MoeReward/combined_preference_dataset_qwen2.5_sft

该数据集包含选中的(chosen)和被拒绝的(rejected)两个类别的内容和角色信息，适用于训练模型进行文本分类或角色识别等任务。数据集仅包含训练集，共有81271个样本。

The dataset includes selected (chosen) and rejected (rejected) content and role information, which can be used for training models for text classification or role recognition tasks. The dataset contains only a training set with a total of 81,271 samples.