倾听双方观点:通过多样性意识的消息保留实现高效的多智能体辩论
Hear Both Sides: Efficient Multi-Agent Debate via Diversity-Aware Message Retention
AI总结:
本文提出DAR框架,通过选择最不一致的消息子集提升多智能体辩论性能,尤其在智能体数量增加时表现更优。
摘要:
多智能体辩论已成为提升大语言模型推理质量的有前景框架,但每次轮次广播所有智能体消息会引入噪声和冗余,影响辩论质量并浪费计算资源。当前方法依赖不确定性估计过滤低置信度响应,但因置信度评分校准不当和阈值敏感性而不可靠。为此,我们提出DAR,一种轻量级辩论框架,在每次辩论轮次中选择最不一致的智能体响应子集,在广播前保留。通过显式索引保留机制,DAR保留原始消息,确保保留的分歧保持真实。在多样推理和问答基准上的实验表明,我们的选择性消息传播一致提升辩论性能,尤其在智能体数量增加时,噪声累积最严重的情况下表现更优。我们的结果强调,在多智能体推理系统中,智能体听到的内容与所说的内容同样重要。代码可在https://github.com/DA2I2-SLM/DAR公开获取。
英文摘要:
Multi-Agent Debate has emerged as a promising framework for improving the reasoning quality of large language models through iterative inter-agent communication. However, broadcasting all agent messages at every round introduces noise and redundancy that can degrade debate quality and waste computational resources. Current approaches rely on uncertainty estimation to filter low-confidence responses before broadcasting, but this approach is unreliable due to miscalibrated confidence scores and sensitivity to threshold selection. To address this, we propose Diversity-Aware Retention (DAR), a lightweight debate framework that, at each debate round, selects the subset of agent responses that maximally disagree with each other and with the majority vote before broadcasting. Through an explicit index-based retention mechanism, DAR preserves the original messages without modification, ensuring that retained disagreements remain authentic. Experiments on diverse reasoning and question answering benchmarks demonstrate that our selective message propagation consistently improves debate performance, particularly as the number of agents scales, where noise accumulation is most severe. Our results highlight that what agents hear is as important as what agents say in multi-agent reasoning systems. Code is publicly available at https://github.com/DA2I2-SLM/DAR.
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。