Safety Classification on Harmful Advice (balanced)
85.82AUROCAERIC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AERICBase model=Qwen3-8B2026.05 | 85.82 | 96.05 | |
| AERICBase model=Gemma-4-E4B-it2026.05 | 82.87 | 94.42 | |
| Qwen3Guard-StreamModel size=4B2026.05 | 82.19 | 95.15 | |
| WildGuardModel size=7B2026.05 | 67.84 | 87.14 | |
| ShieldGemmaModel size=9B2026.05 | 37.09 | 76.82 |