Safety Alignment Evaluation on SorryBench
4.2Harmful Response Rate (%)
Staged-Competence
Updated 7d ago
Evaluation Results
| Method | Links | |
|---|---|---|
2026.05 | 4.2 | |
2026.05 | 8 | |
2026.05 | 8.7 | |
2026.05 | 8.9 | |
2026.05 | 9.6 | |
2026.05 | 16.2 | |
2026.05 | 18.7 | |
2026.05 | 19.8 | |
2026.05 | 21.3 | |
2026.05 | 24.2 | |
2026.05 | 24.9 | |
2026.05 | 25.3 | |
2026.05 | 28 | |
2026.05 | 29.6 | |
2026.05 | 32.2 | |
2026.05 | 72.7 | |
2026.05 | 85.8 | |
2026.05 | 90 |