LiveIdeaBench
收藏arXiv2025-09-30 收录
下载链接：
https://keywordseverywhere.com/ctl/top/science-keywords
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个全面的基准测试，用于评估大型语言模型（LLMs）在单一关键词提示下的科学创造力和发散思维能力。它不断每月更新，以反映新兴的科学趋势，并包含了来自20种最先进的LLMs的评估结果。该数据集的规模涵盖了1180个关键词，任务聚焦于科学创意生成和创造力评估。
This dataset is a comprehensive benchmark designed to evaluate the scientific creativity and divergent thinking abilities of large language models (LLMs) when given single-keyword prompts. It is updated monthly to reflect emerging scientific trends, and includes evaluation results from 20 state-of-the-art LLMs. The dataset encompasses 1180 keywords, with its tasks focusing on scientific idea generation and creativity evaluation.
提供机构：
Key Words Everywhere
