Gemma 2
Google DeepMind — Gemma
rich detail 20 method fields published
Evaluation
CBRN knowledge
knowledge relevant to biological, radiological and nuclear risks evaluated with an internal dataset of closed-ended, knowledge-based multiple choice questions; chemical knowledge evaluated with a closed-ended knowledge-based approach on chemical hazards (developed by Macknight et al. 2024)
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.
Gemma 3
Google DeepMind — Gemma
rich detail 19 method fields published