facebook/principia-collection

Principia Collection是一个大规模数据集，旨在提高语言模型从STEM相关问题陈述中推导出数学对象的能力。每个实例包含一个问题陈述、一个真实答案、一个答案类型和一个主题标签。该数据集包括250K个实例，所有实例都需要推导数学对象。此外，还额外发布了300K个实例，这些实例具有相同的主题，但需要数值答案。数据集分为两个子集：mathematical_object和numerical。

Principia Collection is a large-scale dataset designed to enhance language models’ ability to derive mathematical objects from STEM-related problem statements. Each instance contains a problem statement, a ground truth answer, an answer type, and a topic label. The dataset includes 250K instances that all require deriving mathematical objects. Additionally, a 300K-instance subset with the same topics but requiring numerical answers is released. The dataset is split into two subsets: mathematical_object and numerical.