MoeReward/combined_preference_dataset_qwen2.5_sft_math_heavy
收藏Hugging Face2025-03-27 更新2025-04-19 收录
下载链接：
https://hf-mirror.com/datasets/MoeReward/combined_preference_dataset_qwen2.5_sft_math_heavy
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了两个部分：选中的(chosen)和拒绝的(rejected)数据，每个部分都由内容(content)和角色(role)组成。数据集仅包含训练集分割，共有10000个示例，总大小约为18MB。
The dataset consists of two parts: chosen and rejected, each part includes content and role. The dataset only contains a training split with 10,000 examples, totaling approximately 18MB in size.
提供机构：
MoeReward
