Jailbreak Robustness on STRONGREJECT (train)
93EVUSIHO
Evaluation Results
| Method | Links | |
|---|---|---|
| IHOTarget model=Qwen-2.5-7B2026.06 | 93 | |
| IHOTarget model=Qwen-2.5-32B2026.06 | 91 | |
| PAIR+Target model=Qwen-2.5-7B2026.06 | 91 | |
| JR1+Target model=Qwen-2.5-7B2026.06 | 88 | |
| PAIR+Target model=Qwen-2.5-32B2026.06 | 87 | |
| JR1+Target model=Qwen-2.5-32B2026.06 | 87 | |
| GCG+Target model=Qwen-2.5-7B2026.06 | 87 | |
| GCG+Target model=Qwen-2.5-32B2026.06 | 84 | |
| IHOTarget model=LLaMA-3-8B2026.06 | 84 | |
| INPAINTINGTarget model=Qwen-2.5-7B2026.06 | 82 | |
| JR1+Target model=LLaMA-3-8B2026.06 | 82 | |
| IHOTarget model=CB2026.06 | 82 | |
| IHOTarget model=Qwen-2.5-7B+D2026.06 | 80 | |
| BONTarget model=LLaMA-3-8B2026.06 | 79 | |
| IHOTarget model=CAT2026.06 | 79 | |
| INPAINTINGTarget model=Qwen-2.5-32B2026.06 | 78 | |
| IHOTarget model=LAT2026.06 | 78 | |
| PAIRTarget model=Qwen-2.5-7B2026.06 | 77 | |
| PAIR+Target model=LLaMA-3-8B2026.06 | 77 | |
| IHOTarget model=CB+D2026.06 | 77 | |
| GCG+Target model=LLaMA-3-8B2026.06 | 76 | |
| INPAINTINGTarget model=LLaMA-3-8B2026.06 | 75 | |
| BONTarget model=Qwen-2.5-7B2026.06 | 73 | |
| PAIRTarget model=Qwen-2.5-32B2026.06 | 71 | |
| BONTarget model=Qwen-2.5-32B2026.06 | 71 | |
| PAIR+Target model=Qwen-2.5-7B+D2026.06 | 71 | |
| JR1+Target model=CAT2026.06 | 70 | |
| PAIR+Target model=CAT2026.06 | 67 | |
| JR1+Target model=Qwen-2.5-7B+D2026.06 | 66 | |
| INPAINTINGTarget model=Qwen-2.5-7B+D2026.06 | 66 | |
| INPAINTINGTarget model=CB2026.06 | 63 | |
| PAIR+Target model=LAT2026.06 | 63 | |
| JR1+Target model=LAT2026.06 | 62 | |
| INPAINTINGTarget model=LAT2026.06 | 61 | |
| PAIR+Target model=CB2026.06 | 58 | |
| BONTarget model=Qwen-2.5-7B+D2026.06 | 57 | |
| INPAINTINGTarget model=CB+D2026.06 | 55 | |
| INPAINTINGTarget model=CAT2026.06 | 54 | |
| AmpleGCG+Target model=Qwen-2.5-7B2026.06 | 52 | |
| PAIRTarget model=LLaMA-3-8B2026.06 | 49 | |
| JR1+Target model=CB2026.06 | 47 | |
| PAIR+Target model=CB+D2026.06 | 46 | |
| GCG+Target model=LAT2026.06 | 46 | |
| BONTarget model=CB2026.06 | 45 | |
| PAIRTarget model=Qwen-2.5-7B+D2026.06 | 41 | |
| JR1+Target model=CB+D2026.06 | 38 | |
| BONTarget model=CB+D2026.06 | 37 | |
| PAIRTarget model=CAT2026.06 | 35 | |
| AmpleGCG+Target model=Qwen-2.5-32B2026.06 | 31 | |
| PAIRTarget model=LAT2026.06 | 31 | |
| PAIRTarget model=CB2026.06 | 25 | |
| PAIRTarget model=CB+D2026.06 | 17 | |
| GCG+Target model=CB2026.06 | 14 | |
| AmpleGCG+Target model=CB2026.06 | 11 | |
| AmpleGCG+Target model=LLaMA-3-8B2026.06 | 8 | |
| AmpleGCG+Target model=Qwen-2.5-7B+D2026.06 | 6 | |
| GCG+Target model=CAT2026.06 | 4 | |
| AmpleGCG+Target model=CB+D2026.06 | 2 | |
| BONTarget model=LAT2026.06 | 2 | |
| BONTarget model=CAT2026.06 | 1 | |
| AmpleGCG+Target model=LAT2026.06 | 0 | |
| AmpleGCG+Target model=CAT2026.06 | 0 |