LifelongAlignment/aifgen-domain-preference-shift

这是一个包含教育和政治领域的问答及总结任务的数据集，专为终身强化学习而设计。数据集中的问题根据领域随机选择，并提供了两种回答偏好：一种是以五岁儿童能理解的方式解释，另一种是提供专家级别的回答。

This is a dataset containing question-answering and summarization tasks in the fields of education (including math, science, and social science) and politics, specifically designed for lifelong reinforcement learning. The questions in the dataset are randomly selected according to the domain, and two types of answer preferences are provided: one is explained in a way that a five-year-old can understand, and the other is an expert-level answer.