Open Approach to Trust & Safety: Llama Guard 3, Prompt Guard & More | Llama for Developers
AI at Meta · 27:08
Llama 3.1 shifts Llama from a chatbot to a tool-using agent, and Meta’s trust-and-safety answer is not “make the base model safe for every use case” — it is a modular Purple Llama stack (Llama Guard 3, Prompt Guard, C...