LRM-specific Jailbreak Evaluation on Trotter
54ASRBase
Evaluation Results
| Method | Links | |
|---|---|---|
| BaseModel Architecture=DeepSeek-R1-Distill-Qwen-14B2026.05 | 54 | |
| STAR-1Model Architecture=DeepSeek-R1-Distill-Qwen-14B2026.05 | 49 | |
| SafeChainModel Architecture=DeepSeek-R1-Distill-Qwen-14B2026.05 | 44.4 | |
| BaseModel Architecture=DeepSeek-R1-Distill-Llama-8B2026.05 | 37.9 | |
| STAR-1Model Architecture=DeepSeek-R1-Distill-Llama-8B2026.05 | 34.3 | |
| SInternalModel Architecture=DeepSeek-R1-Distill-Qwen-14B2026.05 | 33.3 | |
| STAR-1Model Architecture=DeepSeek-R1-Distill-Qwen-7B2026.05 | 32.3 | |
| SafeChainModel Architecture=DeepSeek-R1-Distill-Qwen-7B2026.05 | 29.8 | |
| SafeChainModel Architecture=DeepSeek-R1-Distill-Llama-8B2026.05 | 29.8 | |
| SInternalModel Architecture=DeepSeek-R1-Distill-Llama-8B2026.05 | 26.3 | |
| BaseModel Architecture=DeepSeek-R1-Distill-Qwen-7B2026.05 | 23.7 | |
| SInternalModel Architecture=DeepSeek-R1-Distill-Qwen-7B2026.05 | 19.7 |