Mathematical Reasoning on MinervaMath (R0, R1, R2, R3)
41.5R0 ScoreDA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DABackbone=Gemini-2.5-flash-lite, Allocation Strategy=Difficulty-Aware, Thinking Mode=enabled2026.04 | 41.5 | 48.2 | 51.1 | 51.5 | |
| ElimBackbone=Gemini-2.5-flash-lite, Allocation Strategy=Elimination, Thinking Mode=enabled2026.04 | 41.5 | 47.8 | 51.8 | 52.9 | |
| Self-Evolving Test-Time Allocation (Rand)Backbone=Gemini-2.5-flash-lite, Allocation Strategy=Random, Thinking Mode=enabled2026.04 | 41.5 | 48.9 | 50.7 | 52.2 | |
| Self-Evolving Test-Time Allocation (Adpt)Backbone=Gemini-2.5-flash-lite, Allocation Strategy=Adaptive, Thinking Mode=enabled2026.04 | 41.5 | 60.3 | 64.3 | 66.2 |