def mnist(cuda=True, model_root=None):
print("Building and initializing mnist parameters")
from mnist import model, dataset
m = model.mnist(pretrained=os.path.join(model_root, 'mnist.pth'))
if cuda:
m = m.cuda()
return m, dataset.get, False
def mnist(cuda=True, model_root=None):
print("Building and initializing mnist parameters")
from mnist import model, dataset
m = model.mnist(pretrained=os.path.join(model_root, 'mnist.pth'))
if cuda:
m = m.cuda()
return m, dataset.get, False
def mnist(cuda=True, model_root=None, **kwargs):
print("Building and initializing mnist parameters")
from mnist import model, dataset
use_model_zoo = False if model_root else True
m = model.mnist(pretrained=os.path.join(model_root, 'mnist_paper.pth'),
use_model_zoo=use_model_zoo,
**kwargs)
if cuda:
m = m.cuda()
return m, dataset.get, False
print_to_log('{}: {}'.format(k, v))
print_to_log("=======================")
# seed
is_cuda = torch.cuda.is_available()
print_to_log("is_cuda: {}".format(is_cuda))
torch.manual_seed(args.seed)
if is_cuda and args.to_cuda:
torch.cuda.manual_seed(args.seed)
# data loader
train_loader, test_loader = dataset.get(batch_size=args.batch_size, data_root=args.data_root, num_workers=1)
# model
model = model.mnist(input_dims=784, n_hiddens=[256, 256], n_class=10)
if is_cuda and args.to_cuda:
model.cuda()
# optimizer
optimizer = optim.SGD(model.parameters(), lr=args.lr, weight_decay=args.wd, momentum=0.9)
decreasing_lr = list(map(int, args.decreasing_lr.split(',')))
print_to_log('decreasing_lr: ' + str(decreasing_lr))
best_acc, old_file = 0, None
t_begin = time.time()
try:
# ready to go
for epoch in range(args.epochs):
model.train()
if epoch in decreasing_lr: