niniack / resnet-cifar10

ResNet implementation for CIFAR-10 dataset with PyTorch (unofficial).

Implementation of ResNet for CIFAR-10 classification

Requirements

• Python 3.6+
• PyTorch 1.6.0+

Usage

1. Train

mkdir path/to/checkpoint_dir
python train.py --n 3 --checkpoint_dir path/to/checkpoint_dir

n means the network depth, you can choose from {3, 5, 7, 9}, which means ResNet-{20, 32, 44, 56}. For other options, please refer helps: python train.py -h. When you run the code for the first time, the dataset will be downloaded automatically.

2. Test

When your training is done, the model parameter file path/to/checkpoint_dir/model_final.pth will be generated.

python test.py --n 3 --params_path path/to/checkpoint_dir/model_final.pth

Note

If you want to specify GPU to use, you should set environment variable CUDA_VISIBLE_DEVICES=0, for example.

References

• Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun, "Deep residual learning for image recognition," In Proceedings of the IEEE conference on computer vision and pattern recognition, 2016.