print('Min FID:', np.min(history['FID-c']))
print('Epoch with min FID:', np.argmin(history['FID-c']))
if not args.no_plots:
plot_learningcurves(history, 'gen_loss')
plot_learningcurves(history, 'disc_loss')
plot_learningcurves(history, 'gen_loss_minibatch')
plot_learningcurves(history, 'disc_loss_minibatch')
plot_learningcurves(history, 'FID-c')
test_model(model=model,
n_tests=args.n_tests,
cuda_mode=args.cuda,
SNGAN=True)
save_samples(prefix='CIFAR10_SNGAN',
generator=model,
cp_name=args.cp_path.split('/')[-1].split('.')[0],
cuda_mode=args.cuda,
im_size=32,
SNGAN=True)
if args.inception:
print(
inception_score(model,
batch_size=4,
N=args.n_inception,
cuda=args.cuda,
resize=True,
splits=10,
SNGAN=True))
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')
plot_learningcurves(history, 'disc_loss_minibatch')
test_model(model=model,
n_tests=args.n_tests,
cuda_mode=args.cuda,
SNGAN=True)
save_samples(prefix='cats',
generator=model,
cp_name=args.cp_path.split('/')[-1].split('.')[0],
cuda_mode=args.cuda,
enhance=False,
im_size=256,
fig_size=(4, 8),
SNGAN=True)
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:
plot_learningcurves(history, 'gen_loss')
plot_learningcurves(history, 'disc_loss')
plot_learningcurves(history, 'gen_loss_minibatch')
plot_learningcurves(history, 'disc_loss_minibatch')
plot_learningcurves(history, 'FID-c')
test_model(model=model, n_tests=args.n_tests, cuda_mode=args.cuda)
save_samples(prefix='CIFAR10_DCGAN',
generator=model,
cp_name=args.cp_path.split('/')[-1].split('.')[0],
cuda_mode=args.cuda)
if args.inception:
print(
inception_score(model,
N=args.n_inception,
cuda=args.cuda,
resize=True,
splits=10))
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']
if not args.no_plots:
plot_learningcurves(history, 'gen_loss')
plot_learningcurves(history, 'disc_loss')
plot_learningcurves(history, 'gen_loss_minibatch')
plot_learningcurves(history, 'disc_loss_minibatch')
test_model(model=model, n_tests=args.n_tests, cuda_mode=args.cuda)
save_samples(prefix='CELEBA',
generator=model,
cp_name=args.cp_path.split('/')[-1].split('.')[0],
cuda_mode=args.cuda,
fig_size=(8, 8))
raise ValueError(
'There is no checkpoint/model path. Use arg --cp-path to indicate the path!'
)
model = Generator(100, [2048, 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']
if not args.no_plots:
plot_learningcurves(history, 'gen_loss')
plot_learningcurves(history, 'disc_loss')
plot_learningcurves(history, 'gen_loss_minibatch')
plot_learningcurves(history, 'disc_loss_minibatch')
test_model(model=model, n_tests=args.n_tests, cuda_mode=args.cuda)
save_samples(prefix='CELEBA',
generator=model,
cp_name=args.cp_path.split('/')[-1].split('.')[0],
cuda_mode=args.cuda,
enhance=False,
im_size=128,
fig_size=(4, 8))
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:
plot_learningcurves(history, 'gen_loss')
plot_learningcurves(history, 'disc_loss')
plot_learningcurves(history, 'gen_loss_minibatch')
plot_learningcurves(history, 'disc_loss_minibatch')
plot_learningcurves(history, 'FID-c')
#plot_learningcurves(history, 'steepest_dir_norm')
test_model(model=model, n_tests=args.n_tests, cuda_mode=args.cuda)
save_samples(prefix='CIFAR10_LSGAN_proj', generator=model, cp_name=args.cp_path.split('/')[-1].split('.')[0], cuda_mode=args.cuda, fig_size=(10, 10))
if args.inception:
print(inception_score(model, N=args.n_inception, cuda=args.cuda, resize=True, splits=10))
# Testing settings
parser = argparse.ArgumentParser(
description='Testing GANs under max hyper volume training')
parser.add_argument('--models-path',
type=str,
default=None,
metavar='Path',
help='Checkpoint/model path')
args = parser.parse_args()
if args.models_path is None:
raise ValueError(
'There is no checkpoint/model path. Use arg --models-path to indicate the path!'
)
model = Generator(100, [1024, 512, 256, 128], 3)
files_list = glob.glob(args.models_path + 'G_*.pt')
files_list.sort()
for file_ in files_list:
ckpt = torch.load(file_, map_location=lambda storage, loc: storage)
model.load_state_dict(ckpt['model_state'])
save_samples(prefix='CELEBA_hGAN_VaryingD',
generator=model,
cp_name=file_.split('/')[-1].split('.')[0],
cuda_mode=False,
fig_size=(2, 14))