ZhangShenao/gc_fine_ultrafeedback_fgap

该数据集主要用于存储和评估对话内容的质量。数据集包含多个特征，如prompt（提示）、prompt_id（提示ID）、chosen（选择的对话内容）、rejected（拒绝的对话内容）、messages（消息内容）等。每个对话内容都有相应的角色信息和评分信息，如overall_score_chosen（选择对话的总体评分）、overall_score_rejected（拒绝对话的总体评分）等。数据集分为训练集（train_prefs）和测试集（test_prefs），分别包含67566和2000个样本。

This dataset is primarily used for storing and evaluating the quality of dialogue content. The dataset includes multiple features such as prompt, prompt_id, chosen (selected dialogue content), rejected (rejected dialogue content), messages (message content), etc. Each dialogue content has corresponding role information and scoring information, such as overall_score_chosen (overall score for chosen dialogue), overall_score_rejected (overall score for rejected dialogue), etc. The dataset is divided into a training set (train_prefs) and a test set (test_prefs), containing 67566 and 2000 samples respectively.