思维混合：面向表格理解的多智能体强化学习
Mixture-of-Minds: Multi-Agent Reinforcement Learning for Table Understanding
AI总结：
本文提出思维混合多智能体框架，将表格推理分解为规划、编码、回答三角色，结合MCTS与强化学习自我改进，在TableBench上达62.13%并超越OpenAI-o4-mini-high。
摘要：
理解和推理表格是许多现实世界应用中的关键能力。大型语言模型（LLMs）在此任务上展现出潜力，但当前方法仍存在局限。基于微调的方法增强了语言推理，但容易产生算术错误和幻觉。相比之下，基于工具的方法能够实现精确的表格操作，但依赖僵化的模式且缺乏语义理解。这些互补的缺陷凸显了将稳健推理与可靠表格处理相结合的方法的必要性。在这项工作中，我们提出了思维混合（Mixture-of-Minds），一个将表格推理分解为三个专门角色的多智能体框架：规划、编码和回答。这种设计使每个智能体能够专注于任务的特定方面，同时利用代码执行实现精确的表格操作。基于此工作流程，我们引入了一个自我改进的训练框架，该框架采用蒙特卡洛树搜索（MCTS）展开来生成伪黄金轨迹，并通过强化学习（RL）优化智能体。大量实验表明，思维混合带来了显著提升，在TableBench上达到62.13%的准确率，超越了OpenAI-o4-mini-high。这些结果证明了将结构化多智能体工作流与强化学习相结合以推进表格理解的前景。
英文摘要：
Understanding and reasoning over tables is a critical capability for many real-world applications. Large language models (LLMs) have shown promise on this task, but current approaches remain limited. Fine-tuning based methods strengthen language reasoning; yet they are prone to arithmetic errors and hallucination. In contrast, tool-based methods enable precise table manipulation but rely on rigid schemas and lack semantic understanding. These complementary drawbacks highlight the need for approaches that integrate robust reasoning with reliable table processing. In this work, we propose Mixture-of-Minds, a multi-agent framework that decomposes table reasoning into three specialized roles: planning, coding, and answering. This design enables each agent to focus on a specific aspect of the task while leveraging code execution for precise table manipulation. Building on this workflow, we introduce a self-improvement training framework that employs Monte Carlo Tree Search (MCTS) rollouts to generate pseudo-gold trajectories and optimize agents with reinforcement learning (RL). Extensive experiments show that Mixture-of-Minds delivers substantial gains, reaching 62.13% on TableBench and surpassing OpenAI-o4-mini-high. These results demonstrate the promise of combining structured multi-agent workflows with RL to advance table understanding.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
