List your company

LLM Serving Performance on Qwen 7B (Zipf alpha=1.5, 32 variants)

0.87Throughput (req/s)Delta-CoMe
0.15240.33870.5250.7113Jun 5, 2025

Evaluation results

MethodLinks
Delta-CoMearrival rate (lambda)=1.02025.060.870.68
PRINMIXarrival rate (lambda)=1.02025.060.870.62
Delta-CoMearrival rate (lambda)=0.52025.060.420.55
PRINMIXarrival rate (lambda)=0.52025.060.420.52
Naivearrival rate (lambda)=0.52025.060.2152.42
Naivearrival rate (lambda)=1.02025.060.18198.48