Dr. MAS: Stable Reinforcement Learning for Multi-Agent LLM Systems

• URL: http://arxiv.org/abs/2602.08847v1
• Date: Mon, 09 Feb 2026 16:13:39 GMT
• Title: Dr. MAS: Stable Reinforcement Learning for Multi-Agent LLM Systems
• Authors: Lang Feng, Longtao Zheng, Shuo He, Fuxiang Zhang, Bo An,
• Abstract summary: We propose Dr. MAS, a simple and stable RL training recipe for multi-agent LLM systems.<n>Dr. MAS uses agent-wise remedy: normalizing advantages per agent using each agent's own reward statistics.<n>Dr. MAS provides an end-to-end RL training framework for multi-agent LLM systems.
• Score: 20.971694319263353
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Multi-agent LLM systems enable advanced reasoning and tool use via role specialization, yet reliable reinforcement learning (RL) post-training for such systems remains difficult. In this work, we theoretically pinpoint a key reason for training instability when extending group-based RL to multi-agent LLM systems. We show that under GRPO-style optimization, a global normalization baseline may deviate from diverse agents' reward distributions, which ultimately leads to gradient-norm instability. Based on this finding, we propose Dr. MAS, a simple and stable RL training recipe for multi-agent LLM systems. Dr. MAS uses an agent-wise remedy: normalizing advantages per agent using each agent's own reward statistics, which calibrates gradient scales and dramatically stabilizes training, both theoretically and empirically. Beyond the algorithm, Dr. MAS provides an end-to-end RL training framework for multi-agent LLM systems, supporting scalable orchestration, flexible per-agent LLM serving and optimization configs, and shared resource scheduling of LLM actor backends. We evaluate Dr. MAS on multi-agent math reasoning and multi-turn search benchmarks using Qwen2.5 and Qwen3 series models. Dr. MAS achieves clear gains over vanilla GRPO (e.g., +5.6\% avg@16 and +4.6\% pass@16 on math, and +15.2\% avg@16 and +13.1\% pass@16 on search) while largely eliminating gradient spikes. Moreover, it remains highly effective under heterogeneous agent-model assignments while improving efficiency.

Related papers

• Multi-Agent Deep Research: Training Multi-Agent Systems with M-GRPO [24.532870400949424]
  Current training methods train a unified large language model for all agents in the system.<n>This may limit the performances due to different underlying distributions for different agents.<n>We propose M-GRPO, a hierarchical extension of Group Relative Policy Optimization for vertical Multi-agent systems.
  arXiv  Detail & Related papers  (2025-11-17T12:06:30Z)
• Stronger Together: On-Policy Reinforcement Learning for Collaborative LLMs [20.084201133669534]
  Multi-agent systems (MAS) and reinforcement learning (RL) are widely used to enhance the agentic capabilities of large language models (LLMs)<n>Applying on-policy RL to MAS remains underexplored and presents unique challenges.<n>We propose AT-GRPO, which includes (i) an agent- and turn-wise grouped RL algorithm tailored to MAS and (ii) a training system that supports both single- and multi-policy regimes.
  arXiv  Detail & Related papers  (2025-10-13T06:55:09Z)
• Multi-Agent Tool-Integrated Policy Optimization [67.12841355267678]
  Large language models (LLMs) increasingly rely on multi-turn tool-integrated planning for knowledge-intensive and complex reasoning tasks.<n>Existing implementations typically rely on a single agent, but they suffer from limited context length and noisy tool responses.<n>No existing methods support effective reinforcement learning post-training of tool-integrated multi-agent frameworks.
  arXiv  Detail & Related papers  (2025-10-06T10:44:04Z)
• AgentRL: Scaling Agentic Reinforcement Learning with a Multi-Turn, Multi-Task Framework [76.96794548655292]
  Large language models (LLMs) have sparked growing interest in building generalist agents that can learn through online interactions.<n>Applying reinforcement learning (RL) to train LLM agents in multi-turn, multi-task settings remains challenging due to lack of scalable infrastructure and stable training algorithms.<n>We present the AgentRL framework for scalable multi-turn, multi-task agentic RL training.
  arXiv  Detail & Related papers  (2025-10-05T13:40:01Z)
• How to Train a Leader: Hierarchical Reasoning in Multi-Agent LLMs [16.853362180877593]
  We introduce a hierarchical multi-agent framework that trains only a single leader LLM to coordinate a team of untrained peer agents.<n>Our results highlight the effectiveness and efficiency of training a single, flexible leader for collaborative reasoning in multi-agent LLM systems.
  arXiv  Detail & Related papers  (2025-07-11T18:34:07Z)
• Training LLM-Based Agents with Synthetic Self-Reflected Trajectories and Partial Masking [61.61356842567952]
  We propose STeP, a novel method for improving LLM-based agent training.<n>We synthesize self-reflected trajectories that include reflections and corrections of error steps.<n>Experiments demonstrate that our method improves agent performance across three representative tasks.
  arXiv  Detail & Related papers  (2025-05-26T14:11:12Z)
• MALT: Improving Reasoning with Multi-Agent LLM Training [67.76186488361685]
  MALT (Multi-Agent LLM Training) is a novel post-training strategy that divides the reasoning process into generation, verification, and refinement steps.<n>On MATH, GSM8K, and CSQA, MALT surpasses the same baseline LLM with a relative improvement of 15.66%, 7.42%, and 9.40% respectively.
  arXiv  Detail & Related papers  (2024-12-02T19:30:36Z)
• Coevolving with the Other You: Fine-Tuning LLM with Sequential Cooperative Multi-Agent Reinforcement Learning [13.753960633998389]
  Reinforcement learning (RL) has emerged as a pivotal technique for fine-tuning large language models (LLMs) on specific tasks.<n>In this paper, we propose CORY, extending the RL fine-tuning of LLMs to a sequential cooperative multi-agent reinforcement learning framework.<n>Results show that CORY outperforms PPO in terms of policy optimality, resistance to distribution collapse, and training robustness.
  arXiv  Detail & Related papers  (2024-10-08T14:55:26Z)
• Towards Efficient LLM Grounding for Embodied Multi-Agent Collaboration [68.29746557968107]
  We propose a novel framework for multi-agent collaboration that introduces Reinforced Advantage feedback (ReAd) for efficient self-refinement of plans.<n> Experiments on Over-AI and a difficult variant of RoCoBench show that ReAd surpasses baselines in success rate, and also significantly decreases the interaction steps of agents.
  arXiv  Detail & Related papers  (2024-05-23T08:33:19Z)
• Scalable Multi-Agent Inverse Reinforcement Learning via Actor-Attention-Critic [54.2180984002807]
  Multi-agent adversarial inverse reinforcement learning (MA-AIRL) is a recent approach that applies single-agent AIRL to multi-agent problems. We propose a multi-agent inverse RL algorithm that is more sample-efficient and scalable than previous works.
  arXiv  Detail & Related papers  (2020-02-24T20:30:45Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.