The unoffical PyTorch implementation of WeightNet, based on pycls.
See INSTALL.md
and GETTING_STARTED.md
. Learn more at pycls's documentation.
NOTE: We use the pre-trained ShuffleNet V2 1.0x as the default backbone.
python tools/train_net.py --cfg configs/shufflenet/ShuffleNet_1x_imagenet_4gpu.yaml
# folder: output/ShuffleNet_1x_imagenet_4gpu
# result: top1: top5:
python tools/train_net.py --cfg configs/shufflenet/WeightNet_1x_imagenet_4gpu.yaml
# folder: output/WeightNet_1x_imagenet_4gpu
# result: top1: top5: