def main():
args = get_arguments()
if not os.path.isdir(args.output_dir):
os.makedirs(args.output_dir)
face_dataset = ImageData(root_dir=args.input_dir,\
transform=transforms.Compose([PreprocessData(args.scale_size, args.crop_size)]))
dataloader = DataLoader(face_dataset,
batch_size=args.batch_size,
shuffle=True)
########### setup network ##############
net = ConvNet(3, args.K).to(device)
#----------Weight Initialization---------------
def init_weights(m):
if type(m) == nn.Conv2d:
nn.init.normal_(m.weight, 0, 0.02)
net.apply(init_weights)
#---------------------------------------------
summary(net, (3, 128, 128))
# for name, param in net.named_parameters():
# if param.requires_grad:
# print(name, param.data.size(), type(param))
optimizer = torch.optim.Adam(net.parameters(),
lr=args.learning_rate,
weight_decay=args.weight_decay)
# #########################################
print()
train(dataloader, net, optimizer, args)
optimizer = opt_class(param_dict.values(), lr=args.lr_base, **opt_args)
print(net)
print(param_dict.keys())
#############
# train
#############
# load snapshot
start_iter = load_snapshot(net, args.param_prefix)
if start_iter:
print('==> Load model from {0}{1}'.format(args.param_prefix,
start_iter - 1))
else:
start_iter = 0
net.apply(weights_init)
print('==> Random init')
# lr scheduler
if 'rot' in args.kernel_mode:
lr_schedulers = [
StepLR(optimizer.optim0,
step_size=args.lr_step_size,
gamma=0.1,
last_epoch=-1),
StepLR(optimizer.optim1,
step_size=args.lr_step_size,
gamma=0.3,
last_epoch=-1),
]
lr_schedulers[0].step(start_iter - 1)
