else:
print('initializing generator weights')
netG.initialize_weights()
#netG = torch.nn.Sequential(
# Normalize(div2kmean, div2kstd),
# netG,
# DeNormalize(div2kmean, div2kstd),
# )
print('# generator parameters: ', sum(param.numel() for param in netG.parameters()))
netD = Discriminator()
if opt.discriminator_pretrain_path is not None:
print('loading pretrained model at ' + opt.discriminator_pretrain_path)
netD.load_state_dict(torch.load(opt.discriminator_pretrain_path))
else:
print('initializing discriminator weights')
netD.initialize_weights()
print('# discriminator parameters: ', sum(param.numel() for param in netD.parameters()))
generator_criterion = GeneratorLoss()
discriminator_criterion = nn.BCEWithLogitsLoss()
interpolation_layer = nn.functional.interpolate
if torch.cuda.is_available():
netG.cuda()
netD.cuda()
generator_criterion.cuda()
discriminator_criterion.cuda()
optimizerG = optim.Adam(netG.parameters(), lr=opt.lr_G)
if opt.generator_optim_pretrain_path is not None:
print('loading pretrained model at ' + opt.generator_optim_pretrain_path)
