51f2246f1a149a05447229a6f57d30580a6dc0d6,examples/d-optimizer.py,,create_random_discriminator,#,90

Before Change

    return gan

def create_random_discriminator():
    return pyramid_discriminator.config(layers=[5,6,7])

def run_gan(gan, steps):
    d_class_loss = 0
    gan.initialize_graph()

After Change

    return gan

def create_random_discriminator():
    return pyramid_discriminator.config(
            activation=[tf.nn.relu, lrelu, tf.nn.relu6, tf.nn.elu],
            depth_increase=[1.5,1.7,2,2.1],
            final_activation=[tf.nn.relu, tf.tanh, None],
            layer_regularizer=[batch_norm_1, layer_norm_1, None],
            layers=[4,5,6,7],
            fc_layer_size=[1024,512,2048],
            fc_layers=[0,1,2],
            noise=[False, 1e-2],
            progressive_enhancement=[True, False],
            strided=[True, False]
        )

def run_gan(gan, steps):
    d_class_loss = 0
    gan.initialize_graph()

Instances