def main():
device = 'cuda' if torch.cuda.is_available() else 'cpu'
torch.backends.cudnn.benchmark = True
net = create_network()
net.to(device)
criterion = config.create_loss_function().to(device)
optimizer = config.create_optimizer(net.parameters())
lr_scheduler = config.create_lr_scheduler(optimizer)
ds_train = create_train_dataset(args.batch_size)
ds_val = create_test_dataset(args.batch_size)
train_attack = config.create_attack_method(device)
eval_attack = config.create_evaluation_attack_method(device)
now_epoch = 0
if args.auto_continue:
args.resume = os.path.join(config.model_dir, 'last.checkpoint')
if args.resume is not None and os.path.isfile(args.resume):
now_epoch = load_checkpoint(args.resume, net, optimizer, lr_scheduler)
for i in range(now_epoch, config.num_epochs):
descr_str = 'Epoch:{}/{} -- lr:{}'.format(
i, config.num_epochs,
lr_scheduler.get_last_lr()[0])
cleanacc, advacc = train_one_epoch(net,
ds_train,
optimizer,
criterion,
device,
descr_str,
train_attack,
adv_coef=args.adv_coef)
tb_train_dic = {'Acc': cleanacc, 'YofoAcc': advacc}
print('Train: {}'.format(tb_train_dic))
if config.eval_interval > 0 and i % config.eval_interval == 0:
acc, advacc = eval_one_epoch(net, ds_val, device, eval_attack)
tb_val_dic = {'Acc': acc, 'AdvAcc': advacc}
print('Eval: {}'.format(tb_val_dic))
lr_scheduler.step()
save_checkpoint(i,
net,
optimizer,
lr_scheduler,
file_name=os.path.join(config.model_dir,
'advtrain.{}'.format(i)))
acc, advacc = eval_one_epoch(net, ds_val, device, eval_attack)
tb_val_dic = {'Acc': acc, 'AdvAcc': advacc}
print('Eval: {}'.format(tb_val_dic))
def main():
device = 'cuda' if torch.cuda.is_available() else 'cpu'
net = create_network()
net.to(device)
criterion = config.create_loss_function().to(device)
optimizer = config.create_optimizer(net.other_layers.parameters())
lr_scheduler = config.create_lr_scheduler(optimizer)
Hamiltonian_func = Hamiltonian(net.layer_one, config.weight_decay)
layer_one_optimizer = optim.SGD(net.layer_one.parameters(), lr=lr_scheduler.get_last_lr()[0], momentum=0.9,
weight_decay=5e-4)
lyaer_one_optimizer_lr_scheduler = optim.lr_scheduler.MultiStepLR(layer_one_optimizer,
milestones=[70, 90, 100], gamma=0.1)
LayerOneTrainer = FastGradientLayerOneTrainer(Hamiltonian_func, layer_one_optimizer,
config.inner_iters, config.sigma, config.eps)
ds_train = create_train_dataset(args.batch_size)
ds_eval = create_test_dataset(args.batch_size)
eval_attack = config.create_evaluation_attack_method(device)
now_epoch = 0
if args.auto_continue:
args.resume = os.path.join(config.model_dir, 'last.checkpoint')
if args.resume is not None and os.path.isfile(args.resume):
now_epoch = load_checkpoint(args.resume, net, optimizer, lr_scheduler)
for i in range(now_epoch, config.num_epochs):
descrip_str = 'Training epoch:{}/{} -- lr:{}'.format(i, config.num_epochs,
lr_scheduler.get_last_lr()[0])
acc, yofoacc = train_one_epoch(net, ds_train, optimizer, criterion, LayerOneTrainer, config.K,
device, descrip_str)
tb_train_dic = {'Acc': acc, 'YofoAcc': yofoacc}
print('Train: {}'.format(tb_train_dic))
if config.eval_interval > 0 and i % config.eval_interval == 0:
acc, advacc = eval_one_epoch(net, ds_eval, device, eval_attack)
tb_val_dic = {'Acc': acc, 'AdvAcc': advacc}
print('Eval: {}'.format(tb_val_dic))
lr_scheduler.step()
lyaer_one_optimizer_lr_scheduler.step()
save_checkpoint(i, net, optimizer, lr_scheduler,
