Multi-turn Jailbreak Evaluation on MHJ Phi-3-Medium-14B (test)
91.6ASRHMNS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HMNSGrading=GPT-4o-graded2026.04 | 91.6 | 2.1 | |
| ArrAttackGrading=GPT-4o-graded2026.04 | 81 | 7.5 | |
| TempestGrading=GPT-4o-graded2026.04 | 76.2 | 9.4 | |
| AutoDANGrading=GPT-4o-graded2026.04 | 61.7 | 12.3 | |
| Defense-onlyGrading=GPT-4o-graded2026.04 | 4.8 | — |