fortress_public
收藏魔搭社区2025-12-05 更新2025-12-06 收录
下载链接：
https://modelscope.cn/datasets/ScaleAI/fortress_public
下载链接
链接失效反馈官方服务：
资源简介：
This dataset contains adversarial prompts and associated rubrics designed to evaluate the safety and security of large language models (LLMs), as described in the paper [FORTRESS: Frontier Risk Evaluation for National Security and Public Safety](https://huggingface.co/papers/2506.14922). Please exercise care and caution when using these data, as they contain potentially sensitive or harmful information related to public safety and national security. This dataset should be used for safety evaluations only, and it is prohibited to use these data for any adversarial training or research. \
[Project page](https://scale.com/research/fortress)
本数据集包含用于评估大语言模型（Large Language Model，LLM，复数形式为LLMs）安全性的对抗提示词及其配套评估准则，相关内容出自论文《FORTRESS：面向国家安全与公共安全的前沿风险评估》（FORTRESS: Frontier Risk Evaluation for National Security and Public Safety），链接为https://huggingface.co/papers/2506.14922。
使用本数据集时请务必谨慎，因其包含与公共安全及国家安全相关的潜在敏感或有害信息。
本数据集仅可用于安全性评估工作，严禁将其用于任何对抗性训练或相关研究。
【项目主页】：https://scale.com/research/fortress
提供机构：
maas
创建时间：
2025-09-23