file_name=os.path.join(config.model_dir, 'epoch-{}.checkpoint'.format(i)))
acc, advacc = eval_one_epoch(net, ds_eval, device, eval_attack)
tb_val_dic = {'Acc': acc, 'AdvAcc': advacc}
print('Eval: {}'.format(tb_val_dic))
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--resume',
'--resume',
default='log/models/epoch-104.checkpoint',
type=str,
metavar='PATH')
parser.add_argument('-d', type=int, default=0, help='Which gpu to use')
args = parser.parse_args()
device = 'cuda' if torch.cuda.is_available() else 'cpu'
torch.backends.cudnn.benchmark = True
net = create_network()
if torch.cuda.device_count() > 1:
print('Let\'s use {} GPUs!'.format(torch.cuda.device_count()))
net = nn.DataParallel(net)
net.to(device)
ds_val = create_test_dataset()
attack_method = config.create_evaluation_attack_method(device)
if os.path.isfile(args.resume):
load_checkpoint(args.resume, net)
print('Evaluating')
clean_acc, adv_acc = eval_one_epoch(net, ds_val, device, attack_method)
print('clean acc: {}, adv acc: {}'.format(clean_acc, adv_acc))
file1 = open(os.path.join(config.model_dir, 'mi_eval_results.txt'), 'w')
file1.write('MI Eval, clean acc: {}, adv acc: {}'.format(
clean_acc, adv_acc))
file1.close()
def main():
model_dir = args.model_dir
if not os.path.exists(model_dir):
os.makedirs(model_dir)
# use_cuda = not args.no_cuda and torch.cuda.is_available()
torch.manual_seed(args.seed)
device = 'cuda' if torch.cuda.is_available() else 'cpu'
# kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {}
model = create_network().to(device)
optimizer = optim.SGD(model.parameters(),
lr=args.lr,
momentum=args.momentum,
weight_decay=args.weight_decay)
eval_attack = config.create_evaluation_attack_method(device)
train_loader = create_train_dataset(args.batch_size)
test_loader = create_test_dataset(args.batch_size)
now_train_time = 0
for i in range(1, args.epochs + 1):
# adjust learning rate for SGD
adjust_learning_rate(optimizer, i)
s_time = time()
descr_str = 'Epoch: {}/{}'.format(i, args.epochs)
# adversarial training
train(args, model, device, train_loader, optimizer, i, descr_str)
now_train_time += time() - s_time
if args.log_interval > 0 and i % args.log_interval == 0:
acc, advacc = eval_one_epoch(model, test_loader, device,
eval_attack)
tb_val_dic = {'Acc': acc, 'AdvAcc': advacc}
print('Eval: {}'.format(tb_val_dic))
torch.save(
model.state_dict(),
os.path.join(config.model_dir, 'model-wideres-epoch{}.pt'.format(i)))
acc, advacc = eval_one_epoch(model, test_loader, device, eval_attack)
tb_val_dic = {'Acc': acc, 'AdvAcc': advacc}
print('Eval: {}'.format(tb_val_dic))
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--resume', '--resume', default='log/models/last.checkpoint',
type=str, metavar='PATH',
help='path to latest checkpoint (default:log/last.checkpoint)')
parser.add_argument('-d', type=int, default=0, help='Which gpu to use')
args = parser.parse_args()
device = 'cuda' if torch.cuda.is_available() else 'cpu'
torch.backends.cudnn.benchmark = True
net = create_network()
net.to(device)
ds_eval = create_test_dataset(100)
attack_method = config.create_evaluation_attack_method(device)
if os.path.isfile(args.resume):
load_checkpoint(args.resume, net)
print('Evaluating')
clean_acc, adv_acc = eval_one_epoch(net, ds_eval, device, attack_method)
print('clean acc: {}; adv acc: {}'.format(clean_acc, adv_acc))
break
now_epoch = now_epoch + 1
descrip_str = 'Training epoch:{}/{} -- lr:{}'.format(
