Mathematical Reasoning on MATH500 (Pass@1 Accuracy and Step Count)
43.4Pass@1 AccuracyHEX
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HEXGen. Len.=256, Training Necessity=Training-free2026.02 | 43.4 | 3,200 | |
| DVOTINGGen. Len.=256, Training Necessity=Training-free2026.02 | 41.2 | 443.1 | |
| Majority VotingGen. Len.=256, Training Necessity=Training-free2026.02 | 41 | 640 | |
| HEXGen. Len.=128, Training Necessity=Training-free2026.02 | 39.6 | 1,600 | |
| DVOTINGGen. Len.=128, Training Necessity=Training-free2026.02 | 34.4 | 280.1 | |
| LLaDA-1.5*Gen. Len.=256, Training Necessity=None2026.02 | 33.4 | 256 | |
| Majority VotingGen. Len.=128, Training Necessity=Training-free2026.02 | 32 | 320 | |
| LLaDA-1.5*Gen. Len.=128, Training Necessity=None2026.02 | 30.6 | 128 |