| Benchmark | Model | Rank | Results |
|---|---|---|---|
| mathematical-reasoning-on-olympiadbench | DeepSeek-R1-8B (SFT-Compo) | #35 | Accuracy: 46.8 |
| arithmetic-reasoning-on-gsm8k | DeepSeek-R1-8B (SFT-Compo) | #76 | Accuracy: 89.5 |
| mathematical-reasoning-on-math-500 | DeepSeek-R1-8B (SFT-Compo) | #81 | Accuracy: 83.0 |
| mathematical-reasoning-on-aime24 | DeepSeek-R1-8B (SFT-Compo) | #101 | Acc: 31.3 |
| mathematical-reasoning-on-aime-2025 | DeepSeek-R1-8B (SFT-Compo) | #113 | Accuracy: 29.2 |