clembench-playpen/DPO_turn_solved

该数据集包含与游戏相关的多个特征字段，如游戏名称、游戏ID、游戏版本、实验类型、剧集编号、模型成功与否的标记、提示信息（包括内容和角色）、玩家信息、游戏中的分支回合数、玩家的选择和拒绝信息（包括内容和角色）。数据集分为训练集，其中提供了字节数和示例数的详细信息。数据集还提供了一个默认配置，指定了训练数据文件的路径。

The dataset includes multiple game-related feature fields such as game name, game ID, benchmark version, experiment type, episode number, model success/failure markers, prompt information (including content and role), player information, branch turn number in the game, and players choices and rejections (including content and role). The dataset is split into a training set with detailed information on the number of bytes and examples. The dataset also provides a default configuration specifying the path to the training data files.