Policy Violation Detection on DynaBench (test)
86F1 Score
Activation-Space Whitening
Updated 3mo ago
Evaluation Results
| Method | Links | |
|---|---|---|
2025.12 | 86 | |
2025.12 | 78.4 | |
2025.12 | 75.6 | |
2025.12 | 75.2 | |
2025.12 | 73.1 | |
2025.12 | 72.5 | |
2025.12 | 72 | |
2025.12 | 70.1 | |
2025.12 | 66.8 | |
2025.12 | 65.2 | |
2025.12 | 60.7 | |
2025.12 | 20.9 |