Code for "Characterising Across Stack Optimisations for Deep Convolutional Neural Networks"
License: MIT License
Makefile 0.71%C 99.29%
characterising-neural-compression's Introduction
You need to include the weights file (e.g. vgg_plain.txt) file generated from scratchpad/codes/pytorch-cifar/
compile: gcc -O3 VGG16_CPU_cifar.c -lm -fopenmp -o VGG16_CPU_cifar
run (example): ./VGG-16_CPU_cifar weights/vgg.txt img/filelist_cf.txt result_cf_conv.txt 1 (only convolutional layers)
run (example): ./VGG-16_CPU_cifar weights/vgg.txt img/filelist_cf.txt result_cf.txt (all layers)