Reasoning on Reasoning
100Math ScoreMath expert
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Math expert2026.06 | 100 | 95.7 | 78.4 | 43.8 | 79.5 | |
| IR3DEk=102026.06 | 98.4 | 99.9 | 94.5 | 100.6 | 98.4 | |
| IR3DE-avgk=10, aggregation=avg2026.06 | 97.3 | 99.9 | 85.6 | 101.2 | 96 | |
| kNN routerk=12026.06 | 95.8 | 99.5 | 96.3 | 98.7 | 97.6 | |
| IR3DE-allk=all2026.06 | 91.5 | 99.1 | 89.2 | 100.1 | 95 | |
| Experts average2026.06 | 45.9 | 100 | 83.9 | 53.4 | 70.8 | |
| Random router2026.06 | 37.7 | 98.4 | 83.8 | 59.3 | 69.8 | |
| MoDEM-smallSize=small2026.06 | 29.5 | 98.6 | 69.5 | 100.3 | 74.5 | |
| MoDEM-largeSize=large2026.06 | 29.5 | 96.7 | 67.7 | 95.3 | 72.3 | |
| Instr. following expert2026.06 | 28.9 | 99.7 | 80.1 | 100 | 77.2 | |
| Coding expert2026.06 | 11.7 | 99.1 | 100 | 55.2 | 66.5 | |
| Multilingual expert2026.06 | 6.1 | 100 | 73.1 | 21.2 | 50.1 |