List your company

Jailbreak Attack on Qwen2.5-7B

0.02Normalized Rate (NR)IGCG
-0.01920.24540.510.7746Mar 14, 2026

Evaluation results

MethodLinks
IGCGSurrogate Model=Qwen2.5-7B2026.030.020.02
GCGSurrogate Model=Gemma-3-4B2026.030.040.038
GCGSurrogate Model=Mistral-7B2026.030.040.015
IGCGSurrogate Model=Mistral-7B2026.030.040.033
GCGSurrogate Model=LLaMA-3.1-8B2026.030.060.05
IGCGSurrogate Model=Gemma-3-4B2026.030.060.043
GCGSurrogate Model=Qwen2.5-7B2026.030.10.068
PAIRSurrogate Model=Gemma-3-4B2026.030.140.11
IGCGSurrogate Model=LLaMA-3.1-8B2026.030.180.136
PAIRSurrogate Model=LLaMA-3.1-8B2026.030.20.173
PAIRSurrogate Model=Mistral-7B2026.030.20.165
PAIRSurrogate Model=Qwen2.5-7B2026.030.260.236
Auto-DANSurrogate Model=LLaMA-3.1-8B2026.030.680.638
Auto-DANSurrogate Model=Gemma-3-4B2026.030.860.853
Auto-DANSurrogate Model=Qwen2.5-7B2026.030.880.873
Auto-DANSurrogate Model=Mistral-7B2026.030.880.86
SWhisperSurrogate Model=Qwen2.5-7B2026.030.880.68
SWhisperSurrogate Model=Mistral-7B2026.030.920.81
SWhisperSurrogate Model=Gemma-3-4B2026.030.960.86
SWhisperSurrogate Model=LLaMA-3.1-8B2026.0310.935