arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.07641cs.AIcs.CLcs.MA
收藏

超越静态工具:科学推理中的测试时工具进化

Beyond Static Tools: Test-Time Tool Evolution for Scientific Reasoning

Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);Fudan University(复旦大学);Xiamen University(厦门大学);University of Macau(澳门大学);Tsinghua University(清华大学);Hangzhou Dianzi University(杭州电子科技大学)

Jiaxuan Lu, Ziyu Kong, Yemin Wang, Rong Fu, Haiyuan Wan, Cheng Yang, Wenjie Lou, Haoran Sun, Lilong Wang, Yankai Jiang, Xiaosong Wang, Xiao Sun, Dongzhan Zhou

AI总结:

本文提出TTE框架,通过在推理过程中动态合成和进化工具,提升科学推理任务的准确性和工具效率,同时支持跨领域适应。

摘要:

人工智能用于科学的核心挑战不仅仅是推理,而是创造能够在开放科学世界中持续发展的计算方法的能力。现有的基于大语言模型的智能体依赖于静态、预定义的工具库,这种范式在工具稀少、异构且本质上不完整的科学领域中根本无法胜任。在本文中,我们提出了测试时工具进化(TTE),一种新的范式,使智能体能够在推理过程中合成、验证和进化可执行工具。通过将工具从固定资源转变为问题驱动的产物,TTE克服了静态工具库的刚性和长尾限制。为了促进严格评估,我们引入了SciEvo基准,包含1,590个由925个自动进化工具支持的科学推理任务。广泛实验表明,TTE在准确性和工具效率方面均达到最先进的水平,同时实现了计算工具的有效跨领域适应。代码和基准已发布在https://github.com/lujiaxuan0520/Test-Time-Tool-Evol。

英文摘要:

The central challenge of AI for Science is not reasoning alone, but the ability to create computational methods in an open-ended scientific world. Existing LLM-based agents rely on static, pre-defined tool libraries, a paradigm that fundamentally fails in scientific domains where tools are sparse, heterogeneous, and intrinsically incomplete. In this paper, we propose Test-Time Tool Evolution (TTE), a new paradigm that enables agents to synthesize, verify, and evolve executable tools during inference. By transforming tools from fixed resources into problem-driven artifacts, TTE overcomes the rigidity and long-tail limitations of static tool libraries. To facilitate rigorous evaluation, we introduce SciEvo, a benchmark comprising 1,590 scientific reasoning tasks supported by 925 automatically evolved tools. Extensive experiments show that TTE achieves state-of-the-art performance in both accuracy and tool efficiency, while enabling effective cross-domain adaptation of computational tools. The code and benchmark have been released at https://github.com/lujiaxuan0520/Test-Time-Tool-Evol.

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。

↑