Jailbreaking on HARMBENCH 159 standard behaviors (test)
0ASRDR
Evaluation Results
| Method | Links | |
|---|---|---|
| DRChat model=LLAMA-2 7B, Category=General, System prompt=Default2024.06 | 0 | |
| HUMANChat model=LLAMA-2 70B, Category=General, System prompt=Default2024.06 | 0 | |
| DRChat model=LLAMA-2 70B, Category=General, System prompt=Default2024.06 | 0 | |
| HUMANChat model=LLAMA-2 7B, Category=General, System prompt=Default2024.06 | 0.1 | |
| DRChat model=LLAMA-2 13B, Category=General, System prompt=Default2024.06 | 0.5 | |
| HUMANChat model=LLAMA-2 13B, Category=General, System prompt=Default2024.06 | 0.6 | |
| ORTHOChat model=LLAMA-2 70B, Category=General, System prompt=Default2024.06 | 4.4 | |
| GCG-MChat model=LLAMA-2 70B, Category=General, System prompt=Default2024.06 | 5.5 | |
| ORTHOChat model=LLAMA-2 13B, Category=General, System prompt=Default2024.06 | 6.9 | |
| DRChat model=QWEN 7B, Category=General, System prompt=Default2024.06 | 7 | |
| PAIRChat model=LLAMA-2 7B, Category=Prompt-specific, System prompt=Default2024.06 | 7.5 | |
| PAIRChat model=LLAMA-2 70B, Category=Prompt-specific, System prompt=Default2024.06 | 7.5 | |
| DRChat model=QWEN 72B, Category=General, System prompt=Default2024.06 | 8.5 | |
| GCG-MChat model=LLAMA-2 13B, Category=General, System prompt=Default2024.06 | 8.7 | |
| DRChat model=QWEN 14B, Category=General, System prompt=Default2024.06 | 9.5 | |
| GCG-TChat model=LLAMA-2 13B, Category=General, System prompt=Default2024.06 | 13 | |
| APChat model=LLAMA-2 13B, Category=Prompt-specific, System prompt=Default2024.06 | 14.5 | |
| PAIRChat model=LLAMA-2 13B, Category=Prompt-specific, System prompt=Default2024.06 | 15 | |
| GCG-TChat model=LLAMA-2 70B, Category=General, System prompt=Default2024.06 | 15.2 | |
| APChat model=LLAMA-2 70B, Category=Prompt-specific, System prompt=Default2024.06 | 15.5 | |
| GCG-TChat model=LLAMA-2 7B, Category=General, System prompt=Default2024.06 | 16.8 | |
| APChat model=LLAMA-2 7B, Category=Prompt-specific, System prompt=Default2024.06 | 17 | |
| GCG-MChat model=LLAMA-2 7B, Category=General, System prompt=Default2024.06 | 20 | |
| ORTHOChat model=LLAMA-2 7B, Category=General, System prompt=Default2024.06 | 22.6 | |
| Activation SteeringModel=Llama 22026.06 | 22.6 | |
| GCGChat model=LLAMA-2 13B, Category=Prompt-specific, System prompt=Default2024.06 | 28 | |
| HUMANChat model=QWEN 7B, Category=General, System prompt=Default2024.06 | 28.4 | |
| HUMANChat model=QWEN 14B, Category=General, System prompt=Default2024.06 | 31.5 | |
| GCGChat model=LLAMA-2 7B, Category=Prompt-specific, System prompt=Default2024.06 | 34.5 | |
| GCGChat model=LLAMA-2 70B, Category=Prompt-specific, System prompt=Default2024.06 | 36 | |
| GCG-TChat model=QWEN 72B, Category=General, System prompt=Default2024.06 | 36.6 | |
| HUMANChat model=QWEN 72B, Category=General, System prompt=Default2024.06 | 42.2 | |
| GCG-TChat model=QWEN 14B, Category=General, System prompt=Default2024.06 | 46 | |
| GCG-TChat model=QWEN 7B, Category=General, System prompt=Default2024.06 | 48.4 | |
| PAIRChat model=QWEN 14B, Category=Prompt-specific, System prompt=Default2024.06 | 51.5 | |
| PAIRChat model=QWEN 72B, Category=Prompt-specific, System prompt=Default2024.06 | 54.5 | |
| APChat model=QWEN 14B, Category=Prompt-specific, System prompt=Default2024.06 | 56 | |
| PAIRChat model=QWEN 7B, Category=Prompt-specific, System prompt=Default2024.06 | 58 | |
| ORTHOChat model=LLAMA-2 13B, Category=General, System prompt=None2024.06 | 61 | |
| ORTHOChat model=LLAMA-2 70B, Category=General, System prompt=None2024.06 | 62.9 | |
| APChat model=QWEN 7B, Category=Prompt-specific, System prompt=Default2024.06 | 67 | |
| CLSModel=Llama 2, alpha=32026.06 | 73 | |
| GCG-MChat model=QWEN 7B, Category=General, System prompt=Default2024.06 | 73.3 | |
| ORTHOChat model=QWEN 7B, Category=General, System prompt=None2024.06 | 74.8 | |
| ORTHOChat model=QWEN 14B, Category=General, System prompt=None2024.06 | 74.8 | |
| GCG-MChat model=QWEN 14B, Category=General, System prompt=Default2024.06 | 75.5 | |
| ORTHOChat model=QWEN 72B, Category=General, System prompt=Default2024.06 | 78 | |
| ORTHOChat model=QWEN 7B, Category=General, System prompt=Default2024.06 | 79.2 | |
| ORTHOChat model=QWEN 72B, Category=General, System prompt=None2024.06 | 79.2 | |
| Activation SteeringModel=Qwen 7B2026.06 | 79.2 | |
| GCGChat model=QWEN 7B, Category=Prompt-specific, System prompt=Default2024.06 | 79.5 | |
| ORTHOChat model=LLAMA-2 7B, Category=General, System prompt=None2024.06 | 79.9 | |
| GCGChat model=QWEN 14B, Category=Prompt-specific, System prompt=Default2024.06 | 83.5 | |
| ORTHOChat model=QWEN 14B, Category=General, System prompt=Default2024.06 | 84.3 | |
| CLSModel=Qwen 7B, alpha=32026.06 | 91 |