m, C = statistics['m'], statistics['C']
if args.cp_path is None:
raise ValueError(
'There is no checkpoint/model path. Use arg --cp-path to indicate the path!'
)
if args.sub_key is None:
raise ValueError(
'There is no key to substitute. Use arg --sub-key to indicate the key!'
)
print(args.sub_key, args.cp_path)
generator = Generator(100, [1024, 512, 256, 128], 3).eval()
gen_state = torch.load(args.cp_path,
map_location=lambda storage, loc: storage)
generator.load_state_dict(gen_state['model_state'])
if args.cuda:
generator = generator.cuda()
fid = []
for i in range(args.ntests):
fid.append(
compute_fid(
generator,
fid_model,
args.batch_size,
mod_state = torch.load(args.fid_model_path,
map_location=lambda storage, loc: storage)
fid_model.load_state_dict(mod_state['model_state'])
if not os.path.isfile('../test_data_statistics.p'):
testset = datasets.CIFAR10(root=args.data_path,
train=False,
download=True,
transform=transform)
test_loader = torch.utils.data.DataLoader(testset,
batch_size=1000,
shuffle=False,
num_workers=args.workers)
save_testdata_statistics(fid_model, test_loader, cuda_mode=args.cuda)
generator = Generator(100, [1024, 512, 256, 128], 3).train()
disc_list = []
for i in range(args.ndiscriminators):
disc = Discriminator_wgan(3, [128, 256, 512, 1024], 1, optim.Adam, args.lr,
(args.beta1, args.beta2)).train()
disc_list.append(disc)
if args.cuda:
generator = generator.cuda()
for disc in disc_list:
disc = disc.cuda()
torch.backends.cudnn.benchmark = True
optimizer = optim.Adam(generator.parameters(),
lr=args.lr,
betas=(args.beta1, args.beta2))
default=False,
help='Disables plot of train/test losses')
parser.add_argument('--no-cuda',
action='store_true',
default=False,
help='Disables GPU use')
args = parser.parse_args()
args.cuda = True if not args.no_cuda and torch.cuda.is_available(
) else False
if args.cp_path is None:
raise ValueError(
'There is no checkpoint/model path. Use arg --cp-path to indicate the path!'
)
model = Generator(128, [1024, 512, 256, 128, 64, 32], 3)
ckpt = torch.load(args.cp_path, map_location=lambda storage, loc: storage)
model.load_state_dict(ckpt['model_state'])
if args.cuda:
model = model.cuda()
print('Cuda Mode is: {}'.format(args.cuda))
history = ckpt['history']
if not args.no_plots:
plot_learningcurves(history, 'gen_loss')
plot_learningcurves(history, 'disc_loss')
plot_learningcurves(history, 'gen_loss_minibatch')
parser.add_argument('--alpha', type=float, default=0.8, metavar='alhpa', help='Used in GMAN and loss_del modes (default: 0.8)')
parser.add_argument('--no-cuda', action='store_true', default=False, help='Disables GPU use')
args = parser.parse_args()
args.cuda = True if not args.no_cuda and torch.cuda.is_available() else False
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
transform = transforms.Compose([transforms.Resize((256, 256), interpolation=Image.BICUBIC), transforms.RandomHorizontalFlip(), transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
cats_data = datasets.ImageFolder(args.data_path, transform=transform)
train_loader = torch.utils.data.DataLoader(cats_data, batch_size=args.batch_size, shuffle=True, num_workers=args.workers)
generator = Generator(128, [1024, 512, 256, 128, 64, 32], 3).train()
if args.disc_mode == 'RP':
disc_list = []
for i in range(args.ndiscriminators):
disc = Discriminator(3, [32, 64, 128, 256, 512, 1024], 1, optim.Adam, args.lr, (args.beta1, args.beta2)).train()
disc_list.append(disc)
elif args.disc_mode == 'MD':
D1 = Discriminator_vanilla(ndf=64, nc=3, optimizer=optim.Adam, lr=args.lr, betas=(args.beta1, args.beta2)).train()
D2 = Discriminator_f6(ndf=64, nc=3, optimizer=optim.Adam, lr=args.lr, betas=(args.beta1, args.beta2)).train()
D3 = Discriminator_f8(ndf=32, nc=3, optimizer=optim.Adam, lr=args.lr, betas=(args.beta1, args.beta2)).train()
D4 = Discriminator_f4s3(ndf=64, nc=3, optimizer=optim.Adam, lr=args.lr, betas=(args.beta1, args.beta2)).train()
D5 = Discriminator_dense(ndf=64, nc=3, optimizer=optim.Adam, lr=args.lr, betas=(args.beta1, args.beta2)).train()
D6 = Discriminator_f16(ndf=16, nc=3, optimizer=optim.Adam, lr=args.lr, betas=(args.beta1, args.beta2)).train()
parser.add_argument(
'--n-inception',
type=int,
default=1024,
metavar='N',
help='number of samples to calculate inception score (default: 1024)')
args = parser.parse_args()
args.cuda = True if not args.no_cuda and torch.cuda.is_available(
) else False
if args.cp_path is None:
raise ValueError(
'There is no checkpoint/model path. Use arg --cp-path to indicate the path!'
)
model = Generator(100, [1024, 512, 256, 128], 3)
ckpt = torch.load(args.cp_path, map_location=lambda storage, loc: storage)
model.load_state_dict(ckpt['model_state'])
if args.cuda:
model = model.cuda()
print('Cuda Mode is: {}'.format(args.cuda))
history = ckpt['history']
print('Min FID:', np.min(history['FID-c']))
print('Epoch with min FID:', np.argmin(history['FID-c']))
if not args.no_plots: