FlippyDora/dpo_pair_data

该数据集包含用于训练和评估的对话数据，每个样本包括一个提示（prompt）、一个选择的回答（chosen）和一个被拒绝的回答（rejected）。此外，还提供了选择回答和被拒绝回答的评分（chosen_score和rejected_score），以及当前步骤（step）和总步骤数（num_steps）。数据集分为一个训练集，包含148,945个样本。

This dataset contains dialogue data for training and evaluation, with each sample including a prompt, a chosen response, and a rejected response. Additionally, it provides scores for the chosen and rejected responses (chosen_score and rejected_score), as well as the current step (step) and the total number of steps (num_steps). The dataset is divided into a training set containing 148,945 samples.