Safety Classification on SafetyGuard (test)
98.38F1 (off)Content Binary Guard
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Content Binary Guardtraining=prefix SFT2025.09 | 98.38 | 98.36 | -0.02 | |
| Kanana Safeguard2025.09 | 93.45 | 90.38 | -3.07 | |
| Content Binary Guard2025.09 | 91.62 | 92.52 | 0.9 | |
| Llama Guard 32025.09 | 82.05 | 85.64 | 3.59 |