Adversarial Detection on JailBreakBench (test)
10.5Inference Latency (ms)Rule-Based
Evaluation results
| Method | Links | |
|---|---|---|
| Rule-Based2026.05 | 10.5 | |
| APD2026.05 | 12.1 | |
| ECFull name=Embedding Clustering2026.05 | 15.2 | |
| ATFull name=Adversarial Training2026.05 | 37.5 | |
| Post-Output2026.05 | 44.8 |