PyTorch wrapper for image classification
- Calculating Params and FLOPs
- Calculate mean std
- Get mean std (for CIFAR-10, CIFAR-100, and ImageNet)
- Cross entropy label smooth
- Get learning rate
- Avgrage meter (including accuracy)
- Create exp dir
- Save checkpoint
The library can be installed with pip:
pip install pooraka
import pooraka as prk
flops, params = prk.get_flops_params(model, (224, 224))
import pooraka as prk
CLASSES = 1000
criterion_smooth = prk.CrossEntropyLabelSmooth(CLASSES, 0.1)
criterion_smooth = criterion_smooth.cuda()