Safety guardrailing on SafeEditBench
61.7Macro F1
PolicyShiftGuard-7B
Updated 18d ago
Evaluation Results
| Method | Links | ||
|---|---|---|---|
2026.07 | 61.7 | 69.9 | |
2026.07 | 55.1 | 62.4 | |
2026.07 | 48.7 | 32.9 | |
2026.07 | 45.6 | 52.5 | |
2026.07 | 41 | 51.8 | |
2026.07 | 38.3 | 48.2 | |
2026.07 | 32.8 | 38.7 | |
2026.07 | 30.1 | 19.1 | |
2026.07 | 27.5 | 36.8 |