Mathematical Reasoning on GSM8K (0-shot)
88.78Pass@1HEX
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HEXGen. Len.=512, Training Necessity=Training-free2026.02 | 88.78 | 6,400 | |
| DVOTINGGen. Len.=512, Training Necessity=Training-free2026.02 | 87.72 | 289.6 | |
| Majority VotingGen. Len.=512, Training Necessity=Training-free2026.02 | 86.28 | 1,280 | |
| HEXGen. Len.=256, Training Necessity=Training-free2026.02 | 85.75 | 3,200 | |
| DVOTINGGen. Len.=256, Training Necessity=Training-free2026.02 | 83.78 | 237.1 | |
| IGPOGen. Len.=256, Training Necessity=RL2026.02 | 83.6 | 256 | |
| Majority VotingGen. Len.=256, Training Necessity=Training-free2026.02 | 82.33 | 640 | |
| wd1Gen. Len.=512, Training Necessity=RL2026.02 | 82.3 | 256 | |
| d1Gen. Len.=512, Training Necessity=SFT + RL2026.02 | 82.1 | 256 | |
| LLaDA-8B-InstructGen. Len.=512, Training Necessity=Pre-training2026.02 | 81.5 | 512 | |
| RFGGen. Len.=256, Training Necessity=FT or RL2026.02 | 81.3 | 512 | |
| d1Gen. Len.=256, Training Necessity=SFT + RL2026.02 | 81.1 | 128 | |
| wd1Gen. Len.=256, Training Necessity=RL2026.02 | 80.8 | 128 | |
| HEXGen. Len.=128, Training Necessity=Training-free2026.02 | 80.14 | 1,600 | |
| DVOTINGGen. Len.=128, Training Necessity=Training-free2026.02 | 78.24 | 170.4 | |
| LLaDA-8B-InstructGen. Len.=256, Training Necessity=Pre-training2026.02 | 76.72 | 256 | |
| Majority VotingGen. Len.=128, Training Necessity=Training-free2026.02 | 76.72 | 320 | |
| d1Gen. Len.=128, Training Necessity=SFT + RL2026.02 | 73.2 | 64 | |
| LLaDA-8B-InstructGen. Len.=128, Training Necessity=Pre-training2026.02 | 70.58 | 128 |