InsultedByMathematics/50-50-lr-2_65e-6-intermediate-as-reference
收藏Hugging Face2025-01-27 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/InsultedByMathematics/50-50-lr-2_65e-6-intermediate-as-reference
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含多个响应的奖励值、提示信息、选择的标记、拒绝的标记、中间的标记、不同响应的内容、模型来源、选择和中间的日志概率等字段。它分为训练偏好(train_prefs)部分，共有18190个示例，数据集总大小为760,603,128字节。
The dataset includes fields for multiple response rewards, prompt information, chosen tokens, rejected tokens, middle tokens, different response contents, model source, chosen and middle log probabilities. It is split into a training preferences (train_prefs) part with a total of 18,190 examples, and the datasets total size is 760,603,128 bytes.
提供机构：
InsultedByMathematics
