CIFAR10 with PyTorch

Pytorch implementation for CIFAR10 classification.

Accuracy

Models Acc
VGG16 .9387
resnet18 .9525
resnet34 .9555
resnet50 .9521
resnet101 .9572

Issue

Multi-gpu training is implemented, but not yet tested.

Memory issue when evaluating.

Model in this repository use more channels than original residual network.

Reference

Model files are the copies of [https://github.com/kuangliu/pytorch-cifar]