Math on AIME 25 (Speedup, τ)
7.96SpeedupBASTION
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BASTIONModel=Qwen3-4B, Temperature=0, Hardware=H1002026.05 | 7.96 | 9.43 | |
| BASTIONModel=Qwen3-4B, Temperature=1, Hardware=H1002026.05 | 7.41 | 9.16 | |
| BASTIONModel=Qwen3-8B, Temperature=1, Hardware=H1002026.05 | 7.1 | 8.81 | |
| BASTIONModel=Qwen3-8B, Temperature=0, Hardware=H1002026.05 | 6.88 | 9.35 | |
| DFlashModel=Qwen3-4B, Temperature=0, Hardware=H100, Block size=162026.05 | 6.45 | 7.26 | |
| DFlashModel=Qwen3-8B, Temperature=0, Hardware=H100, Block size=162026.05 | 5.95 | 7.18 | |
| DFlashModel=Qwen3-4B, Temperature=1, Hardware=H100, Block size=162026.05 | 4.13 | 4.97 | |
| DFlashModel=Qwen3-8B, Temperature=1, Hardware=H100, Block size=162026.05 | 3.98 | 4.83 | |
| EAGLE-3Model=Qwen3-4B, Temperature=0, Hardware=H100, Tree budget size=602026.05 | 3.13 | 2.51 | |
| EAGLE-3Model=Qwen3-8B, Temperature=0, Hardware=H100, Tree budget size=602026.05 | 2.85 | 2.5 | |
| EAGLE-3Model=Qwen3-4B, Temperature=1, Hardware=H100, Tree budget size=602026.05 | 2.5 | 2.16 | |
| EAGLE-3Model=Qwen3-8B, Temperature=1, Hardware=H100, Tree budget size=602026.05 | 2.47 | 2.06 |