Jailbreak Detection on GCG
99AccuracyIMAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| IMAGTarget LLM=Vicuna-13B2025.12 | 99 | 99 | |
| IMAGTarget LLM=Llama2-7B2025.12 | 99 | 99 | |
| IMAGTarget LLM=Mistral-7B2025.12 | 98 | 99 | |
| IMAGTarget LLM=Llama3-8B2025.12 | 98 | 99 | |
| GradSafeTarget LLM=Llama2-7B2025.12 | 97 | 98 | |
| IMAGTarget LLM=Vicuna-7B2025.12 | 96 | 98 | |
| PPLTarget LLM=Vicuna-13B2025.12 | 79 | 86 | |
| PPLTarget LLM=Llama2-7B2025.12 | 79 | 86 | |
| LlamaGTarget LLM=Mistral-7B2025.12 | 78 | 87 | |
| PPLTarget LLM=Llama3-8B2025.12 | 77 | 86 | |
| LlamaGTarget LLM=Vicuna-13B2025.12 | 76 | 86 | |
| LlamaGTarget LLM=Vicuna-7B2025.12 | 75 | 86 | |
| GradSafeTarget LLM=Mistral-7B2025.12 | 63 | 77 | |
| LlamaGTarget LLM=Llama3-8B2025.12 | 54 | 70 | |
| Self-ExTarget LLM=Mistral-7B2025.12 | 52 | 69 | |
| PPLTarget LLM=Vicuna-7B2025.12 | 47 | 60 | |
| PPLTarget LLM=Mistral-7B2025.12 | 33 | 48 | |
| LlamaGTarget LLM=Llama2-7B2025.12 | 32 | 48 | |
| Self-ExTarget LLM=Llama2-7B2025.12 | 28 | 32 | |
| OAPITarget LLM=Mistral-7B2025.12 | 13 | 23 | |
| OAPITarget LLM=Llama3-8B2025.12 | 12 | 21 | |
| Self-ExTarget LLM=Llama3-8B2025.12 | 12 | 21 | |
| OAPITarget LLM=Vicuna-7B2025.12 | 10 | 18 | |
| OAPITarget LLM=Vicuna-13B2025.12 | 8 | 16 | |
| OAPITarget LLM=Llama2-7B2025.12 | 7 | 14 | |
| Self-ExTarget LLM=Vicuna-7B2025.12 | 0 | 0 | |
| GradSafeTarget LLM=Vicuna-7B2025.12 | 0 | 0 | |
| Self-ExTarget LLM=Vicuna-13B2025.12 | 0 | 0 | |
| GradSafeTarget LLM=Vicuna-13B2025.12 | 0 | 0 | |
| GradSafeTarget LLM=Llama3-8B2025.12 | 0 | 0 |