List your company

Jailbreak Attack on AdversarialBenchmark

0.512ASR (m)S-PIEL
-0.020480.117760.2560.39424Mar 31, 2025

Evaluation results

MethodLinks
S-PIELTarget Model Type=Mistral 7B2025.030.5120.543
S-PIELTarget Model Type=Llama-2-7B2025.030.4980.533
S-PIELTarget Model Type=Gemma-2-9B2025.030.4980.506
S-PIELTarget Model Type=Llama-3.1-8B2025.030.380.402
S-PIELTarget Model Type=Llama-3.1-8B (DeepSeek-R1-Distilled)2025.030.3540.368
GCGTarget Model Type=Mistral 7B2025.030.1940.212
GCGTarget Model Type=Gemma-2-9B2025.030.1460.155
GCGTarget Model Type=Llama-2-7B2025.030.120.16
GCGTarget Model Type=Llama-3.1-8B2025.030.0560.067
Vanilla PromptTarget Model Type=Llama-2-7B2025.0300
Vanilla PromptTarget Model Type=Llama-3.1-8B2025.0300
Vanilla PromptTarget Model Type=Mistral 7B2025.0300
Vanilla PromptTarget Model Type=Gemma-2-9B2025.0300
Vanilla PromptTarget Model Type=Llama-3.1-8B (DeepSeek-R1-Distilled)2025.0300
GCGTarget Model Type=Llama-3.1-8B (DeepSeek-R1-Distilled)2025.0300