Mathematical Reasoning on iGSM (test)
97AccuracyLow-SFT
Evaluation Results
| Method | Links | |
|---|---|---|
| Low-SFTModel Scale=7B2026.05 | 97 | |
| Anchored LearningModel Scale=7B2026.05 | 96.8 | |
| Low-SFTModel Scale=1.5B2026.05 | 95.4 | |
| Anchored LearningModel Scale=1.5B2026.05 | 95 | |
| Anchored LearningModel Scale=3B2026.05 | 93.6 | |
| Low-SFTModel Scale=3B2026.05 | 93 | |
| BaseModel Scale=7B2026.05 | 22.8 | |
| BaseModel Scale=3B2026.05 | 18.4 | |
| BaseModel Scale=1.5B2026.05 | 14.4 |