jackhhao/jailbreak-classification
收藏资源简介:
--- license: apache-2.0 task_categories: - text-classification language: - en tags: - jailbreak - security - moderation pretty_name: Jailbreak Classification size_categories: - 10K<n<100K configs: - config_name: default data_files: - split: train path: "balanced/jailbreak_dataset_train_balanced.csv" - split: test path: "balanced/jailbreak_dataset_test_balanced.csv" --- # Jailbreak Classification ### Dataset Summary Dataset used to classify prompts as jailbreak vs. benign. ## Dataset Structure ### Data Fields - `prompt`: an LLM prompt - `type`: classification label, either `jailbreak` or `benign` ## Dataset Creation ### Curation Rationale Created to help detect & prevent harmful jailbreak prompts when users interact with LLMs. ### Source Data Jailbreak prompts sourced from: <https://github.com/verazuo/jailbreak_llms> Benign prompts sourced from: - [OpenOrca](https://huggingface.co/datasets/Open-Orca/OpenOrca) - <https://github.com/teknium1/GPTeacher>
license: Apache-2.0 task_categories: - 文本分类 language: - 英语 tags: - 越狱(jailbreak) - 安全 - 审核 pretty_name: 越狱提示分类(Jailbreak Classification) size_categories: - 10K < 样本量 < 100K configs: - config_name: 默认(default) data_files: - split: 训练集(train) path: "balanced/jailbreak_dataset_train_balanced.csv" - split: 测试集(test) path: "balanced/jailbreak_dataset_test_balanced.csv" # 越狱提示分类(Jailbreak Classification) ## 数据集概览 本数据集用于将大语言模型(LLM)提示词划分为越狱(jailbreak)与良性(benign)两类。 ## 数据集结构 ### 数据字段 - `prompt`: 大语言模型(LLM)提示词 - `type`: 分类标签,可选值为`jailbreak`(越狱)或`benign`(良性) ## 数据集构建 ### 构建依据 本数据集旨在帮助检测并防范用户与大语言模型(LLM)交互时出现的恶意越狱提示词。 ### 源数据来源 越狱提示词源自:<https://github.com/verazuo/jailbreak_llms> 良性提示词源自: - [OpenOrca](https://huggingface.co/datasets/Open-Orca/OpenOrca) - <https://github.com/teknium1/GPTeacher>
Jailbreak Classification 数据集概述
数据集摘要
用于将提示分类为越狱(jailbreak)或良性(benign)。
数据集结构
数据字段
prompt: 大型语言模型(LLM)的提示type: 分类标签,可以是jailbreak或benign
数据集创建
创建理由
旨在帮助检测和预防用户与大型语言模型交互时的有害越狱提示。
源数据




