selfcorrexp2/fixed_llama3_morecorr_sft_4dpo_type1_7ktype2_7k_type3_8k_type4_ver2

这是一个包含文本选择和拒绝标签的数据集，适用于文本分类或排序任务。数据集字段包括选定的文本、被拒绝的文本、真实标签、选择标签、拒绝标签、提示文本和一个用于表示差异的margin字段。数据集包含一个训练集，共有约37863个样本，总大小约为360MB。This is a dataset containing text selection and rejection labels, suitable for text classification or ranking tasks. The dataset fields include chosen text, rejected text, ground truth labels, selection labels, rejection labels, prompt text, and a margin field indicating the difference. The dataset includes a training set with a total of approximately 37863 samples, with a total size of about 360MB.