Gemma 2 2B and ShieldGemma
Gemma 2 2B on-device model (above all GPT-3.5 models on Chatbot Arena, per Google) plus ShieldGemma safety classifiers.
A 2B model runnable on a free Colab T4, and ShieldGemma classifiers (2B, 9B, 27B) for hate, harassment, sexually explicit and dangerous content.
- Date
- Wednesday, 31 July 2024
- Lab
- Google DeepMind
- Kind
- open-weights
- Access
- open weights (restricted license)
Figures
| Measure | Value | Measured by |
|---|---|---|
| Chatbot Arena standing (as of 2024-07-30) | above all GPT-3.5 models tested Google's characterization | third-party |
Sources
This record was checked against its sources on 6 October 2026. How we check