def __init__(self, z_dim, output_dim=28**2):
super(Generator, self).__init__()
self.z_dim = z_dim
self.fc1 = nn.Linear(z_dim, 500, bias=False)
self.bn1 = nn.BatchNorm1d(500, affine=False, eps=1e-6, momentum=0.5)
self.fc2 = nn.Linear(500, 500, bias=False)
self.bn2 = nn.BatchNorm1d(500, affine=False, eps=1e-6, momentum=0.5)
self.fc3 = LinearWeightNorm(500, output_dim, weight_scale=1)
self.bn1_b = Parameter(torch.zeros(500))
self.bn2_b = Parameter(torch.zeros(500))
nn.init.xavier_uniform_(self.fc1.weight)
nn.init.xavier_uniform_(self.fc2.weight)
def __init__(self, z_dim, output_dim=28**2):
super(Generator, self).__init__()
self.z_dim = z_dim
self.fc1 = Linear(z_dim, 500)
self.bn1 = tf.keras.layers.BatchNormalization(trainable=False,
epsilon=1e-6,
momentum=0.5)
self.fc2 = Linear(500, 500)
self.bn2 = tf.keras.layers.BatchNormalization(trainable=False,
epsilon=1e-6,
momentum=0.5)
self.fc3 = LinearWeightNorm(500, output_dim)
self.bn1_b = tf.Variable(tf.zeros(500))
self.bn2_b = tf.Variable(tf.zeros(500))
def __init__(self, input_dim=28**2, output_dim=10):
super(Discriminator, self).__init__()
self.input_dim = input_dim
self.layers = torch.nn.ModuleList([
LinearWeightNorm(input_dim, 1000),
LinearWeightNorm(1000, 500),
LinearWeightNorm(500, 250),
LinearWeightNorm(250, 250),
LinearWeightNorm(250, 250)
])
self.final = LinearWeightNorm(250, output_dim, weight_scale=1)
def __init__(self, input_dim=28**2, output_dim=10):
super(Discriminator, self).__init__()
self.input_dim = input_dim
self.layers_hidden = [
LinearWeightNorm(input_dim, 500),
LinearWeightNorm(500, 500),
LinearWeightNorm(500, 250),
LinearWeightNorm(250, 250),
LinearWeightNorm(250, 250)
]
self.final = LinearWeightNorm(250, output_dim)