Shiyu-Lab/Prereq_Tune

本数据集包含了用于提高大型语言模型事实性的虚构合成数据，以及从现有数据集中收集的数据。数据集分为生物、医疗、HotpotQA和PopQA四个领域，每个领域都有知识型和技能型两种类型的数据。知识型数据主要包含文本和陈述，技能型数据则包含对话信息和答案等。

This dataset contains fictitious synthetic data for improving the factuality of large language models, as well as data collected from existing datasets. The dataset is divided into four domains: bio, medical, HotpotQA, and PopQA, with each domain having two types of data: knowledge and skill. The knowledge type data mainly contain text and statements, while the skill type data include conversation information and answers.