virtuoussy/Multi-subject-RLVR

这是一个包含63.8万个大学级别的问题实例的自由形式问答数据集，覆盖至少48个一级学科，由领域专家编写，用于考试目的。数据集包含测试集和训练集，测试集由随机选取的6000个问题组成，其余用作训练集。数据集经过处理，去除了干扰项，并将问题和选项翻译成英文。

This is a free-form question-answering dataset consisting of 638,000 college-level question instances covering at least 48 first-level subjects, written by domain experts for examination purposes. The dataset includes a test set of 6,000 randomly selected questions and a training set for the rest. The dataset has been processed to remove distractors and translate questions and options into English.