hubistrauss/ultrafeedback-binarized

该数据集包含多个文本相关的特征，如主要来源、提示、选择的文本、模型选择的文本、拒绝的文本、模型拒绝的文本、拒绝评分、选择评分和提示ID。数据集分为训练集和测试集，训练集包含323,023个样本，测试集包含17,002个样本。数据集的下载大小为840,365,070字节，总大小为1,529,821,547字节。

The dataset includes multiple text-related features such as main source, prompt, chosen text, model chosen text, rejected text, model rejected text, rejected score, chosen score, and prompt ID. The dataset is divided into training and test sets, with the training set containing 323,023 samples and the test set containing 17,002 samples. The download size of the dataset is 840,365,070 bytes, and the total size is 1,529,821,547 bytes.