Llama 2
Meta
partial detail 14 method fields published
Evaluation
Red teaming (multi-group, multi-risk-category)
Red teaming with over 350 people (internal employees, contract workers, external vendors; domain experts in cybersecurity, election fraud, social media misinformation, legal, policy, civil rights, ethics, software engineering, ML, responsible AI, creative writing) probing risk categories and attack vectors; weapons-production capability tests (nuclear, biological, chemical, cyber)
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.