Math Reasoning on MATH 500 (Speedup, τ)
4.37SpeedupBlockPilot
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BlockPilotModel=Qwen2.5-Coder-32B-Instruct, Temperature=02026.06 | 4.37 | 6.14 | |
| BlockPilotModel=Qwen2.5-Coder-32B-Instruct, Temperature=12026.06 | 4.01 | 5.87 | |
| BlockPilotModel=Llama-3.1-8B-Instruct, Temperature=02026.06 | 3.21 | 4.79 | |
| HyperDFlashTemperature=0, Number of drafted steps=6, Mode=Non-thinking2026.06 | 2.93 | 3.95 | |
| HyperDFlashTemperature=1, Number of drafted steps=6, Mode=Non-thinking2026.06 | 2.72 | 3.67 | |
| MTPTemperature=0, Number of drafted steps=3, Mode=Non-thinking2026.06 | 2.36 | 3.13 | |
| MTPTemperature=1, Number of drafted steps=3, Mode=Non-thinking2026.06 | 2.24 | 2.96 | |
| BlockPilotModel=Llama-3.1-8B-Instruct, Temperature=12026.06 | 2.1 | 3.45 | |
| MTPTemperature=0, Number of drafted steps=6, Mode=Non-thinking2026.06 | 1.88 | 3.39 | |
| Vanilla DFlashTemperature=0, Number of drafted steps=6, Mode=Non-thinking2026.06 | 1.82 | 2.35 | |
| Vanilla DFlashTemperature=1, Number of drafted steps=6, Mode=Non-thinking2026.06 | 1.75 | 2.25 | |
| MTPTemperature=1, Number of drafted steps=6, Mode=Non-thinking2026.06 | 1.73 | 3.15 |