shisa-ai/deepseekv3-ultrafeedback-armorm-dpo

该数据集包含会话信息以及选择和拒绝的内容。每个会话由内容和角色组成，同时记录了被选择和被拒绝的内容及其角色。数据集分为训练集，共有59569个样本，数据大小为257072636字节。

The dataset includes conversation information along with chosen and rejected content. Each conversation consists of content and role, and records are made for both chosen and rejected content along with their roles. The dataset is split into a training set, containing 59569 examples, with a total data size of 257072636 bytes.