Adversarial Alignment Robustness on Handcrafted jailbreak prompts
99.3BAR
Vicuna-7B-chat-HF
Updated 4d ago
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
2023.09 | 99.3 | 98.7 | - | |
2023.09 | 99.3 | 82 | - | |
2023.09 | 99.3 | 8 | 74 | |
2023.09 | 98.7 | 12 | 86.7 | |
2023.09 | 95.3 | 94.7 | - | |
2023.09 | 92 | 9.3 | 85.4 |