Tag: #가지치기
Kubernetes, DevOps and MLOps, written from things we ran and measured ourselves · 1 posts
Building Smaller Models: Depthwise Separable Convolution, Pruning, Distillation, and EfficientTAM
If reducing precision and leaving things to the compiler is not enough, you must make the model itself smaller. We look at MobileNet, which splits convolutions into spatial viewing (depthwise 3x3) and channel mixing (1x1
2026-10-15 · 17 min read #비전 모델#실시간#MobileNet#EfficientNet#가지치기