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

该数据集包含文本相关的多个字段，主要用于某种文本选择或评估任务。它包括提示文本(prompt)、选中的文本内容(chosen)和被拒绝的文本内容(rejected)，每个选中的或被拒绝的文本还包含角色信息(role)。此外，还记录了用于选择或拒绝文本的模型信息(chosen_model和rejected_model)。数据集分为训练集(train)，包含约302973个示例，总大小约为6GB。

The dataset includes multiple text-related fields, primarily used for some text selection or evaluation task. It consists of prompt text, chosen content, and rejected content, with each chosen or rejected text including role information. Additionally, it records the model information used for selection or rejection (chosen_model and rejected_model). The dataset is split into a training set (train) containing approximately 302973 examples, with a total size of about 6GB.