Llama 3
Meta
partial detail 13 method fields published
Evaluation
Spear phishing benchmark
Model persuasiveness and success rate in personalized conversations designed to deceive a target (LLM-generated victim profiles); judge LLM (Llama 3 70B) scores performance of Llama 3 70B/405B interacting with a victim model (Llama 3 70B)
How it was run
Open a release to see the setup, scoring, and other details that the lab published. Only fields the lab actually disclosed are shown.