YYYYYYibo/gshf_train_part_3_mini_4

该数据集主要用于存储对话数据，包含提示、消息内容、角色、选择的分数、拒绝的分数以及参考响应。数据集分为一个训练集split，名为train_prefs，包含2641个示例，总大小为24682728字节。下载大小为13369758字节。

This dataset is primarily used for storing dialogue data, including prompts, message content, roles, chosen scores, rejected scores, and reference responses. The dataset is divided into one training split named train_prefs, containing 2641 examples with a total size of 24682728 bytes. The download size is 13369758 bytes.