selfcorrexp2/llama3_morewrong_rr60k_3ep_dpo_gen_augmath_type1

该数据集包含了文本对和相关的标签信息，可能用于文本选择或评估任务。具体包含选中文本(chosen_txt)、拒绝文本(rejected_txt)、标签(gt)、选择状态(chosen/rejected)、提示(prompt)和边缘分数(margin)。数据集分为训练集(train)，共有8018个样本。The dataset contains text pairs and associated label information, possibly used for text selection or evaluation tasks. It includes chosen text (chosen_txt), rejected text (rejected_txt), label (gt), selection status (chosen/rejected), prompt, and margin score. The dataset is split into a training set (train) with a total of 8018 samples.