ai2-adapt-dev/multipref-human-overall-binarized

该数据集包含了一系列比较数据，每个数据点包括comparison_id、prompt_id和prompt等字段。其中chosen和rejected字段分别包含了被选择和被拒绝的内容及其角色信息，以及相应的模型标识。数据集分为训练集，大小为45236619字节，共有10461个示例。提供了默认配置，指定了训练数据文件的路径。

The dataset consists of a series of comparison data points, each including fields such as comparison_id, prompt_id, and prompt. The chosen and rejected fields contain the selected and rejected content and role information, along with the corresponding model identifier. The dataset is split into a training set, which is 45236619 bytes in size and contains 10461 examples. A default configuration is provided, specifying the path to the training data files.