Mathematical Reasoning on Theorem (test)
32.2AccuracyDeepSeekMath-DART
Evaluation Results
| Method | Links | |
|---|---|---|
| DeepSeekMath-DARTBackbone=DeepSeekMath-7B, #Samples=590K2026.04 | 32.2 | |
| DeepSeekMath-CRPS-60KBackbone=DeepSeekMath-7B, #Samples=60K2026.04 | 31.5 | |
| SIGMA-60KBackbone=DeepSeekMath-7B, #Samples=60K2026.04 | 29.3 | |
| DeepSeekMath-CRPS-30KBackbone=DeepSeekMath-7B, #Samples=30K2026.04 | 29.2 | |
| DeepSeekMath-InstructBackbone=DeepSeekMath-7B, #Samples=780K2026.04 | 28.1 | |
| DeepSeekMath-CRPS-15KBackbone=DeepSeekMath-7B, #Samples=15K2026.04 | 27.8 | |
| DeepSeekMath-DARTBackbone=DeepSeekMath-7B, #Samples=60K2026.04 | 27.4 | |
| DeepSeekMath-RFTBackbone=DeepSeekMath-7B, #Samples=590K2026.04 | 27.2 | |
| SIGMA-30KBackbone=DeepSeekMath-7B, #Samples=30K2026.04 | 26.6 | |
| MathFusionBackbone=DeepSeekMath-7B, #Samples=60K2026.04 | 24.6 | |
| DeepSeekMath-MMIQCBackbone=DeepSeekMath-7B, #Samples=2.3M2026.04 | 23.4 | |
| MathFusion (Sequential)Backbone=DeepSeekMath-7B, #Samples=30K2026.04 | 22.8 | |
| DeepSeekMath-MetaMathBackbone=DeepSeekMath-7B, #Samples=60K2026.04 | 18.9 |