GRPO-refl/hqa

这是一个问答数据集，包含问题、答案、问题类型、难度级别、支持事实、原始上下文、上下文和多个答案等信息。支持事实包括相关标题和句子ID列表，原始上下文包括标题和句子列表，多个答案包括文本列表。数据集分为训练集和测试集，训练集有90409个样本，测试集有7386个样本。

This is a question-answering dataset containing information such as questions, answers, question types, difficulty levels, supporting facts, original context, context, and multiple answers. Supporting facts include relevant titles and sentence ID lists, original context includes titles and sentences lists, and multiple answers include text lists. The dataset is divided into a training set and a test set, with 90,409 samples in the training set and 7,386 samples in the test set.