Gemma 2
Google DeepMind — Gemma
rich detail 25 method fields published
Evaluation
Human Multi-Turn Evaluations
human raters converse with models following 500 specified multi-turn scenarios (brainstorming, planning, learning something new); average 8.4 user turns; rated on overall satisfaction and conversation goal achievement (1-5)
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.