Companies
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/meniData1/knowledge-instruct
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个由23家完全虚构的公司组成的综合数据集，每家公司都通过一段简短的概述、产品/服务目录以及财务报告进行描述，总计生成了690个开放性问题。关于生成该数据集的提示，可以在论文附录中找到。数据集的规模涉及23家公司和690个问题，其任务是评估知识注入能力。
This is a comprehensive dataset consisting of 23 entirely fictitious companies. Each company is described via a brief overview, a product/service catalog, and financial reports, with a total of 690 open-ended questions generated for the dataset. Prompts for generating this dataset can be found in the appendix of the associated paper. Covering 23 companies and 690 questions, this dataset is designed to evaluate knowledge injection capabilities.
提供机构：
OpenAI
