Awesome

BoostResNet

This repository contains a simple PyTorch implementation of the article Learning Deep ResNet Blocks Sequentially using Boosting Theory.

The program brn.py assumes the existence of a dataset in torch format that is already normalized. It uses a 50-layer ResNet architecture from Facebook that takes 32 x 32 images as input, but can easily be modified to accomodate other architectures.

python brn.py --data CIFAR.t7 --transform