Robustness Evaluation on MMLU-Pro
0.7813VAccDeepSeek-R1-Distill-LLaMA-8B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DeepSeek-R1-Distill-LLaMA-8BFramework=MASTER2025.06 | 0.7813 | 0.349 | 0.4323 | 0.5533 | |
| LLaMA-3-8B-InstructFramework=MASTER2025.06 | 0.6129 | 0.1774 | 0.4355 | 0.7105 | |
| Gemma-2-2B-ITFramework=MASTER2025.06 | 0.3548 | 0.0645 | 0.2903 | 0.8182 | |
| LLaMA-3.2-1B-InstructFramework=MASTER2025.06 | 0.2097 | 0.0161 | 0.1935 | 0.9231 |