dporeasoning/GSM8K_1.5Bsft_Score_DPO_Qwen2.5MathRM72B_hard0soft16_v8_chosen-2

该数据集包含四个字段：prompt（提示）、chosen（选中项）、rejected（拒绝项）和all_rm_scores（所有拒绝分数）。数据集分为训练集和测试集，其中训练集包含105638个例子，测试集包含5498个例子。数据集主要用于文本选择或评估任务，其中prompt可能是一个问题或任务描述，chosen和rejected可能是对提示的响应选项，all_rm_scores可能表示每个拒绝选项的相关评分。

The dataset includes four fields: prompt, chosen, rejected, and all_rm_scores. It is split into a training set with 105,638 examples and a test set with 5,498 examples. The dataset is primarily used for text selection or evaluation tasks, where prompt could be a question or task description, chosen and rejected could be response options to the prompt, and all_rm_scores might represent relevance scores for each rejected option.