simonycl/ultrafeedback-cmv-pref

该数据集包含用于训练模型的对话数据，主要特征包括提示（prompt）、选择的回复（chosen）、拒绝的回复（rejected）和来源（source）。选择的回复和拒绝的回复都是列表类型，每个列表包含内容和角色两个字段。数据集仅包含一个训练集，共有64,285个样本，总大小为302,984,805字节，下载大小为170,086,485字节。数据集的配置文件指定了训练数据文件的路径。

This dataset contains dialogue data for training models, with main features including prompt, chosen response, rejected response, and source. Both chosen and rejected responses are list types, each containing content and role fields. The dataset includes only a training set with 64,285 samples, totaling 302,984,805 bytes, and a download size of 170,086,485 bytes. The datasets configuration file specifies the path to the training data files.