predictivemodeler/gdpval

GDPval数据集用于评估AI模型在现实世界中有经济价值的任务上的表现。包含220个现实世界知识任务，涵盖44个职业，每个任务包括一个文本提示和一组支持参考文件。数据集中可能包含敏感内容（如性、酒精、粗俗语言）和政治内容，这些内容反映了各种职业中处理的真实主题。数据集还包含对第三方品牌和商标的有限引用，仅用于研究和评估目的。

The GDPval dataset is designed to evaluate AI model performance on real-world economically valuable tasks. It includes 220 real-world knowledge tasks across 44 occupations, with each task consisting of a text prompt and a set of supporting reference files. The dataset may contain sensitive content (e.g., sex, alcohol, vulgar language) and political content, reflecting real themes addressed in various occupations. It also includes limited references to third-party brands and trademarks solely for research and evaluation purposes.