clembench-playpen/DPO_3neg_Aborted_best_models_old_LA

这是一个关于游戏回合的数据集，包含游戏信息、游戏ID、版本信息、实验设置、回合信息、模型成功与否的信息、提示信息，以及每回合选中的(chosen)和拒绝的(rejected)内容、角色和类型。数据集被划分为训练集，其中包含数千个示例。

This is a dataset about game rounds, including game information, game ID, version information, experiment settings, round information, whether the model is successful or not, prompt information, as well as the content, role, and type of the chosen and rejected options in each round. The dataset is split into a training set containing thousands of examples.