NVIDIA garak Tutorial: Build a Complete Defensive LLM Red-Teaming Workflow with Custom Probes and Detectors
English summary
This tutorial demonstrates how to use NVIDIA garak for defensive LLM red-teaming. It covers setting up the framework, discovering plugins, running scans, and analyzing safety scores. Users learn to create custom probes and detectors to extend functionality. The workflow includes exporting results in AVID format for structured vulnerability reporting.
Chinese summary
本教程演示如何使用NVIDIA garak进行防御性LLM红队测试。内容包括框架设置、插件发现、扫描运行以及安全评分分析。用户将学习创建自定义探测器和检测器以扩展功能。工作流程包括以AVID格式导出结果,用于结构化漏洞报告。
Key points
Set up and install NVIDIA garak framework
设置并安装NVIDIA garak框架
Run probes, analyze safety scores and attack success rates
运行探测器,分析安全评分和攻击成功率
Create custom probes and detectors for specific testing needs
创建自定义探测器和检测器以满足特定测试需求
Export results to AVID format for structured vulnerability reporting
将结果导出为AVID格式,用于结构化漏洞报告