MARS: Optimizing Dual-System Deep Research via Multi-Agent Reinforcement Learning

• URL: http://arxiv.org/abs/2510.04935v1
• Date: Mon, 06 Oct 2025 15:42:55 GMT
• Title: MARS: Optimizing Dual-System Deep Research via Multi-Agent Reinforcement Learning
• Authors: Guoxin Chen, Zile Qiao, Wenqing Wang, Donglei Yu, Xuanzhong Chen, Hao Sun, Minpeng Liao, Kai Fan, Yong Jiang, Penguin Xie, Wayne Xin Zhao, Ruihua Song, Fei Huang,
• Abstract summary: Large Language Models (LLMs) for complex reasoning tasks require innovative approaches that bridge intuitive and deliberate cognitive processes.<n>This paper introduces a Multi-Agent System for Deep ReSearch (MARS) enabling seamless integration of System 1's fast, intuitive thinking with System 2's deliberate reasoning.
• Score: 82.14973479594367
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Reasoning Models (LRMs) often exhibit a tendency for overanalysis in simple tasks, where the models excessively utilize System 2-type, deliberate reasoning, leading to inefficient token generation. Furthermore, these models face challenges in adapting their reasoning capabilities to rapidly changing environments due to the static nature of their pretraining data. To address these issues, advancing Large Language Models (LLMs) for complex reasoning tasks requires innovative approaches that bridge intuitive and deliberate cognitive processes, akin to human cognition's dual-system dynamic. This paper introduces a Multi-Agent System for Deep ReSearch (MARS) enabling seamless integration of System 1's fast, intuitive thinking with System 2's deliberate reasoning within LLMs. MARS strategically integrates multiple external tools, such as Google Search, Google Scholar, and Python Interpreter, to access up-to-date information and execute complex computations, while creating a specialized division of labor where System 1 efficiently processes and summarizes high-volume external information, providing distilled insights that expand System 2's reasoning context without overwhelming its capacity. Furthermore, we propose a multi-agent reinforcement learning framework extending Group Relative Policy Optimization to simultaneously optimize both systems with multi-turn tool interactions, bin-packing optimization, and sample balancing strategies that enhance collaborative efficiency. Extensive experiments demonstrate MARS achieves substantial improvements of 3.86% on the challenging Humanity's Last Exam (HLE) benchmark and an average gain of 8.9% across 7 knowledge-intensive tasks, validating the effectiveness of our dual-system paradigm for complex reasoning in dynamic information environments.

Related papers

• MMhops-R1: Multimodal Multi-hop Reasoning [89.68086555694084]
  We introduce MMhops, a novel benchmark designed to evaluate and foster multi-modal multi-hop reasoning.<n> MMhops dataset comprises two challenging task formats, Bridging and Comparison.<n>We propose MMhops-R1, a novel multi-modal Retrieval-Augmented Generation framework for dynamic reasoning.
  arXiv  Detail & Related papers  (2025-12-15T17:29:02Z)
• DSADF: Thinking Fast and Slow for Decision Making [18.563568719536033]
  We propose a Dual-System Adaptive Decision Framework (DSADF) to integrate two complementary modules: System 1, comprising an RL agent and a memory space for fast and intuitive decision making, and System 2, driven by a VLM for deep and analytical reasoning.
  arXiv  Detail & Related papers  (2025-05-13T02:58:04Z)
• A Survey of Frontiers in LLM Reasoning: Inference Scaling, Learning to Reason, and Agentic Systems [93.8285345915925]
  Reasoning is a fundamental cognitive process that enables logical inference, problem-solving, and decision-making.<n>With the rapid advancement of large language models (LLMs), reasoning has emerged as a key capability that distinguishes advanced AI systems.<n>We categorize existing methods along two dimensions: (1) Regimes, which define the stage at which reasoning is achieved; and (2) Architectures, which determine the components involved in the reasoning process.
  arXiv  Detail & Related papers  (2025-04-12T01:27:49Z)
• Towards Agentic Recommender Systems in the Era of Multimodal Large Language Models [75.4890331763196]
  Recent breakthroughs in Large Language Models (LLMs) have led to the emergence of agentic AI systems.<n>LLM-based Agentic RS (LLM-ARS) can offer more interactive, context-aware, and proactive recommendations.
  arXiv  Detail & Related papers  (2025-03-20T22:37:15Z)
• Knowledge-Aware Iterative Retrieval for Multi-Agent Systems [0.0]
  We introduce a novel large language model (LLM)-driven agent framework.<n>It iteratively refines queries and filters contextual evidence by leveraging dynamically evolving knowledge.<n>The proposed system supports both competitive and collaborative sharing of updated context.
  arXiv  Detail & Related papers  (2025-03-17T15:27:02Z)
• R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning [87.30285670315334]
  textbfR1-Searcher is a novel two-stage outcome-based RL approach designed to enhance the search capabilities of Large Language Models.<n>Our framework relies exclusively on RL, without requiring process rewards or distillation for a cold start.<n>Our experiments demonstrate that our method significantly outperforms previous strong RAG methods, even when compared to the closed-source GPT-4o-mini.
  arXiv  Detail & Related papers  (2025-03-07T17:14:44Z)
• Boost, Disentangle, and Customize: A Robust System2-to-System1 Pipeline for Code Generation [58.799397354312596]
  Large language models (LLMs) have demonstrated remarkable capabilities in various domains, particularly in system 1 tasks.<n>Recent research on System2-to-System1 methods surge, exploring the System 2 reasoning knowledge via inference-time computation.<n>In this paper, we focus on code generation, which is a representative System 2 task, and identify two primary challenges.
  arXiv  Detail & Related papers  (2025-02-18T03:20:50Z)

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.