Mathematical formalization on Gaokao-Formal 495 problems
84.2Correctness @1Kimina-7B
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Kimina-7BTemperature=0.62026.04 | 84.2 | 97.2 | 44.6 | 68.1 | 39.6 | |
| Mathesis-7BTemperature=0.62026.04 | 84.1 | 95.2 | 49.8 | 71.1 | 34.3 | |
| SFT+GRPO-0%Temperature=1.02026.04 | 77.6 | 93.9 | 51.4 | 72.7 | 26.2 | |
| SFT+GRPO-30%Temperature=1.02026.04 | 76.4 | 92.9 | 48.6 | 70.7 | 27.8 | |
| SFT+GRPO-100%Temperature=1.02026.04 | 62.9 | 92.7 | 40.6 | 69.9 | 22.3 | |
| SFTTemperature=1.02026.04 | 61.8 | 92.3 | 41 | 70.9 | 20.8 | |
| GRPO-onlyTemperature=1.02026.04 | 50.9 | 69.3 | 28.1 | 40.2 | 22.8 | |
| BaseTemperature=1.02026.04 | 19.9 | 40.2 | 10.2 | 19.4 | 9.7 |