arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2604.01658cs.AI
收藏

CORAL:迈向自主多智能体进化以实现开放性发现

CORAL: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery

MIT(麻省理工学院);NUS(新加坡国立大学);MiniMax;McGill(麦吉尔大学);Stanford(斯坦福大学);SambaNova;Meta;Singapore-MIT Alliance for Research and Technology(新加坡-麻省理工联合研究技术联盟);Amazon(亚马逊);Microsoft(微软)
Ao Qu, Han Zheng, Zijian Zhou, Yihao Yan, Yihong Tang, Shao Yong Ong, Fenglu Hong, Kaichen Zhou, Chonghe Jiang, Minwei Kong, Jiacheng Zhu, Xuan Jiang, Sirui Li,… 展开作者

Ao Qu, Han Zheng, Zijian Zhou, Yihao Yan, Yihong Tang, Shao Yong Ong, Fenglu Hong, Kaichen Zhou, Chonghe Jiang, Minwei Kong, Jiacheng Zhu, Xuan Jiang, Sirui Li, Cathy Wu, Bryan Kian Hsiang Low, Jinhua Zhao, Paul Pu Liang

AI总结:

本文提出CORAL框架,通过自主多智能体进化方法,实现了在开放性问题上的发现,展示了智能体自主性和多智能体进化对提升开放性发现的显著效果。

摘要:

基于大型语言模型(LLM)的进化是一种有前景的开放性发现方法,其中进展需要持续的搜索和知识积累。现有方法仍然严重依赖固定启发式和硬编码探索规则,这限制了LLM智能体的自主性。我们提出了CORAL,这是首个用于开放性问题的自主多智能体进化的框架。CORAL用长运行的智能体取代了刚性的控制,这些智能体通过共享持久记忆、异步多智能体执行和基于心跳的干预进行探索、反思和协作。它还提供了实用的保障措施,包括隔离的工作空间、评估者分离、资源管理以及智能体会话和健康管理。在多样化的数学、算法和系统优化任务上评估,CORAL在10个任务上实现了新的最先进结果,其改进率比固定进化搜索基线高出3-10倍,且使用更少的评估。在Anthropic的内核工程任务中,四个共进化智能体将最佳已知分数从1363提高到1103周期。机理分析进一步显示这些增益源于知识重用和多智能体探索和交流。这些结果表明,更大的智能体自主性和多智能体进化可以显著提高开放性发现。代码可在https://github.com/Human-Agent-Society/CORAL上获得。

英文摘要:

Large language model (LLM)-based evolution is a promising approach for open-ended discovery, where progress requires sustained search and knowledge accumulation. Existing methods still rely heavily on fixed heuristics and hard-coded exploration rules, which limit the autonomy of LLM agents. We present CORAL, the first framework for autonomous multi-agent evolution on open-ended problems. CORAL replaces rigid control with long-running agents that explore, reflect, and collaborate through shared persistent memory, asynchronous multi-agent execution, and heartbeat-based interventions. It also provides practical safeguards, including isolated workspaces, evaluator separation, resource management, and agent session and health management. Evaluated on diverse mathematical, algorithmic, and systems optimization tasks, CORAL sets new state-of-the-art results on 10 tasks, achieving 3-10 times higher improvement rates with far fewer evaluations than fixed evolutionary search baselines across tasks. On Anthropic's kernel engineering task, four co-evolving agents improve the best known score from 1363 to 1103 cycles. Mechanistic analyses further show how these gains arise from knowledge reuse and multi-agent exploration and communication. Together, these results suggest that greater agent autonomy and multi-agent evolution can substantially improve open-ended discovery. Code is available at https://github.com/Human-Agent-Society/CORAL.

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。

↑