List your company

LLM Inference on Aggregate Suite (Alpaca, CodeAlpaca, HumanEval, LiveCodeBench, Math500, MBPP, MT-Bench)

2.87Mean SpeedupDART
0.90441.41471.9252.4353Jan 27, 2026

Evaluation results

MethodLinks
DARTModel=Qwen4B, Temperature=02026.012.873.87
DARTModel=L2 7B, Temperature=02026.012.854.08
DARTModel=Qwen14B, Temperature=02026.012.773.67
DARTModel=Qwen8B, Temperature=02026.012.713.61
HydraModel=L2 7B, Temperature=02026.012.663.55
DARTModel=Qwen1.7B, Temperature=02026.012.613.6
DARTModel=Qwen14B, Temperature=12026.012.473.61
DARTModel=Qwen32B, Temperature=02026.012.423.76
MedusaModel=L2 7B, Temperature=02026.012.242.68
EAGLE3Model=Qwen8B, Temperature=02026.012.23.72
DARTModel=Qwen32B, Temperature=12026.012.193.55
EAGLE3Model=Qwen4B, Temperature=02026.012.123.54
EAGLE3Model=Qwen32B, Temperature=02026.012.113.85
EAGLE3Model=Qwen14B, Temperature=02026.012.023.48
EAGLE3Model=Qwen1.7B, Temperature=02026.012.013.8
EAGLE3Model=Qwen32B, Temperature=12026.011.973.67
EAGLE3Model=Qwen14B, Temperature=12026.011.893.38
PLDModel=L2 7B, Temperature=02026.011.741.92
LookaheadModel=L2 7B, Temperature=02026.011.611.81
SPSModel=Qwen32B, Temperature=02026.011.093.45
SPSModel=Qwen14B, Temperature=02026.010.984.17