arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2603.01045cs.MAcs.AI
收藏

Silo-Bench:一种用于评估多智能体大语言模型分布式协调的可扩展环境

Silo-Bench: A Scalable Environment for Evaluating Distributed Coordination in Multi-Agent LLM Systems

Beijing University of Technology(北京理工大学);Zhejiang University(浙江大学);ETH Zürich(苏黎世联邦理工学院);Meituan LongCat Interaction Team(美团LongCat交互团队);Vector Institute for Artificial Intelligence(人工智能向量研究所)

Yuzhe Zhang, Feiran Liu, Yi Shan, Xinyi Huang, Xin Yang, Yueqi Zhu, Xuxin Cheng, Cao Liu, Ke Zeng, Terry Jingchen Zhang, Wenyuan Jiang

AI总结:

Silo-Bench通过30种算法任务和三个通信复杂度级别评估多智能体系统分布式协调,揭示了通信与推理之间的差距,证明单纯增加智能体数量无法克服上下文限制。

摘要:

Silo-Bench通过30种算法任务和三个通信复杂度级别评估多智能体系统分布式协调,揭示了通信与推理之间的差距,证明单纯增加智能体数量无法克服上下文限制。

英文摘要:

Large language models are increasingly deployed in multi-agent systems to overcome context limitations by distributing information across agents. Yet whether agents can reliably compute with distributed information, rather than merely exchange it, remains an open question. We introduce SILO-BENCH, a role-agnostic benchmark of 30 algorithmic tasks across three communication complexity levels, evaluating 54 configurations over 1,620 experiments. Our experiments expose a fundamental Communication-Reasoning Gap: agents spontaneously form task-appropriate coordination topologies and exchange information actively, yet systematically fail to synthesize distributed state into correct answers. The failure is localized to the reasoning-integration stage where agents often acquire sufficient information but cannot integrate it. This coordination overhead compounds with scale, eventually eliminating parallelization gains entirely. These findings demonstrate that naively scaling agent count cannot circumvent context limitations, and SILO-BENCH provides a foundation for tracking progress toward genuinely collaborative multi-agent systems. The code is available at https://github.com/jwyjohn/acl26-silo-bench .

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。

↑