for seed in range(10):
args.seed = seed
# Model name
model_name = args.outpath + 'test_domain_' + str(args.list_test_domain[0]) + '_vae_seed_' + str(
args.seed)
print(model_name)
# Set seed
torch.manual_seed(args.seed)
torch.backends.cudnn.benchmark = False
np.random.seed(args.seed)
# Load supervised training
train_loader = data_utils.DataLoader(
MnistRotated(args.list_train_domains, args.list_test_domain, args.num_supervised, args.seed, './../dataset/',
train=True),
batch_size=args.batch_size,
shuffle=True, **kwargs)
# Load test
test_loader = data_utils.DataLoader(
MnistRotated(args.list_train_domains, args.list_test_domain, args.num_supervised, args.seed, './../dataset/',
train=False),
batch_size=args.batch_size,
shuffle=True, **kwargs)
# setup the VAE
model = VAE(args).to(device)
# setup the optimizer
optimizer = optim.Adam(model.parameters(), lr=args.lr)
def main():
# Training settings
parser = argparse.ArgumentParser(description='PyTorch MNIST Example')
parser.add_argument('--batch-size', type=int, default=100, metavar='N',
help='input batch size for training (default: 64)')
parser.add_argument('--test-batch-size', type=int, default=1000, metavar='N',
help='input batch size for testing (default: 1000)')
parser.add_argument('--epochs', type=int, default=500, metavar='N',
help='number of epochs to train (default: 10)')
parser.add_argument('--lr', type=float, default=0.001, metavar='LR',
help='learning rate (default: 0.01)')
parser.add_argument('--lamb', type=float, default=1.0, metavar='L',
help='weight for domain loss')
parser.add_argument('--no-cuda', action='store_true', default=False,
help='disables CUDA training')
parser.add_argument('--seed', type=int, default=0, metavar='S',
help='random seed (default: 1)')
parser.add_argument('--log-interval', type=int, default=10, metavar='N',
help='how many batches to wait before logging training status')
parser.add_argument('--list_train_domains', type=list, default=['0', '15', '45', '60'],
help='domains used during training')
parser.add_argument('--list_test_domain', type=str, default='75',
help='domain used during testing')
parser.add_argument('--num-supervised', default=1000, type=int,
help="number of supervised examples, /10 = samples per class")
args = parser.parse_args()
use_cuda = not args.no_cuda and torch.cuda.is_available()
device = torch.device("cuda" if use_cuda else "cpu")
kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {}
torch.manual_seed(args.seed)
torch.backends.cudnn.benchmark = False
np.random.seed(args.seed)
model = CNNModel().to(device)
optimizer = optim.Adam(model.parameters(), lr=args.lr)
args.list_test_domain = [args.list_test_domain]
# Choose training domains
all_training_domains = ['0', '15', '30', '45', '75']
all_training_domains.remove(args.list_test_domain[0])
args.list_train_domains = all_training_domains
print(args.list_test_domain, args.list_train_domains)
# Load supervised training
train_loader = data_utils.DataLoader(
MnistRotated(args.list_train_domains, args.list_test_domain, args.num_supervised, args.seed, './../../dataset/',
train=True),
batch_size=args.batch_size,
shuffle=True, **kwargs)
print(len(train_loader.dataset))
# Load test
test_loader = data_utils.DataLoader(
MnistRotated(args.list_train_domains, args.list_test_domain, args.num_supervised, args.seed, './../../dataset/',
train=False),
batch_size=args.batch_size,
shuffle=True, **kwargs)
train(args, model, device, train_loader, test_loader, optimizer)
test_accuracy_y_list = []
for i in range(10):
model_name = 'test_domain_75_sup_only_seed_' + str(i) + '_add_30'
model = torch.load(model_name + '.model')
args = torch.load(model_name + '.config')
args.cuda = not args.no_cuda and torch.cuda.is_available()
device = torch.device("cuda" if args.cuda else "cpu")
kwargs = {'num_workers': 2, 'pin_memory': True} if args.cuda else {}
# Load test
test_loader_sup = data_utils.DataLoader(MnistRotated(
args.list_train_domains,
args.list_test_domain,
args.num_supervised,
args.seed,
'./../../../../dataset/',
train=False),
batch_size=args.batch_size,
shuffle=True)
# Set seed
torch.manual_seed(args.seed)
torch.backends.cudnn.benchmark = False
np.random.seed(args.seed)
test_accuracy_d, test_accuracy_y = get_accuracy(
test_loader_sup, model.classifier, args.batch_size)
test_accuracy_y_list.append(test_accuracy_y)
print(args.list_test_domain, args.list_train_domains)
# Set seed
torch.manual_seed(args.seed)
torch.backends.cudnn.benchmark = False
np.random.seed(args.seed)
# Empty data loader dict
data_loaders = {}
# Load supervised training
train_loader_sup = data_utils.DataLoader(MnistRotated(
args.list_train_domains,
args.list_test_domain,
args.num_supervised,
args.seed,
'./../dataset/',
train=True),
batch_size=args.batch_size,
shuffle=True,
**kwargs)
if args.seed < 9:
additional_data_index = args.seed + 1
elif args.seed == 9:
additional_data_index = 0
# Load unsupervised training
train_loader_unsup = data_utils.DataLoader(MnistRotated(
args.list_train_domains,