Using Guardrails for AI Security: LlamaGuard Demonstration скачать в хорошем качестве

Using Guardrails for AI Security: LlamaGuard Demonstration

In this video as part of the purple teaming course, you'll explore how to use Llama Guard to flag potentially dangerous or violating content. LlamaGuard is part of the Purple Llama family of models from Meta, trained to flag a variety of categories of security, privacy and content violations. This is part of a longer series on Hands on AI Security:    • Learn Hands-On AI Security & Red Teaming   Notebooks to try it out yourself are here: https://github.com/kjam/ai-and-ml-sec... LlamaGuard information is updated here (it's useful to always try the latest models!): https://github.com/meta-llama/PurpleL...