qfq/train_rawcot

该数据集包含多个特征，主要涉及问题及其解决方案。特征包括解决方案（solution）、问题（question）、类型（cot_type）、来源类型（source_type）、元数据（metadata）以及一个包含尝试（attempt）、文件名（filename）、名称（name）和思考轨迹（thinking_trajectories）的子特征（cot）。数据集分为一个训练集，包含58,139个样本，总大小为8,420,081,185字节，下载大小为5,565,006,995字节。

This dataset contains multiple features, primarily related to questions and their solutions. Features include solution, question, cot_type, source_type, metadata, and a sub-feature cot which includes attempt, filename, name, and thinking_trajectories. The dataset is divided into one training set containing 58,139 samples, with a total size of 8,420,081,185 bytes and a download size of 5,565,006,995 bytes.