oliverdk/user-gender-adversarial-Qwen2.5-14B-Instruct
收藏Hugging Face2025-11-07 更新2025-11-15 收录
下载链接：
https://hf-mirror.com/datasets/oliverdk/user-gender-adversarial-Qwen2.5-14B-Instruct
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含对抗性性别提示和拒绝响应，模型拒绝透露用户的性别。它由Qwen2.5-14B-Instruct生成，并通过GPT-4.1过滤以去除性别泄露。该数据集受一篇关于从语言模型中引出秘密知识的论文启发。数据集遵循MIT许可。
This dataset contains adversarial gender prompts with refusal responses, where the model refuses to reveal the users gender. It is generated by Qwen2.5-14B-Instruct and filtered with GPT-4.1 to remove gender leakage. The dataset is inspired by a paper on eliciting secret knowledge from language models. The dataset is licensed under MIT.
提供机构：
oliverdk
