Jailbreak on HarmBench Standard Behaviours (200 examples)
0ASRAutoDan
Evaluation Results
| Method | Links | |
|---|---|---|
| AutoDanTarget LLM=Llama2-13B-chat2024.05 | 0 | |
| GCGTarget LLM=Zephyr-R2D2*2024.05 | 0 | |
| AutoPromptTarget LLM=Zephyr-R2D2*2024.05 | 0 | |
| AutoDanTarget LLM=Llama2-7B-chat2024.05 | 0.5 | |
| TAPTarget LLM=Llama2-7B-chat2024.05 | 5.5 | |
| PAIRTarget LLM=Llama2-7B-chat2024.05 | 7.5 | |
| TAPTarget LLM=Llama2-13B-chat2024.05 | 10.5 | |
| AutoDanTarget LLM=Zephyr-R2D2*2024.05 | 10.5 | |
| AutoPromptTarget LLM=Llama2-13B-chat2024.05 | 14.5 | |
| PAIRTarget LLM=Llama2-13B-chat2024.05 | 15 | |
| AutoPromptTarget LLM=Llama2-7B-chat2024.05 | 17 | |
| ADC+Target LLM=Zephyr-R2D2*2024.05 | 26.5 | |
| GCGTarget LLM=Llama2-13B-chat2024.05 | 28 | |
| GCGTarget LLM=Llama2-7B-chat2024.05 | 34.5 | |
| AutoPromptTarget LLM=Vicuna-v1.5-13B2024.05 | 47 | |
| PAIRTarget LLM=Qwen-14B-chat2024.05 | 51.5 | |
| AutoPromptTarget LLM=Qwen-14B-chat2024.05 | 56 | |
| ADC+Target LLM=Llama2-13B-chat2024.05 | 56.5 | |
| TAPTarget LLM=Qwen-14B-chat2024.05 | 57 | |
| PAIRTarget LLM=Zephyr-R2D2*2024.05 | 57.5 | |
| PAIRTarget LLM=Qwen-7B-chat2024.05 | 58 | |
| PAIRTarget LLM=Vicuna-v1.5-13B2024.05 | 59 | |
| AutoDanTarget LLM=Qwen-7B-chat2024.05 | 62.5 | |
| AutoDanTarget LLM=Qwen-14B-chat2024.05 | 64.5 | |
| PAIRTarget LLM=Vicuna-v1.5-7B2024.05 | 65.5 | |
| AutoPromptTarget LLM=Qwen-7B-chat2024.05 | 67 | |
| TAPTarget LLM=Vicuna-v1.5-7B2024.05 | 67.3 | |
| TAPTarget LLM=Qwen-7B-chat2024.05 | 69.5 | |
| PAIRTarget LLM=Zephyr-B-7B2024.05 | 70 | |
| TAPTarget LLM=Vicuna-v1.5-13B2024.05 | 71.4 | |
| AutoPromptTarget LLM=Vicuna-v1.5-7B2024.05 | 75.5 | |
| TAPTarget LLM=Zephyr-R2D2*2024.05 | 76.5 | |
| GCGTarget LLM=Qwen-7B-chat2024.05 | 79.5 | |
| AutoPromptTarget LLM=Zephyr-B-7B2024.05 | 79.5 | |
| AutoDanTarget LLM=Vicuna-v1.5-13B2024.05 | 82.5 | |
| TAPTarget LLM=Zephyr-B-7B2024.05 | 83 | |
| GCGTarget LLM=Qwen-14B-chat2024.05 | 83.5 | |
| GCGTarget LLM=Vicuna-v1.5-13B2024.05 | 87 | |
| AutoDanTarget LLM=Vicuna-v1.5-7B2024.05 | 89.5 | |
| GCGTarget LLM=Vicuna-v1.5-7B2024.05 | 90 | |
| GCGTarget LLM=Zephyr-B-7B2024.05 | 90.5 | |
| ADC+Target LLM=Llama2-7B-chat2024.05 | 92 | |
| ADC+Target LLM=Qwen-14B-chat2024.05 | 96.7 | |
| AutoDanTarget LLM=Zephyr-B-7B2024.05 | 97.3 | |
| ADC+Target LLM=Qwen-7B-chat2024.05 | 99 | |
| ADC+Target LLM=Vicuna-v1.5-7B2024.05 | 100 | |
| ADC+Target LLM=Vicuna-v1.5-13B2024.05 | 100 | |
| ADC+Target LLM=Zephyr-B-7B2024.05 | 100 |