else:
net = WideResNet(args.layers,
num_classes,
args.widen_factor,
dropRate=args.droprate)
start_epoch = 0
# Restore model if desired
if args.load != '':
for i in range(1000 - 1, -1, -1):
model_name = os.path.join(
args.load, args.dataset + '_' + args.model + '_baseline_epoch_' +
str(i) + '.pt')
if os.path.isfile(model_name):
net.load_state_dict(torch.load(model_name))
print('Model restored! Epoch:', i)
start_epoch = i + 1
break
if start_epoch == 0:
assert False, "could not resume"
if args.ngpu > 1:
net = torch.nn.DataParallel(net, device_ids=list(range(args.ngpu)))
if args.ngpu > 0:
net.cuda()
torch.cuda.manual_seed(1)
cudnn.benchmark = True # fire on all cylinders
test_data = dset.CIFAR10('/data/sauravkadavath/cifar10-dataset/', train=False, transform=test_transform)
num_classes = 10
else:
test_data = dset.CIFAR100('/data/sauravkadavath/cifar10-dataset/', train=False, transform=test_transform)
num_classes = 100
test_loader = torch.utils.data.DataLoader(
test_data, batch_size=args.batch_size, shuffle=False,
num_workers=args.prefetch, pin_memory=True)
# Create model
if args.model == 'allconv':
net = AllConvNet(num_classes).cuda()
net.load_state_dict(torch.load(args.load))
adversary = attacks.PGD(epsilon=8./255, num_steps=10, step_size=0.5/255).cuda()
# def train():
# net.train() # enter train mode
# loss_avg = 0.0
# for bx, by in tqdm(train_loader):
# bx, by, = bx.cuda(), by.cuda()
# adv_bx = adversary(net, bx, by)
# # print(torch.max(bx), torch.min(bx), torch.mean(bx))
# # forward
