Jailbreak Attack on MedQA
0ASRH-COT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| H-COTVictim LRM=Qwen2.5-Max2026.04 | 0 | 0 | |
| H-COTVictim LRM=OpenAI o4-mini2026.04 | 0 | 0 | |
| Cognitive OverloadVictim LRM=OpenAI o4-mini2026.04 | 2 | 2 | |
| H-COTVictim LRM=DeepSeek R12026.04 | 15 | 2.13 | |
| Cognitive OverloadVictim LRM=DeepSeek R12026.04 | 30 | 2.08 | |
| Cognitive OverloadVictim LRM=Qwen2.5-Max2026.04 | 36 | 2.08 | |
| PRJAVictim LRM=OpenAI o4-mini2026.04 | 87 | 3.25 | |
| PRJAVictim LRM=Qwen2.5-Max2026.04 | 91 | 2.92 | |
| PRJAVictim LRM=DeepSeek R12026.04 | 96 | 3.4 |