train=True,
transform=test_transform)
test_data = dset.CIFAR10('/share/data/vision-greg/cifarpy',
train=False,
transform=test_transform)
num_classes = 10
else:
train_data = dset.CIFAR100('/share/data/vision-greg/cifarpy',
train=True,
transform=test_transform)
test_data = dset.CIFAR100('/share/data/vision-greg/cifarpy',
train=False,
transform=test_transform)
num_classes = 100
train_data, val_data = validation_split(train_data, val_share=0.1)
val_loader = torch.utils.data.DataLoader(val_data,
batch_size=args.test_bs,
shuffle=False,
num_workers=args.prefetch,
pin_memory=True)
test_loader = torch.utils.data.DataLoader(test_data,
batch_size=args.test_bs,
shuffle=False,
num_workers=args.prefetch,
pin_memory=True)
# Create model
if 'allconv' in args.method_name:
net = AllConvNet(num_classes)
else:
torch.manual_seed(1)
np.random.seed(1)
train_data_in = svhn.SVHN('/share/data/vision-greg/svhn/',
split='train_and_extra',
transform=trn.ToTensor(),
download=False)
test_data = svhn.SVHN('/share/data/vision-greg/svhn/',
split='test',
transform=trn.ToTensor(),
download=False)
num_classes = 10
calib_indicator = ''
if args.calibration:
train_data_in, val_data = validation_split(train_data_in,
val_share=5000 / 604388.)
calib_indicator = 'calib_'
tiny_images = TinyImages(transform=trn.Compose([
trn.ToTensor(),
trn.ToPILImage(),
trn.RandomHorizontalFlip(),
trn.ToTensor()
]))
train_loader_in = torch.utils.data.DataLoader(train_data_in,
batch_size=args.batch_size,
shuffle=True,
num_workers=args.prefetch,
pin_memory=True)
