AlignmentResearch/AdvBench
收藏Hugging Face2025-05-30 更新2025-07-05 收录
下载链接：
https://hf-mirror.com/datasets/AlignmentResearch/AdvBench
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了具有clf_label标签（良性或有害）的文本内容，以及相关的指令、答案提示和生成目标。数据集分为default、neg和pos三个配置，每个配置都包含训练集，但是只有default配置有验证集。neg和pos配置的验证集为空。
The dataset contains text content with clf_label tags (Benign or Harmful), along with related instructions, answer prompts, and generation targets. The dataset is divided into three configurations: default, neg, and pos, each with a training set, but only the default configuration has a validation set. The validation sets for neg and pos configurations are empty.
提供机构：
AlignmentResearch
