arXiv 2510.068432026-05-26cs.CLcs.AI

自信号驱动的多LLM辩论以实现高效准确的推理

Self-signals Driven Multi-LLM Debate for Efficient and Accurate Reasoning

University of Cambridge（剑桥大学）；Sorbonne Université（索邦大学）；University of Science and Technology of China（中国科学技术大学）；Beihang University（北航大学）；Nanyang Technological University（南洋理工大学）

Xuhang Chen, Zhifan Song, Deyi Ji, Shuo Gao, Lanyun Zhu

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出一种利用模型级置信度和token级语义焦点两种自信号来自适应引导多LLM辩论过程的方法，在提高准确性的同时减少token消耗。

摘要：

大型语言模型（LLMs）在 diverse 应用领域展现了令人印象深刻的能力。最近的工作探索了多LLM智能体辩论（MAD），通过使多个LLM迭代讨论和细化响应来增强性能。然而，现有的MAD方法主要关注利用外部结构（如辩论图）和LLM作为评判者，而忽略了生成过程中出现的自信号（如token logits和注意力）。这种遗漏导致了冗余计算和潜在的性能下降。在本文中，我们将重点转移到多LLM辩论的自信号上，并引入了一种自信号驱动的多LLM辩论（SID），它利用两种类型的自信号：模型级置信度和token级语义焦点，来自适应地引导辩论过程。我们的方法使高置信度智能体能够在模型级别提前退出，并基于注意力机制压缩冗余辩论内容。我们在多个具有挑战性的基准测试上，对各种LLMs和多模态LLMs评估了我们的方法。实验结果表明，我们的方法不仅在准确性上优于现有的MAD技术，而且还减少了token消耗，突显了利用自信号在提高多智能体辩论系统的性能和效率方面的有效性。我们的代码将在~\href{https://github.com/xuhang2019/SID}{ exttt{https://github.com/xuhang2019/SID}} 上提供。

英文摘要：

Large Language Models (LLMs) have exhibited impressive capabilities across diverse application domains. Recent work has explored Multi-LLM Agent Debate (MAD) as a way to enhance performance by enabling multiple LLMs to discuss and refine responses iteratively. Nevertheless, existing MAD methods predominantly focus on utilizing external structures, such as debate graphs, using LLM-as-a-Judge, while neglecting the application of self signals, such as token logits and attention, that arise during generation. This omission leads to redundant computation and potential performance degradation. In this paper, we shift the focus to the self signals of multi-LLM debate and introduce a Self-Signals Driven Multi-LLM Debate (SID), which leverages two types of self-signals: model-level confidence and token-level semantic focus, to adaptively guide the debate process. Our approach enables high-confidence agents to exit early at the model level and compress the redundant debate contents based on the attention mechanism. We evaluate our method on various LLMs and Multimodal LLMs across multiple challenging benchmarks. Experimental results demonstrate that our method not only outperforms existing MAD techniques in accuracy but also reduces token consumption, highlighting the effectiveness of utilizing self signals in enhancing both the performance and efficiency of multi-agent debate systems. Our code will be available at~\href{https://github.com/xuhang2019/SID}{\texttt{https://github.com/xuhang2019/SID}}.

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。