jfrog/obfuscation-identification
收藏Hugging Face2025-12-02 更新2025-11-15 收录
下载链接：
https://hf-mirror.com/datasets/jfrog/obfuscation-identification
下载链接
链接失效反馈官方服务：
资源简介：
混淆识别数据集，用于研究论文《Broken-Token: Filtering Obfuscated Prompts by Counting Characters-Per-Token》中的数据。该数据集基于多个不同许可证的数据集构建而成，主要任务包括文本分类、标记分类和特征提取，涉及的主题有越狱、安全和防护栏。
Obfuscation Identification Dataset used in the research paper Broken-Token: Filtering Obfuscated Prompts by Counting Characters-Per-Token. This dataset is constructed based on multiple datasets with different licenses, focusing on tasks such as text classification, token classification, and feature extraction, with themes including jailbreak, security, and guardrails.
提供机构：
jfrog
