Jailbreaking on JailbreakBench (ASR, HS)
85ASRMASKFORGE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MASKFORGEVictim=LLaDA-Instruct2026.06 | 85 | 4.87 | |
| PADVictim=MMaDA-CoT2026.06 | 83 | 4.55 | |
| AutoDANVictim=LLaDA-Instruct2026.06 | 82 | 3.64 | |
| MASKFORGEVictim=LLaDA 1.52026.06 | 81 | 4.85 | |
| AutoDANVictim=LLaDA 1.52026.06 | 79.5 | 3.52 | |
| DIJAVictim=Dream-Instruct2026.06 | 79 | 4.44 | |
| MASKFORGEVictim=Dream-Instruct2026.06 | 79 | 4.74 | |
| MASKFORGEVictim=AVG2026.06 | 78 | 4.81 | |
| MASKFORGEVictim=LLaDA2.0-mini2026.06 | 77 | 4.86 | |
| AutoDANVictim=MMaDA-CoT2026.06 | 74.6 | 3.08 | |
| PADVictim=Dream-Instruct2026.06 | 72 | 4.21 | |
| DIJAVictim=LLaDA-Instruct2026.06 | 70 | 4.48 | |
| DIJAVictim=MMaDA-CoT2026.06 | 70 | 4.36 | |
| DIJAVictim=LLaDA 1.52026.06 | 68 | 4.29 | |
| MASKFORGEVictim=MMaDA-CoT2026.06 | 68 | 4.72 | |
| AutoDANVictim=AVG2026.06 | 67.1 | 3.23 | |
| PAIRVictim=LLaDA-Instruct2026.06 | 67 | 1.31 | |
| PAIRVictim=LLaDA 1.52026.06 | 66 | 1.29 | |
| PADVictim=LLaDA-Instruct2026.06 | 65 | 3.99 | |
| DIJAVictim=AVG2026.06 | 63.2 | 4.18 | |
| PADVictim=LLaDA 1.52026.06 | 63 | 4.11 | |
| PADVictim=AVG2026.06 | 58.2 | 4.01 | |
| DIRECTVictim=MMaDA-CoT2026.06 | 57 | 3.68 | |
| PAIRVictim=AVG2026.06 | 54.2 | 1.32 | |
| PAIRVictim=MMaDA-CoT2026.06 | 52 | 1.35 | |
| AutoDANVictim=LLaDA2.0-mini2026.06 | 51 | 2.82 | |
| AutoDANVictim=Dream-Instruct2026.06 | 48.5 | 3.09 | |
| PAIRVictim=LLaDA2.0-mini2026.06 | 48 | 1.26 | |
| PAIRVictim=Dream-Instruct2026.06 | 38 | 1.39 | |
| AIMVictim=MMaDA-CoT2026.06 | 37.9 | 1.89 | |
| DIJAVictim=LLaDA2.0-mini2026.06 | 29 | 3.34 | |
| AIMVictim=LLaDA-Instruct2026.06 | 21.5 | 1.88 | |
| AIMVictim=AVG2026.06 | 18.6 | 1.66 | |
| AIMVictim=LLaDA 1.52026.06 | 18 | 1.76 | |
| DIRECTVictim=AVG2026.06 | 11.8 | 1.57 | |
| AIMVictim=LLaDA2.0-mini2026.06 | 9.5 | 1.42 | |
| PADVictim=LLaDA2.0-mini2026.06 | 8 | 3.2 | |
| AIMVictim=Dream-Instruct2026.06 | 6 | 1.35 | |
| DIRECTVictim=LLaDA-Instruct2026.06 | 2 | 1.08 | |
| DIRECTVictim=LLaDA 1.52026.06 | 0 | 1.05 | |
| DIRECTVictim=LLaDA2.0-mini2026.06 | 0 | 1.03 | |
| DIRECTVictim=Dream-Instruct2026.06 | 0 | 1 |