List your company

Mathematical Reasoning on GSM8K (test) (Accept Rate, Accuracy)

91AccuracySD
85.887.1588.589.85Apr 8, 2026

Evaluation results

MethodLinks
SDTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.049159.58
AutoregressiveTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.0490—
DIVERSEDTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.048876.48
SD (lossy)Target/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.048660.1
SpecCascadeTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.048658.65
Static EnsembleTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.048667.58