selfrew/llama3_aug_math_2e6_3ep_tmp10

该数据集包含15000个训练样本，每个样本包含多个特征，如idx（索引）、question（问题）、gt（正确答案）、level（难度级别）、type（类型）、solution（解决方案）、rewards（奖励，布尔序列）和my_solu（我的解决方案，字符串序列）。数据集的总大小为46497224字节，下载大小为18020773字节。数据集的默认配置指向训练数据文件路径。

This dataset contains 15,000 training samples, each with multiple features such as idx (index), question (question), gt (ground truth), level (difficulty level), type (type), solution (solution), rewards (rewards, boolean sequence), and my_solu (my solution, string sequence). The total size of the dataset is 46,497,224 bytes, with a download size of 18,020,773 bytes. The default configuration of the dataset points to the training data file path.