Inference Efficiency on WQ
0.014Relative Execution Time OverheadPerplexity
Evaluation Results
| Method | Links | |
|---|---|---|
| PerplexityModel=Llama-3-8B2026.03 | 0.014 | |
| PerplexityModel=Llama-2-7B2026.03 | 0.015 | |
| PerplexityModel=Mistral-7B2026.03 | 0.017 | |
| PerplexityModel=Qwen2.5-3B2026.03 | 0.029 | |
| PerplexityModel=Qwen2.5-7B2026.03 | 0.031 | |
| PerplexityModel=Qwen2.5-14B2026.03 | 0.041 | |
| STCModel=Llama-2-7B2026.03 | 0.473 | |
| STCModel=Mistral-7B2026.03 | 0.966 | |
| STCModel=Qwen2.5-14B2026.03 | 1.352 | |
| STCModel=Llama-3-8B2026.03 | 1.577 | |
| STCModel=Qwen2.5-7B2026.03 | 1.9 | |
| STCModel=Qwen2.5-3B2026.03 | 1.938 |