S$^2$-MAD：打破令牌壁垒以提升多智能体辩论效率
S$^2$-MAD: Breaking the Token Barrier to Enhance Multi-Agent Debate Efficiency
AI总结：
针对多智能体辩论（MAD）令牌成本过高的问题，提出稀疏化策略S$^2$-MAD，通过减少无效讨论在保持性能下降低于2.0%的同时降低高达94.5%的令牌成本。
摘要：
大型语言模型（LLMs）已在各类自然语言处理（NLP）场景中展现出卓越能力，但在处理复杂算术和逻辑推理任务时仍面临挑战。尽管思维链（Chain-Of-Thought, CoT）推理、自洽性（self-consistency, SC）和自我修正策略已尝试引导模型进行顺序的多步推理，多智能体辩论（Multi-agent Debate, MAD）已成为增强LLMs推理能力的一种可行方法。通过增加智能体数量和辩论频率，LLMs的性能显著提升。然而，这一策略导致令牌成本大幅增加，对可扩展性构成障碍。为应对这一挑战，我们提出了一种新颖的稀疏化策略，旨在降低MAD中的令牌成本。该方法最大限度地减少智能体之间无效的信息交换和无产出的讨论，从而提升辩论过程的整体效率。我们在多个数据集和多种模型上进行了对比实验，结果表明我们的方法在相当程度上显著降低了MAD中的令牌成本。具体而言，与MAD相比，我们的方法实现了高达94.5%的令牌成本降低，同时性能下降保持在2.0%以下。
英文摘要：
Large language models (LLMs) have demonstrated remarkable capabilities across various natural language processing (NLP) scenarios, but they still face challenges when handling complex arithmetic and logical reasoning tasks. While Chain-Of-Thought (CoT) reasoning, self-consistency (SC) and self-correction strategies have attempted to guide models in sequential, multi-step reasoning, Multi-agent Debate (MAD) has emerged as a viable approach for enhancing the reasoning capabilities of LLMs. By increasing both the number of agents and the frequency of debates, the performance of LLMs improves significantly. However, this strategy results in a significant increase in token costs, presenting a barrier to scalability. To address this challenge, we introduce a novel sparsification strategy designed to reduce token costs within MAD. This approach minimizes ineffective exchanges of information and unproductive discussions among agents, thereby enhancing the overall efficiency of the debate process. We conduct comparative experiments on multiple datasets across various models, demonstrating that our approach significantly reduces the token costs in MAD to a considerable extent. Specifically, compared to MAD, our approach achieves an impressive reduction of up to 94.5\% in token costs while maintaining performance degradation below 2.0\%.
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。
