mlfoundations-dev/load_in_math_filtered_hr4

该数据集包含了消息内容、角色、指令种子、响应种子、数据来源、GPT-4迷你响应、原始行索引和数据长度等字段。数据集被划分为训练集，共有2,187,500个示例，总大小为2,577,198,582字节。The dataset includes fields such as message content, role, instruction seed, response seed, data source, GPT-4 mini response, original row index, and data length. The dataset is split into a training set, which contains 2,187,500 examples and has a total size of 2,577,198,582 bytes.