Mathematical Reasoning on AIME24 Full Standard (Avg@16, Maj@16)
1.9Average Score @16+QK-Restore
Evaluation results
| Method | Links | ||
|---|---|---|---|
| +QK-RestoreModel=Jet-Nemotron-2B2026.06 | 1.9 | 6.7 | |
| +SFTModel=Jet-Nemotron-2B2026.06 | 1.5 | 3.3 | |
| +SFTModel=HypeNet-9B2026.06 | 1.2 | 10 | |
| Pre-trainModel=Jet-Nemotron-2B2026.06 | 0.8 | 3.3 | |
| Pre-trainModel=HypeNet-9B2026.06 | 0.8 | 6.7 | |
| +SFTModel=HypeNet-5B2026.06 | 0.6 | 0 | |
| +QK-RestoreModel=HypeNet-9B2026.06 | 0.6 | 3.3 | |
| +QK-RestoreModel=HypeNet-5B2026.06 | 0.4 | 3.3 | |
| +QK-RestoreModel=HypeNet-2B2026.06 | 0.2 | 3.3 | |
| Pre-trainModel=HypeNet-2B2026.06 | 0 | 0 | |
| +SFTModel=HypeNet-2B2026.06 | 0 | 0 | |
| Pre-trainModel=HypeNet-5B2026.06 | 0 | 0 |