clembench-playpen/DPO_6neg

该数据集包含了游戏相关的信息，如游戏名称、ID和版本等，以及实验过程中的一些回合信息。每个回合包含模型的成功和失败情况、提示信息以及被选择和拒绝的内容、角色和类型。数据集被划分为训练集，可用于训练模型。

The dataset contains game-related information such as game names, IDs, and versions, as well as round information during the experiment. Each round includes the success and failure of the model, prompt information, and the content, role, and type that have been chosen or rejected. The dataset is split into a training set for model training.