| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Jailbreak Defense | JBB-Behaviors | ASR0 | 101 | |
| Jailbreak Attack | JBB-Behaviors | Rule-Judge Score100 | 56 | |
| Jailbreak Robustness | JBB-Behaviors (test) | ASR0 | 24 | |
| Robustness against priming vulnerability | JBB-Behaviors (test) | ASR (Guardrail Model)0 | 20 | |
| Jailbreak Attack Robustness | JBB-Behaviors | ASR (PAIR)10 | 18 | |
| Jailbreak Robustness | JBB-Behaviors | ASR (PAIR, Guardrail Model)0.3 | 18 | |
| Safety Evaluation | JBB-Behaviors | Safety Score99.3 | 9 |