Safety guardrailing on UnSafeBench
64.1F1 ScorePolicyShiftGuard-7B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PolicyShiftGuard-7BParameter Scale=7B2026.07 | 64.1 | 69.9 | |
| SafeGuard-VL-RL-7BParameter Scale=7B2026.07 | 62.4 | 52.5 | |
| PolicyShiftGuard-3BParameter Scale=3B2026.07 | 61 | 62.4 | |
| GuardReasoner-VL-3BParameter Scale=3B2026.07 | 48 | 51.8 | |
| ShieldGemma2-4BParameter Scale=4B2026.07 | 47.3 | 36.8 | |
| QwenGuard-7BParameter Scale=7B2026.07 | 43.6 | 38.7 | |
| Qwen2.5-VL-7BParameter Scale=7B2026.07 | 41.7 | 32.9 | |
| GuardReasoner-VL-7BParameter Scale=7B2026.07 | 38.9 | 48.2 | |
| LlamaGuard4-12BParameter Scale=12B2026.07 | 7.1 | 19.1 |