Mathematical Reasoning on MATH 500 (R0-R3 Coverage)
88.8R0 CoverageDA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DABackbone=GPT-4.1-mini2026.04 | 88.8 | 90.6 | 91.8 | 92.2 | |
| ElimBackbone=GPT-4.1-mini2026.04 | 88.8 | 91 | 92.4 | 92.4 | |
| Self-Evolving Test-Time Allocation (Rand)Backbone=GPT-4.1-mini, Strategy=Rand2026.04 | 88.8 | 91.2 | 93.2 | 94.4 | |
| Self-Evolving Test-Time Allocation (Adpt)Backbone=GPT-4.1-mini, Strategy=Adpt2026.04 | 88.8 | 92.8 | 93.8 | 95.6 |