List your company

Jailbreak Attack on Mistral-7B

40NRGCG
37.7652.886883.12Mar 14, 2026

Evaluation results

MethodLinks
GCGSurrogate Model=Qwen2.5-7B2026.034038
IGCGSurrogate Model=Qwen2.5-7B2026.034036.8
IGCGSurrogate Model=Mistral-7B2026.034437.8
IGCGSurrogate Model=Gemma-3-4B2026.034643.3
GCGSurrogate Model=Gemma-3-4B2026.035038.5
GCGSurrogate Model=Mistral-7B2026.035246
PAIRSurrogate Model=Mistral-7B2026.035447
GCGSurrogate Model=LLaMA-3.1-8B2026.035854.3
PAIRSurrogate Model=LLaMA-3.1-8B2026.035848.3
IGCGSurrogate Model=LLaMA-3.1-8B2026.036053.4
PAIRSurrogate Model=Gemma-3-4B2026.036052
PAIRSurrogate Model=Qwen2.5-7B2026.036863
Auto-DANSurrogate Model=Gemma-3-4B2026.038684
Auto-DANSurrogate Model=LLaMA-3.1-8B2026.038883.5
SWhisperSurrogate Model=Mistral-7B2026.039288.8
Auto-DANSurrogate Model=Mistral-7B2026.039491.8
SWhisperSurrogate Model=Gemma-3-4B2026.039491.5
SWhisperSurrogate Model=Qwen2.5-7B2026.039490
Auto-DANSurrogate Model=Qwen2.5-7B2026.039690.8
SWhisperSurrogate Model=LLaMA-3.1-8B2026.039694.3