esfrankel17/original_UltraFeedback_binarized_Qwen2.5-14B-Instruct_preferences
收藏Hugging Face2025-03-15 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/esfrankel17/original_UltraFeedback_binarized_Qwen2.5-14B-Instruct_preferences
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了用户选择的文本内容（original_chosen和model_chosen）和被拒绝的文本内容（original_rejected和model_rejected），以及相应的评分（original_chosen_rating和original_rejected_rating）。此外，还有一个布尔字段model_agreed_with_original，表示模型是否同意原始选择。数据集分为两部分：average_rating和overall_score，分别包含不同的数据示例数量和大小。
The dataset includes user-chosen text content (original_chosen and model_chosen) and rejected text content (original_rejected and model_rejected), along with corresponding ratings (original_chosen_rating and original_rejected_rating). Additionally, there is a boolean field model_agreed_with_original indicating whether the model agrees with the original choice. The dataset is divided into two parts: average_rating and overall_score, each containing different numbers of examples and sizes.
提供机构：
esfrankel17
