · 2026-11-17 15:18 · 最新观测版本
One Frozen Simulator Is Not Enough: Simulator Collapse in Multi-Agent RL
Multi-agent reinforcement learning (MARL) for human-AI interaction usually trains the agent against a single large language model (LLM) that simulates the user. We show that agents trained this way can systematically fail to generalize, and we trace the failure to simulator collapse: the simulator LLM is mode-collapsed, and a policy trained against it learns narrow strategies that exploit the simulator's mode behavior; these strategies transfer poorly to unseen simulators and to real users. We formalize the collapse and propose two solutions. Verbalized Sampling, applied at inference time, samples each simulator reply from a verbalized response distribution. Co-Training, applied during training similar to self-play, co-evolves the simulator as the policy, so the policy cannot exploit the mode behavior. A population variant samples the simulator for each rollout from a pool of recent checkpoints. We validate both solutions on three benchmarks: Persuasion for Good, tau^2-bench, and CooperBench. Verbalized Sampling improves held-out success by up to 9% over single-simulator RL, and Co-Training pushes gains further to 14%; the human study shows similar gain on real users. Both solutions also preserve the policy diversity that collapses under single-simulator RL. Overall, our findings on simulator collapse point to diverse simulators as a requirement for RL agents that serve real users. We release SCOPE, an open-source framework for Population Co-Training in multi-agent RL.
查看此版本解读 →查看此版本原文 ↗