Mathematical Reasoning on MATH (Detailed Accuracy & Efficiency)
96.04Algebra AccuracyTandem
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| TandemSeries=Tandem, Model=7B+32B2026.04 | 96.04 | 82.7 | 73.07 | 67.77 | 88.7 | 92.65 | 71.98 | 83.46 | 2,916 | 99.72 | |
| 7B+32B (high)Series=Baseline, Model=7B+32B (high)2026.04 | 95.53 | 82.49 | 72.86 | 67.55 | 88.52 | 92.54 | 71.61 | 83.18 | 2,930 | 104.62 | |
| DeepSeek-32BSeries=Single, Model=32B2026.04 | 94.78 | 82.07 | 68.89 | 64.45 | 85 | 91.04 | 67.22 | 80.9 | 2,630 | 168.35 | |
| 7B+32B (low)Series=Baseline, Model=7B+32B (low)2026.04 | 94.27 | 77.43 | 68.27 | 60.24 | 80.74 | 90.36 | 65.38 | 78.74 | 2,735 | 44.76 | |
| 7B+32B (medium)Series=Baseline, Model=7B+32B (medium)2026.04 | 93.93 | 79.96 | 69.94 | 63.9 | 83.33 | 91.16 | 67.4 | 80.36 | 2,853 | 71.96 | |
| DeepSeek-7BSeries=Single, Model=7B2026.04 | 93.09 | 75.95 | 65.34 | 60.13 | 76.85 | 89.55 | 62.45 | 77.14 | 2,732 | 38.25 |