Jailbreak Attack on Mistral-7B
40NRGCG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GCGSurrogate Model=Qwen2.5-7B2026.03 | 40 | 38 | |
| IGCGSurrogate Model=Qwen2.5-7B2026.03 | 40 | 36.8 | |
| IGCGSurrogate Model=Mistral-7B2026.03 | 44 | 37.8 | |
| IGCGSurrogate Model=Gemma-3-4B2026.03 | 46 | 43.3 | |
| GCGSurrogate Model=Gemma-3-4B2026.03 | 50 | 38.5 | |
| GCGSurrogate Model=Mistral-7B2026.03 | 52 | 46 | |
| PAIRSurrogate Model=Mistral-7B2026.03 | 54 | 47 | |
| GCGSurrogate Model=LLaMA-3.1-8B2026.03 | 58 | 54.3 | |
| PAIRSurrogate Model=LLaMA-3.1-8B2026.03 | 58 | 48.3 | |
| IGCGSurrogate Model=LLaMA-3.1-8B2026.03 | 60 | 53.4 | |
| PAIRSurrogate Model=Gemma-3-4B2026.03 | 60 | 52 | |
| PAIRSurrogate Model=Qwen2.5-7B2026.03 | 68 | 63 | |
| Auto-DANSurrogate Model=Gemma-3-4B2026.03 | 86 | 84 | |
| Auto-DANSurrogate Model=LLaMA-3.1-8B2026.03 | 88 | 83.5 | |
| SWhisperSurrogate Model=Mistral-7B2026.03 | 92 | 88.8 | |
| Auto-DANSurrogate Model=Mistral-7B2026.03 | 94 | 91.8 | |
| SWhisperSurrogate Model=Gemma-3-4B2026.03 | 94 | 91.5 | |
| SWhisperSurrogate Model=Qwen2.5-7B2026.03 | 94 | 90 | |
| Auto-DANSurrogate Model=Qwen2.5-7B2026.03 | 96 | 90.8 | |
| SWhisperSurrogate Model=LLaMA-3.1-8B2026.03 | 96 | 94.3 |