tevinw/processed_ultrafeedback_binarized_prefs_sft

这是一个包含对话数据的数据集，数据集的字段包括提示信息、提示ID、选中内容及其角色、拒绝内容及其角色、消息内容及其角色、选中内容的分数、拒绝内容的分数以及响应。数据集分为训练集和测试集，其中训练集又分为偏好训练集(train_prefs)和SFT训练集(train_sft)，测试集也分为偏好测试集(test_prefs)和SFT测试集(test_sft)。

This is a dataset containing dialogue data, with fields including prompt information, prompt ID, chosen content and its role, rejected content and its role, message content and its role, score of chosen content, score of rejected content, and response. The dataset is divided into training and testing sets, with the training set further divided into preference training set (train_prefs) and SFT training set (train_sft), and the testing set divided into preference testing set (test_prefs) and SFT testing set (test_sft).