List your company

Inference Efficiency on WQ

0.014Relative Execution Time OverheadPerplexity
-0.062960.456520.9761.49548Mar 20, 2026

Evaluation results

MethodLinks
PerplexityModel=Llama-3-8B2026.030.014
PerplexityModel=Llama-2-7B2026.030.015
PerplexityModel=Mistral-7B2026.030.017
PerplexityModel=Qwen2.5-3B2026.030.029
PerplexityModel=Qwen2.5-7B2026.030.031
PerplexityModel=Qwen2.5-14B2026.030.041
STCModel=Llama-2-7B2026.030.473
STCModel=Mistral-7B2026.030.966
STCModel=Qwen2.5-14B2026.031.352
STCModel=Llama-3-8B2026.031.577
STCModel=Qwen2.5-7B2026.031.9
STCModel=Qwen2.5-3B2026.031.938