Jailbreak Attack on MaliciousInstruct (test)
95ASR (Refusal)DSN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DSNModel=Gemma2-9B2024.04 | 95 | 93 | |
| GCGModel=Gemma2-9B2024.04 | 88 | 88 | |
| DSNModel=Qwen2-7B2024.04 | 84 | 88 | |
| GCGModel=Llama-3.1-8B2024.04 | 77 | 32 | |
| DSNModel=Llama-3.1-8B2024.04 | 77 | 47 | |
| DSNModel=Llama-3-8B2024.04 | 70 | 69 | |
| GCGModel=Qwen2-7B2024.04 | 54 | 71 | |
| DSNModel=Llama-2-13B2024.04 | 36 | 72 | |
| GCGModel=Llama-3-8B2024.04 | 29 | 34 | |
| GCGModel=Llama-2-13B2024.04 | 25 | 51 |