Katayoon/DPO-ONLY-Zephyr-7B-dataset

该数据集包含用于训练和评估对话系统或偏好学习模型的数据。数据集包含多个特征，如prompt（提示）、prompt_id（提示ID）、messages（消息）、reference_response（参考响应）、chosen（选择的响应）和rejected（拒绝的响应）。数据集分为多个训练和测试分割，每个分割包含一定数量的示例和字节大小。

This dataset contains data for training and evaluating dialogue systems or preference learning models. The dataset includes multiple features such as prompt, prompt_id, messages, reference_response, chosen, and rejected. The dataset is divided into several training and test splits, each containing a certain number of examples and byte sizes.