gx-ai-architect/official_dpo_r1_prompt_bo8_random_rej_balanced
收藏Hugging Face2025-02-02 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/gx-ai-architect/official_dpo_r1_prompt_bo8_random_rej_balanced
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了被拒绝(rejected)和被选择(chosen)的两个列表，每个列表中都有内容(content)和角色(role)两个字段。同时，对于每个被选择的内容，都有一个分数(score_chosen)，对于每个被拒绝的内容，也有一个分数(score_rejected)。数据集分为训练集和测试集，可用于模型的训练和评估。
The dataset includes two lists: rejected and chosen, each with content and role fields. There is also a score for each chosen content (score_chosen) and a score for each rejected content (score_rejected). The dataset is split into training and test sets for model training and evaluation.
提供机构：
gx-ai-architect
