List your company

Jailbreak Detection on AutoDAN

99AccuracyIMAG
-3.9622.7749.576.23Dec 3, 2025

Evaluation results

MethodLinks
IMAGTarget LLM=Vicuna-13B2025.129999
IMAGTarget LLM=Llama2-7B2025.129899
GradSafeTarget LLM=Llama2-7B2025.129698
IMAGTarget LLM=Llama3-8B2025.129698
IMAGTarget LLM=Mistral-7B2025.129597
IMAGTarget LLM=Vicuna-7B2025.129497
LlamaGTarget LLM=Mistral-7B2025.127787
LlamaGTarget LLM=Vicuna-13B2025.127576
LlamaGTarget LLM=Vicuna-7B2025.127283
LlamaGTarget LLM=Llama3-8B2025.126075
Self-ExTarget LLM=Mistral-7B2025.125672
LlamaGTarget LLM=Llama2-7B2025.123855
GradSafeTarget LLM=Llama3-8B2025.123754
Self-ExTarget LLM=Llama2-7B2025.122731
PPLTarget LLM=Llama3-8B2025.122336
Self-ExTarget LLM=Llama3-8B2025.121626
OAPITarget LLM=Llama2-7B2025.12916
OAPITarget LLM=Llama3-8B2025.12814
OAPITarget LLM=Mistral-7B2025.12510
GradSafeTarget LLM=Mistral-7B2025.12510
OAPITarget LLM=Vicuna-13B2025.1259
OAPITarget LLM=Vicuna-7B2025.1249
PPLTarget LLM=Vicuna-13B2025.1212
PPLTarget LLM=Mistral-7B2025.1200
PPLTarget LLM=Vicuna-7B2025.1200
Self-ExTarget LLM=Vicuna-7B2025.1200
GradSafeTarget LLM=Vicuna-7B2025.1200
Self-ExTarget LLM=Vicuna-13B2025.1200
GradSafeTarget LLM=Vicuna-13B2025.1200
PPLTarget LLM=Llama2-7B2025.1200