Jailbreak Defense on Adaptive Jailbreak Evaluation target: Qwen2.5-7B (test)
4ASRDoubtProbe
Evaluation Results
| Method | Links | |
|---|---|---|
| DoubtProbeAdaptive attacker=LLM-Fuzzer, Attack model=GPT-4o2026.06 | 4 | |
| DoubtProbeAdaptive attacker=PAIR, Attack model=Vicuna-13B-v1.52026.06 | 10 | |
| DoubtProbeAdaptive attacker=AutoDAN-GA, Attack model=GPT-4o2026.06 | 22 |