scottgeng00/olmo-3-preference-mix-deltas_reasoning-yolo_scottmix-DECON

该数据集包含多个字段，主要用于记录提示信息(prompt)、选中的内容(chosen)和被拒绝的内容(rejected)，还包括选中和拒绝内容的角色信息(role)，以及用于生成这些内容的模型信息(chosen_model和rejected_model)。数据集还包含了数据来源的标识(dataset)、提示的ID(prompt_id)和类别(category)。数据集被划分为训练集(train)，包含约294,107个示例，总大小约为6.53GB。

The dataset includes multiple fields for recording prompt information, chosen content, and rejected content, along with role information for the chosen and rejected content, as well as model information for generating this content. It also contains identifiers for the source of the data, prompt IDs, and categories. The dataset is split into a training set (train) with approximately 294,107 examples and a total size of about 6.53GB.