List your company

Math Reasoning on MATH 500 (Speedup, τ)

4.37SpeedupBlockPilot
1.62442.33723.053.7628Jun 25, 2026Jun 26, 2026Jun 27, 2026Jun 28, 2026Jun 29, 2026Jun 30, 2026

Evaluation results

MethodLinks
BlockPilotModel=Qwen2.5-Coder-32B-Instruct, Temperature=02026.064.376.14
BlockPilotModel=Qwen2.5-Coder-32B-Instruct, Temperature=12026.064.015.87
BlockPilotModel=Llama-3.1-8B-Instruct, Temperature=02026.063.214.79
HyperDFlashTemperature=0, Number of drafted steps=6, Mode=Non-thinking2026.062.933.95
HyperDFlashTemperature=1, Number of drafted steps=6, Mode=Non-thinking2026.062.723.67
MTPTemperature=0, Number of drafted steps=3, Mode=Non-thinking2026.062.363.13
MTPTemperature=1, Number of drafted steps=3, Mode=Non-thinking2026.062.242.96
BlockPilotModel=Llama-3.1-8B-Instruct, Temperature=12026.062.13.45
MTPTemperature=0, Number of drafted steps=6, Mode=Non-thinking2026.061.883.39
Vanilla DFlashTemperature=0, Number of drafted steps=6, Mode=Non-thinking2026.061.822.35
Vanilla DFlashTemperature=1, Number of drafted steps=6, Mode=Non-thinking2026.061.752.25
MTPTemperature=1, Number of drafted steps=6, Mode=Non-thinking2026.061.733.15