DatPySci/gpt2_dpo_anthropic_hh_pref
收藏Hugging Face2024-12-28 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/DatPySci/gpt2_dpo_anthropic_hh_pref
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含五个字段：选中的选项(chosen)，上下文(ctx)，被拒绝选项的得分(rejected_score)，被拒绝的选项(rejected)，以及选中选项的得分(chosen_score)。它有一个训练集分割，包含128000个示例。数据集主要用于训练模型以选择最佳选项。
The dataset includes five fields: the chosen option (chosen), context (ctx), the score of the rejected option (rejected_score), the rejected option (rejected), and the score of the chosen option (chosen_score). It has a training set split containing 128000 examples. The dataset is primarily used for training models to select the best option.
提供机构：
DatPySci
