/PyTorch-MixNet-SS

Extremely light-weight MixNet with Top-1 75.7% and 2.5M params

Primary LanguagePythonBSD 2-Clause "Simplified" LicenseBSD-2-Clause

PyTorch-MixNet-SS

Extremely light-weight MixNet with ImageNet Top-1 75.7% accuracy and 2.5M parameters.

Precision Top-1 (%) Top-5 (%) Params
FP32 75.744 92.576 2.5 M
FP16 75.714 92.570 1.3 M

1. Download Link

2. How to load params ?

from collections import OrderedDict

state_dict = torch.load(args.pretrained)
new_state_dict = OrderedDict()
for key_ori, key_pre in zip(model.state_dict().keys(), state_dict.keys()):
    new_state_dict[key_ori] = state_dict[key_pre]
model.load_state_dict(new_state_dict)

3. Inference setting :

normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                     std=[0.225, 0.225, 0.225])

val_loader = torch.utils.data.DataLoader(
        datasets.ImageFolder(valdir, transforms.Compose([
            transforms.Resize(256, interpolation=Image.BICUBIC), # == 256
            transforms.CenterCrop(224),
            transforms.ToTensor(),
            normalize,
        ])),
        batch_size=args.batch_size, shuffle=False,
        num_workers=args.workers, pin_memory=True)