List your company

Code Generation on HumanEval (Average Acceptance Length (τ), Speedup Ratio)

7.23Avg Acceptance Length (τ)PPOW
5.4625.9216.386.839May 14, 2026

Evaluation results

MethodLinks
PPOWModel=L31-8B, Temperature=0.02026.057.234.14
PPOWModel=Q3-8B, Temperature=0.02026.057.013.62
PPOWModel=L33-70B, Temperature=0.02026.056.964.82
PPOWModel=Q3-32B, Temperature=0.02026.056.914.16
EAGLE-3Model=L33-70B, Temperature=0.02026.056.784.8
GRIFFINModel=L31-8B, Temperature=0.02026.056.723.96
GRIFFINModel=L33-70B, Temperature=0.02026.056.694.78
EAGLE-3Model=Q3-8B, Temperature=0.02026.056.683.4
EAGLE-3Model=L31-8B, Temperature=0.02026.056.633.92
PPOWModel=L33-70B, Temperature=1.02026.056.574.62
EAGLE-3Model=Q3-32B, Temperature=0.02026.056.524.02
PPOWModel=L31-8B, Temperature=1.02026.056.333.37
GRIFFINModel=L33-70B, Temperature=1.02026.056.314.46
EAGLE-3Model=L33-70B, Temperature=1.02026.056.284.43
PPOWModel=Q3-32B, Temperature=1.02026.056.172.77
GRIFFINModel=L31-8B, Temperature=1.02026.056.013.29
PPOWModel=Q3-8B, Temperature=1.02026.055.962.33
EAGLE-3Model=L31-8B, Temperature=1.02026.055.943.25
EAGLE-3Model=Q3-8B, Temperature=1.02026.055.712.11
EAGLE-3Model=Q3-32B, Temperature=1.02026.055.532.59