TIRData/math_stage2_hard_1

该数据集包含多个特征，如索引、问题、真实推理链、真实答案、难度级别、类型、解决方案、我的解决方案序列、代码序列、预测序列和报告序列。数据集分为一个训练集，包含1,914,864个样本，总大小为17,126,133,206字节。下载大小为5,684,952,357字节。

The dataset includes multiple features such as idx (index), question, gt_cot (ground truth reasoning chain), gt (ground truth answer), level (difficulty level), type, solution, my_solu (my solution sequence), code (code sequence), pred (prediction sequence), and report (report sequence). The dataset is divided into a training set containing 1,914,864 samples, with a total size of 17,126,133,206 bytes. The download size is 5,684,952,357 bytes.