Speculative Decoding on Code (Throughput)
138.72Throughput (tokens/s)EvoSpec
Evaluation Results
| Method | Links | |
|---|---|---|
| EvoSpecStatic Vocabulary Size=32k, Dynamic Buffer Size (Dyna)=2562026.04 | 138.72 | |
| EvoSpecStatic Vocabulary Size=32k, Dynamic Buffer Size (Dyna)=5122026.04 | 137.9 | |
| EvoSpecStatic Vocabulary Size=32k, Dynamic Buffer Size (Dyna)=10242026.04 | 137.14 | |
| EvoSpecStatic Vocabulary Size=16k, Dynamic Buffer Size (Dyna)=10242026.04 | 131.19 | |
| EvoSpecStatic Vocabulary Size=16k, Dynamic Buffer Size (Dyna)=5122026.04 | 129.29 | |
| EvoSpecStatic Vocabulary Size=16k, Dynamic Buffer Size (Dyna)=2562026.04 | 125.77 | |
| FR-SpecStatic Vocabulary Size=32k2026.04 | 123.61 | |
| Vanilla SD2026.04 | 119.56 | |
| EvoSpecStatic Vocabulary Size=16k, Dynamic Vocabulary Size=256, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 118 | |
| EvoSpecStatic Vocabulary Size=32k, Dynamic Vocabulary Size=256, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 116.38 | |
| EvoSpecStatic Vocabulary Size=16k, Dynamic Vocabulary Size=1024, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 116.3 | |
| EvoSpecStatic Vocabulary Size=16k, Dynamic Vocabulary Size=512, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 115.64 | |
| EvoSpecStatic Vocabulary Size=8k, Dynamic Buffer Size (Dyna)=10242026.04 | 115.14 | |
| Standard SDDraft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 114.71 | |
| EvoSpecStatic Vocabulary Size=8k, Dynamic Vocabulary Size=256, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 114.07 | |
| EvoSpecStatic Vocabulary Size=8k, Dynamic Buffer Size (Dyna)=5122026.04 | 113.82 | |
| EvoSpecStatic Vocabulary Size=32k, Dynamic Vocabulary Size=512, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 113.62 | |
| EvoSpecStatic Vocabulary Size=32k, Dynamic Vocabulary Size=1024, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 113.38 | |
| EvoSpecStatic Vocabulary Size=8k, Dynamic Vocabulary Size=512, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 113.34 | |
| EvoSpecStatic Vocabulary Size=8k, Dynamic Vocabulary Size=1024, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 112.63 | |
| EvoSpecStatic Vocabulary Size=8k, Dynamic Buffer Size (Dyna)=2562026.04 | 112.51 | |
| FR-SpecStatic Vocabulary Size=32k, Draft Model=Qwen3-1.7B, Target Model=Qwen3-4B, Architecture=Decoupled2026.04 | 103.34 |