def __init__(self, num_classes):
super(Discriminator, self).__init__()
self.net = nn.Sequential(
nn.Dropout(.2),
weight_norm(nn.Conv2d(3, 96, 3, stride=1, padding=1)),
nn.LeakyReLU(),
weight_norm(nn.Conv2d(96, 96, 3, stride=1, padding=1)),
nn.LeakyReLU(),
weight_norm(nn.Conv2d(96, 96, 3, stride=2, padding=1)),
nn.LeakyReLU(),
nn.Dropout(.5),
weight_norm(nn.Conv2d(96, 192, 3, stride=1, padding=1)),
nn.LeakyReLU(),
weight_norm(nn.Conv2d(192, 192, 3, stride=1, padding=1)),
nn.LeakyReLU(),
weight_norm(nn.Conv2d(192, 192, 3, stride=2, padding=1)),
nn.LeakyReLU(),
nn.Dropout(.5),
weight_norm(nn.Conv2d(192, 192, 3, stride=1, padding=0)),
nn.LeakyReLU(),
weight_norm(nn.Conv2d(192, 192, 1, stride=1, padding=0)),
nn.LeakyReLU(),
weight_norm(nn.Conv2d(192, 192, 1, stride=1, padding=0)),
nn.LeakyReLU(),
# nn.AvgPool2d(6,stride=1),
nn.AdaptiveAvgPool2d(1),
helpers.Flatten())
self.fc = weight_norm(nn.Linear(192, num_classes))
def __init__(self, latent_dim):
super(Generator, self).__init__()
self.net = nn.Sequential(helpers.Flatten(),
nn.Linear(latent_dim, 500), nn.Softplus(),
nn.BatchNorm1d(500), nn.Linear(500, 500),
nn.Softplus(), nn.BatchNorm1d(500),
nn.Linear(500, 32 * 32), nn.Tanh(),
helpers.Reshape((1, 32, 32)))
def __init__(self, num_classes):
super(Discriminator, self).__init__()
self.net = nn.Sequential(
nn.Conv2d(1, 64, 3, stride=1, padding=1), nn.LeakyReLU(),
nn.Conv2d(64, 96, 3, stride=2, padding=1), nn.LeakyReLU(),
nn.Dropout(.2), nn.Conv2d(96, 96, 3, stride=1, padding=1),
nn.LeakyReLU(), nn.Conv2d(96, 192, 3, stride=2, padding=1),
nn.LeakyReLU(), nn.Dropout(.2),
nn.Conv2d(192, 192, 3, stride=2, padding=1), nn.LeakyReLU(),
nn.Conv2d(192, 192, 1, stride=1, padding=0), nn.LeakyReLU(),
nn.Conv2d(192, 192, 1, stride=1, padding=0), nn.LeakyReLU(),
nn.MaxPool2d(4, stride=1), helpers.Flatten())
self.fc = nn.Linear(192, num_classes)
def __init__(self, num_classes):
super(Discriminator, self).__init__()
self.net = nn.Sequential(
helpers.Flatten(),
nn.Linear(32 * 32, 1000),
nn.Dropout(.1),
nn.ReLU(),
nn.Linear(1000, 500),
nn.Dropout(.1),
nn.ReLU(),
nn.Linear(500, 250),
nn.Dropout(.1),
nn.ReLU(),
nn.Linear(250, 250),
nn.Dropout(.1),
nn.ReLU(),
nn.Linear(250, 250),
nn.Dropout(.1),
nn.ReLU(),
nn.Linear(250, num_classes),
)