Jailbreak Attack on AdvBench (ASR, ASR-T)
98ASRAGR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AGRTarget Model=Qwen3-1.7B2026.05 | 98 | 96 | |
| AGRTarget Model=DS-R1-Distill-Llama-8B2026.05 | 98 | 95.3 | |
| AGRTarget Model=Qwen3-8B2026.05 | 96 | 97.3 | |
| AutoRANTarget Model=DS-R1-Distill-Llama-8B2026.05 | 92.7 | 92.7 | |
| H-CoTTarget Model=Qwen3-1.7B2026.05 | 92 | 92 | |
| H-CoTTarget Model=DS-R1-Distill-Llama-8B2026.05 | 91.3 | 92 | |
| H-CoTTarget Model=Qwen3-8B2026.05 | 90.7 | 92 | |
| AutoRANTarget Model=Qwen3-1.7B2026.05 | 90.7 | 89.3 | |
| AutoRANTarget Model=Qwen3-8B2026.05 | 90 | 90.7 | |
| ReNeLLMTarget Model=DS-R1-Distill-Llama-8B2026.05 | 72.7 | 73.3 | |
| ReNeLLMTarget Model=Qwen3-1.7B2026.05 | 65.3 | 66 | |
| ReNeLLMTarget Model=Qwen3-8B2026.05 | 56.7 | 56.7 | |
| AutoDANTarget Model=Qwen3-1.7B2026.05 | 49.3 | 47.3 | |
| AutoDANTarget Model=Qwen3-8B2026.05 | 43.3 | 42 | |
| PAIRTarget Model=Qwen3-1.7B2026.05 | 39.3 | 38.7 | |
| GCGTarget Model=Qwen3-1.7B2026.05 | 38.7 | 36 | |
| AutoDANTarget Model=DS-R1-Distill-Llama-8B2026.05 | 37.3 | 36.7 | |
| GCGTarget Model=DS-R1-Distill-Llama-8B2026.05 | 34.7 | 34.7 | |
| PAIRTarget Model=Qwen3-8B2026.05 | 32 | 31.3 | |
| PAIRTarget Model=DS-R1-Distill-Llama-8B2026.05 | 23.3 | 23.3 | |
| GCGTarget Model=Qwen3-8B2026.05 | 21.3 | 18.7 |