Mathematics on AIME 24 (AVG@32, Pass@32)
74.38AVG@32Projected
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ProjectedScale=14B, Category=Cross-Domain Merging2026.07 | 74.38 | 93.33 | |
| OR-ZeroScale=14B, Category=Math Expert2026.07 | 74.27 | 93.33 | |
| Best Single-Domain ExpertScale=14B2026.07 | 74.27 | 93.33 | |
| TIES MergingScale=14B, Category=Cross-Domain Merging2026.07 | 73.54 | 93.33 | |
| DARE + TIESScale=14B, Category=Cross-Domain Merging2026.07 | 73.06 | 90 | |
| DeepCoderScale=14B, Category=Code Expert2026.07 | 71.04 | 86.67 | |
| Task ArithmeticScale=14B, Category=Cross-Domain Merging2026.07 | 71.04 | 93.33 | |
| Distill-Qwen-14BScale=14B, Category=Single-Domain Reference2026.07 | 67.71 | 90 | |
| ProjectedScale=1.5B, Category=Cross-Domain Merging2026.07 | 43.44 | 76.67 | |
| TIES MergingScale=1.5B, Category=Cross-Domain Merging2026.07 | 42.08 | 73.33 | |
| DeepScaleRScale=1.5B, Category=Math Expert2026.07 | 40.31 | 76.67 | |
| Best Single-Domain ExpertScale=1.5B2026.07 | 40.31 | 76.67 | |
| DARE + TIESScale=1.5B, Category=Cross-Domain Merging2026.07 | 40.05 | 73.33 | |
| Archer-CodeScale=1.5B, Category=Code Expert2026.07 | 39.48 | 76.67 | |
| Task ArithmeticScale=1.5B, Category=Cross-Domain Merging2026.07 | 36.25 | 63.33 | |
| Distill-Qwen-1.5BScale=1.5B, Category=Single-Domain Reference2026.07 | 31.67 | 76.67 |