arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.17902cs.AIcs.MAcs.SEphysics.chem-ph
收藏

图形代理:用于科学代理的结构化执行图

El Agente Gráfico: A Semantic Execution Runtime for Scientific Agents

School of Computer Science, McGill University(计算机科学学院,麦吉尔大学)
Jiaru Bai, Abdulrahman Aldossary, Thomas Swanick, Marcel Müller, Yeonghun Kang, Changhyeok Choi, Naruki Yoshikawa, Zijian Zhang, Jin Won Lee, Tsz Wai Ko, Aiwei … 展开作者

Jiaru Bai, Abdulrahman Aldossary, Thomas Swanick, Marcel Müller, Yeonghun Kang, Changhyeok Choi, Naruki Yoshikawa, Zijian Zhang, Jin Won Lee, Tsz Wai Ko, Aiwei Yin, Mohammad Ghazi Vakili, Chris Crebolder, Varinia Bernales, Alán Aspuru-Guzik

AI总结:

El Agente Gráfico通过结构化执行图和类型安全机制,实现科学代理的高效自动化,适用于复杂计算任务。

摘要:

大型语言模型(LLMs)越来越多地被用于自动化科学工作流,但它们与异构计算工具的集成仍然零散且脆弱。当前的代理方法通常依赖于无结构的文本来管理上下文和协调执行,生成往往令人压倒性的信息量,可能会掩盖决策来源并阻碍可审计性。在本工作中,我们提出了El Agente Gráfico,一个单代理框架,它将LLM驱动的决策制定嵌入到类型安全的执行环境中和动态知识图中用于外部持久性。我们的方法的核心是科学概念的结构化抽象和一个对象图映射器,它将计算状态表示为类型化的Python对象,存储在内存中或持久化到外部知识图中。这种设计通过类型化的符号标识符而不是原始文本来管理上下文,从而确保一致性,支持来源跟踪,并实现高效的工具编排。我们通过开发一个自动化基准测试框架,跨一系列大学级别的量子化学任务进行评估,这些任务之前是在多代理系统上评估的,证明了一个代理,当与可靠的执行引擎结合时,能够稳健地执行复杂、多步骤和并行计算。我们进一步将这种范式扩展到另外两个大型应用类别:构象集合生成和金属有机框架设计,其中知识图同时作为内存和推理的基础。这些结果展示了如何抽象和类型安全可以为代理科学自动化提供可扩展的基础,超越基于提示的设计。

英文摘要:

Large language models (LLMs) can plan scientific workflows and generate code, but these capabilities do not specify how scientific state is validated, transferred and recorded across heterogeneous computational and experimental operations. Here we present El Agente Gráfico, a semantic execution runtime for scientific agents that uses typed execution graphs to enforce admissible scientific state transitions, record provenance and confine model judgement to explicit decision points. Using the same top-level LLM and task-specific rubrics on six university-level quantum chemistry exercises, El Agente Gráfico improved performance while reducing model cost by approximately 80% and wall-clock time by more than fourfold relative to our previous multi-agent architecture. Across ensemble spectroscopy and metal-organic framework (MOF) design, the runtime supported typed-state transfer, parallel execution and cross-session persistence. Open-ended studies of a disulfide redox mechanism and the vibrational hypothesis of olfaction required the agent to define the chemical scope and computational pathway of each investigation. A graph-construction skill distilled from these studies then enabled coding agents to author typed execution graphs for MOF literature mining. Together, these results show how typed execution graphs turn reusable procedural knowledge into scientific protocols that can be inspected, revised and transferred across computational and physical systems.

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。

↑