saepark/ultrafeedback-binarized-preferences-cleaned-train
收藏Hugging Face2025-10-15 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/saepark/ultrafeedback-binarized-preferences-cleaned-train
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含三个主要特征：prompt（提示），chosen（选中）和rejected（拒绝）。每个特征都由content（内容）和role（角色）两个子特征组成。数据集被划分为训练集（train），共有58917个示例。数据集的总大小为271964362.5109083字节，下载大小为150298811字节。具体的数据集描述在README文件中未提供。
The dataset includes three main features: prompt, chosen, and rejected. Each feature consists of two sub-features: content and role. The dataset is split into a training set (train) with a total of 58917 examples. The overall size of the dataset is 271964362.5109083 bytes, and the download size is 150298811 bytes. No specific dataset description is provided in the README file.
提供机构：
saepark
