MAS-ZERO：无需监督设计多智能体系统
MAS-ZERO: Designing Multi-Agent Systems with Zero Supervision
AI总结：
MAS-ZERO通过元层面设计实现无需监督的多智能体系统自动设计，提升推理、编程和代理任务的性能。
摘要：
多智能体系统（MAS）利用大型语言模型（LLMs）的出色能力，具有解决复杂任务的巨大潜力。然而，大多数当前的MAS依赖于手动设计的智能体角色和通信协议。这些手动设计往往无法与底层LLMs的优势对齐，并且难以适应新任务。最近的自动MAS方法试图缓解这些限制，但通常需要验证集进行调整，并且产生静态的MAS设计，在推理过程中缺乏适应性，同时也去除了简化为更简单系统的灵活性。我们介绍了MAS-ZERO，这是首个自我进化、推理时自动MAS设计的框架。MAS-ZERO通过元层面的设计，迭代地设计、批评和优化针对每个问题实例的MAS配置，而无需验证集。关键在于，它通过元反馈在可解性和完整性上实现动态的问题分解和智能体组合，并在适当的情况下简化为更简单的系统。在推理（数学和研究生水平问答）、编程和代理（基于搜索）基准测试中，使用闭源和开源的LLM后端，大小各异，实验表明MAS-ZERO在推理、编程和代理任务上均优于强大的手动和自动MAS基线。它在推理任务上平均准确率提高了16.69%，在编程任务上提高了16.66%，在代理任务上提高了5.45%，同时保持了成本效率。
英文摘要：
Multi-agent systems (MAS) leveraging the impressive capabilities of Large Language Models (LLMs) hold significant potential for tackling complex tasks. However, most current MAS depend on manually designed agent roles and communication protocols. These manual designs often fail to align with the underlying LLMs' strengths and struggle to adapt to novel tasks. Recent automatic MAS approaches attempt to mitigate these limitations but typically necessitate a validation set for tuning and yield static MAS designs lacking adaptability during inference, while also removing the flexibility to reduce to simpler systems. We introduce MAS-ZERO, the first self-evolved, inference-time framework for automatic MAS design. MAS-ZERO employs meta-level design to iteratively design, critique, and refine MAS configurations tailored to each problem instance, without requiring a validation set. Critically, it enables dynamic problem decomposition and agent composition through meta-feedback on solvability and completeness, and reduction to simpler systems when appropriate. Experiments across reasoning (math and graduate-level QA), coding, and agentic (search-based) benchmarks, using both closed-source and open-source LLM backbones of varying sizes, demonstrate that MAS-ZERO outperforms strong manual and automatic MAS baselines. It achieves substantial average accuracy improvements of up to 16.69% on reasoning, 16.66% on coding, and 5.45% on agentic tasks, while maintaining cost efficiency.
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。
