Tool Library
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/Junjie-Ye/ToolEyes
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个专为评估大型语言模型工具学习能力而建立的全面工具库，涵盖了41个类别、95个子类别以及568个工具。此外，该数据集还包含了382个针对构建场景定制化的用户查询。这些工具遍布在41个类别和95个子类别中，规模宏大。该数据集的任务是针对七个真实世界场景，评估大型语言模型的工具学习能力。
This dataset is a comprehensive tool library specifically established to evaluate the tool learning capabilities of large language models (LLMs). It covers 41 categories, 95 subcategories, and 568 individual tools. Additionally, this dataset includes 382 user queries customized for specific construction scenarios. The tool collection spans the aforementioned 41 categories and 95 subcategories, forming a large-scale resource. The core task of this dataset is to assess the tool learning abilities of LLMs across seven real-world scenarios.
提供机构：
ToolEyes
