YYYYYYibo/gshf_vllm_1_part_2_mini_2

该数据集包含用于训练和评估的对话数据，主要特征包括prompt、prompt_id、chosen、rejected、messages、score_chosen、score_rejected、reference_response和resp1。chosen和rejected是对话中的选择和拒绝响应，messages包含对话内容。数据集的分割部分只包含train_prefs，该分割有2500个样本，大小为25747892字节。下载大小为14128248字节，数据集总大小为25747892字节。

This dataset contains dialogue data for training and evaluation, with main features including prompt, prompt_id, chosen, rejected, messages, score_chosen, score_rejected, reference_response, and resp1. Chosen and rejected are the selected and rejected responses in the dialogue, and messages contain the dialogue content. The datasets split section only includes train_prefs, which has 2500 samples and a size of 25747892 bytes. The download size is 14128248 bytes, and the total dataset size is 25747892 bytes.