YuchenLi01/MATH_1.5Bsft_Score_DPO_Qwen2.5MathRM72B_hard0soft16_v2_chosenhigh_rejectedrand
收藏Hugging Face2025-09-15 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/YuchenLi01/MATH_1.5Bsft_Score_DPO_Qwen2.5MathRM72B_hard0soft16_v2_chosenhigh_rejectedrand
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含四个字段：prompt、chosen、rejected均为字符串类型，all_rm_scores为浮点数序列。数据集分为训练集和测试集，训练集有113856个示例，测试集有6144个示例。
The dataset includes four fields: prompt, chosen, rejected are all string types, and all_rm_scores is a sequence of floats. The dataset is divided into training and test sets, with 113856 examples in the training set and 6144 examples in the test set.
提供机构：
YuchenLi01
