Math Reasoning on MINERVA (Coverage and Average Success Rate)
39.3CoverageTOFU
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TOFUBackbone=DEEPSEEK-MATH-7B2026.04 | 39.3 | 12 | |
| TOFUBackbone=QWEN-2.5-MATH-7B2026.04 | 38.6 | 14.7 | |
| GEMBackbone=DEEPSEEK-MATH-7B2026.04 | 35.3 | 12.5 | |
| GEMBackbone=QWEN-2.5-MATH-7B2026.04 | 34.9 | 15.1 | |
| CEBackbone=DEEPSEEK-MATH-7B2026.04 | 34.2 | 12.5 | |
| TOFUBackbone=QWEN-2.5-MATH-1.5B2026.04 | 33.5 | 10.7 | |
| GEMBackbone=QWEN-2.5-MATH-1.5B2026.04 | 31.3 | 12.4 | |
| CEBackbone=QWEN-2.5-MATH-7B2026.04 | 30.9 | 15.1 | |
| CEBackbone=QWEN-2.5-MATH-1.5B2026.04 | 28.3 | 12.1 |