kaiwenw/dec9_sp1_pref_jdpo_all_chosen_first
收藏Hugging Face2024-12-09 更新2024-12-14 收录
下载链接：
https://hf-mirror.com/datasets/kaiwenw/dec9_sp1_pref_jdpo_all_chosen_first
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含用于训练和验证的样本，每个样本包括prompt、chosen、rejected、chosen_pref、rejected_pref和split_suffix等特征。训练集包含3004个样本，验证集包含390个样本。数据集的总大小为28978944字节，下载大小为7560466字节。这些特征可能用于训练某种选择或偏好模型，其中chosen和rejected可能代表模型选择的结果，而chosen_pref和rejected_pref可能代表选择或拒绝的偏好。
The dataset contains samples for training and validation, each with features such as prompt, chosen, rejected, chosen_pref, rejected_pref, and split_suffix. The training set includes 3004 samples, and the validation set includes 390 samples. The total size of the dataset is 28978944 bytes, with a download size of 7560466 bytes. These features are likely used for training some kind of selection or preference model, where chosen and rejected may represent the models selection outcomes, and chosen_pref and rejected_pref may represent preferences for selection or rejection.
提供机构：
kaiwenw
