Math Reasoning on Math Level-5
77.8AccuracyTAB
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TABModel Variant=4B, Budget=10k2026.04 | 77.8 | 6,895 | |
| TABBudget=10k2026.04 | 77.6 | 7,333 | |
| TABModel Variant=4B, Budget=8k2026.04 | 77 | 5,531 | |
| TABBudget=8k2026.04 | 76.8 | 5,473 | |
| TAB All-SubQBudget=10k2026.04 | 76.8 | 6,311 | |
| TAB All-SubQBudget=8k2026.04 | 75.8 | 5,901 | |
| StaticBudget=40962026.04 | 74.6 | 11,058 | |
| L-JProtocol=Multi-Turn, Strategy=All-SubQ2026.04 | 71.2 | 5,744 | |
| L-JProtocol=Multi-Turn2026.04 | 70.8 | 5,588 | |
| TAB All-SubQBudget=5k2026.04 | 70.4 | 3,255 | |
| StaticBudget=20482026.04 | 70.1 | 5,901 | |
| L-JProtocol=Multi-Turn, Model Variant=4B2026.04 | 69.5 | 4,324 | |
| L-JProtocol=Individual2026.04 | 69.3 | 5,961 | |
| L-JProtocol=Individual, Model Variant=4B2026.04 | 69.3 | 4,318 | |
| TABModel Variant=4B, Budget=5k2026.04 | 69.1 | 3,697 | |
| TABBudget=5k2026.04 | 68.1 | 3,498 | |
| StaticBudget=10242026.04 | 66.6 | 3,986 | |
| TABModel Variant=4B, Budget=3k2026.04 | 65.6 | 2,596 | |
| TAB All-SubQBudget=3k2026.04 | 65.4 | 2,636 | |
| TABBudget=3k2026.04 | 65 | 2,516 | |
| StaticBudget=5122026.04 | 63.2 | 2,660 | |
| StaticBudget=2562026.04 | 60.8 | 2,013 |