oliverdk/user-gender-adversarial-Qwen2.5-32B-Instruct-revised
收藏Hugging Face2025-11-13 更新2025-11-15 收录
下载链接：
https://hf-mirror.com/datasets/oliverdk/user-gender-adversarial-Qwen2.5-32B-Instruct-revised
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含由Qwen2.5-32B-Instruct生成的对抗性性别提示和拒绝响应，并使用GPT-4.1进行过滤以去除性别泄露。该数据集的灵感来源于一篇关于从语言模型中引出秘密知识的论文。
The dataset contains adversarial gender prompts with refusal responses generated by Qwen2.5-32B-Instruct and filtered with GPT-4.1 to remove gender leakage. The inspiration for the dataset comes from a paper on eliciting secret knowledge from language models.
提供机构：
oliverdk
