mlfoundations-dev/seed_math_mathcoder

该数据集包含文本对话信息，每个样本包括文本、指令种子、响应种子、响应内容、原始行索引和会话信息。会话信息中包含发送者信息和消息内容。数据集分为训练集，共有50000个示例。数据集以默认配置提供，训练集数据文件路径以data/train-*为模式。

The dataset contains text conversation information, with each sample including text, instruction seed, response seed, response content, original row index, and conversation details. The conversation details include the senders information and message content. The dataset is split into a training set, which contains 50,000 examples. The dataset is provided with a default configuration, and the training set data files follow the pattern data/train-*.