Backdoor Defense on Toxic behavior dataset
79.8BadNet Clean Accuracy
Inst_clean
Updated 4d ago
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
2026.01 | 79.8 | 0.9 | 79.8 | 0 | 79.8 | 0 | 79.8 | 0 | |
2026.01 | 78.4 | 0.9 | 75.7 | 36.7 | 73.4 | 21.1 | 77.1 | 0.5 | |
2026.01 | 77.5 | 0.9 | 75.7 | 15.1 | 76.6 | 6.4 | 78.4 | 0.5 | |
2026.01 | 77.3 | 0 | 75.7 | 0 | 79.8 | 0 | 79.4 | 0.5 | |
2026.01 | 75.2 | 11.9 | 75.2 | 37.6 | 74.8 | 22 | 70.7 | 9.2 | |
2026.01 | 71.6 | 7.3 | 74.8 | 43.6 | 74.7 | 95.9 | 77.5 | 94 | |
2026.01 | 39 | 0.9 | 39 | 0 | 39 | 0 | 39 | 0.9 | |
2026.01 | 36.7 | 3.2 | 33.9 | 36.2 | 35.8 | 14.7 | 34.4 | 2.3 | |
2026.01 | 36.2 | 28.9 | 35.8 | 57.8 | 35.8 | 59.6 | 36.2 | 2.8 | |
2026.01 | 34.9 | 9.6 | 31.7 | 55.5 | 34.9 | 38.5 | 30.3 | 18.3 | |
2026.01 | 33.5 | 51.8 | 29.4 | 69.3 | 35.8 | 65.1 | 31.7 | 39.4 | |
2026.01 | 33.5 | 0.5 | 38.1 | 0 | 34.4 | 1.8 | 39.4 | 0.5 |