now_epoch, config.num_epochs,
lr_scheduler.get_lr()[0])
s_time = time.time()
acc, yofoacc = train_one_epoch(net, ds_train, optimizer, criterion,
LayerOneTrainer, config.K, DEVICE,
descrip_str)
now_train_time = now_train_time + time.time() - s_time
tb_train_dic = {'Acc': acc, 'YofoAcc': yofoacc}
print(tb_train_dic)
writer.add_scalars('Train', tb_train_dic, now_epoch)
if config.val_interval > 0 and now_epoch % config.val_interval == 0:
acc, advacc = eval_one_epoch(net, ds_val, DEVICE, EvalAttack)
tb_val_dic = {'Acc': acc, 'AdvAcc': advacc}
writer.add_scalars('Val', tb_val_dic, now_epoch)
tb_val_dic['time'] = now_train_time
log_str = json.dumps(tb_val_dic)
with open('time.log', 'a') as f:
f.write(log_str + '\n')
lr_scheduler.step()
lyaer_one_optimizer_lr_scheduler.step()
delta = datetime.datetime.now() - start_time
duration = delta.days * 24 * 60 + (delta.seconds / 60) # in minutes
log_file.write("{} {}\n".format(now_epoch, duration))
print("Epoch: {}, Duration: {}".format(now_epoch, duration))
save_checkpoint(now_epoch,
net,
def main():
device = 'cuda' if torch.cuda.is_available() else 'cpu'
torch.backends.cudnn.benchmark = True
# writer = SummaryWriter(log_dir=config.log_dir)
net = create_network()
net.to(device)
criterion = config.create_loss_function().to(device)
# criterion = CrossEntropyWithWeightPenalty(net.other_layers, DEVICE, config.weight_decay)#.to(DEVICE)
# ce_criterion = nn.CrossEntropyLoss().to(DEVICE)
optimizer = config.create_optimizer(net.other_layers.parameters())
lr_scheduler = config.create_lr_scheduler(optimizer)
# Make Layer One trainer This part of code should be writen in config.py
Hamiltonian_func = Hamiltonian(net.layer_one, config.weight_decay)
layer_one_optimizer = optim.SGD(net.layer_one.parameters(),
lr=lr_scheduler.get_last_lr()[0],
momentum=0.9,
weight_decay=5e-4)
layer_one_optimizer_lr_scheduler = optim.lr_scheduler.MultiStepLR(
layer_one_optimizer, milestones=[55, 75, 95], gamma=0.1)
LayerOneTrainer = FastGradientLayerOneTrainer(Hamiltonian_func,
layer_one_optimizer,
config.inner_iters,
config.sigma, config.eps)
ds_train = create_train_dataset(args.batch_size)
ds_val = create_test_dataset(args.batch_size)
# TrainAttack = config.create_attack_method(DEVICE)
eval_attack = config.create_evaluation_attack_method(device)
now_epoch = 0
if args.auto_continue:
args.resume = os.path.join(config.model_dir, 'last.checkpoint')
if args.resume is not None and os.path.isfile(args.resume):
now_epoch = load_checkpoint(args.resume, net, optimizer, lr_scheduler)
for i in range(now_epoch, config.num_epochs):
descrip_str = 'Epoch:{}/{} -- lr:{}'.format(
i, config.num_epochs,
lr_scheduler.get_last_lr()[0])
acc, yofoacc = train_one_epoch(net, ds_train, optimizer, criterion,
LayerOneTrainer, config.K, device,
descrip_str)
tb_train_dic = {'Acc': acc, 'YofoAcc': yofoacc}
print('Train: {}'.format(tb_train_dic))
if config.eval_interval > 0 and i % config.eval_interval == 0:
acc, advacc = eval_one_epoch(net, ds_val, device, eval_attack)
tb_val_dic = {'Acc': acc, 'AdvAcc': advacc}
print('Eval: {}'.format(tb_val_dic))
lr_scheduler.step()
layer_one_optimizer_lr_scheduler.step()
save_checkpoint(i,
net,
optimizer,
lr_scheduler,
file_name=os.path.join(config.model_dir,
'epoch-{}.checkpoint'.format(i)))
acc, advacc = eval_one_epoch(net, ds_val, device, eval_attack)
tb_val_dic = {'Acc': acc, 'AdvAcc': advacc}
print('Eval: {}'.format(tb_val_dic))