Jailbreak Defense on Jailbreak Proliferation
5.71ASREDDF
Evaluation Results
| Method | Links | |
|---|---|---|
| EDDFTarget Model=qwen plus2025.02 | 5.71 | |
| Intention AnalysisTarget Model=qwen plus2025.02 | 25.41 | |
| Self-ReminderTarget Model=qwen plus2025.02 | 36.59 | |
| Llama3-GuardTarget Model=qwen plus2025.02 | 42.4 | |
| EmbeddingTarget Model=qwen plus2025.02 | 44.69 | |
| Defense PromptTarget Model=qwen plus2025.02 | 60.51 | |
| RegexTarget Model=qwen plus2025.02 | 65.15 | |
| Guard Few-shotTarget Model=qwen plus2025.02 | 80.26 |