Mathematical Reasoning on GSM8K (test) (Accept Rate, Accuracy)
91AccuracySD
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SDTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.04 | 91 | 59.58 | |
| AutoregressiveTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.04 | 90 | — | |
| DIVERSEDTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.04 | 88 | 76.48 | |
| SD (lossy)Target/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.04 | 86 | 60.1 | |
| SpecCascadeTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.04 | 86 | 58.65 | |
| Static EnsembleTarget/Draft model pair=Qwen3-8B/Qwen3-0.6B, Temperature=1, Draft length N=52026.04 | 86 | 67.58 |