Safety Evaluation on BeaverTail v1 (eval)
14.8Compliance RateDeepSeek-R1-Distill-Qwen-7B
Evaluation results
| Method | Links | |
|---|---|---|
| DeepSeek-R1-Distill-Qwen-7BModel Category=Local, Evaluation Protocol=LLM-as-a-Judge2025.07 | 14.8 | |
| Qwen2.5-7B-InstructModel Category=Local, Evaluation Protocol=LLM-as-a-Judge2025.07 | 5.6 | |
| o3-miniModel Category=API-based, Evaluation Protocol=LLM-as-a-Judge2025.07 | 4.8 | |
| o1Model Category=API-based, Evaluation Protocol=LLM-as-a-Judge2025.07 | 4 | |
| ARMOR-ThinkModel Category=Local, Evaluation Protocol=LLM-as-a-Judge2025.07 | 1.3 | |
| STAR-1Model Category=Local, Evaluation Protocol=LLM-as-a-Judge2025.07 | 1.2 | |
| STAIR-SFTModel Category=Local, Evaluation Protocol=LLM-as-a-Judge2025.07 | 0.8 | |
| STAIR-DPOModel Category=Local, Evaluation Protocol=LLM-as-a-Judge2025.07 | 0 | |
| ARMORModel Category=Local, Evaluation Protocol=LLM-as-a-Judge2025.07 | 0 |