Code examples / Generative Deep Learning / WGAN-GP overriding `Model.train_step`

WGAN-GP overriding Model.train_step

Author: A_K_Nain
Date created: 2020/05/9
Last modified: 2023/08/3
Description: Implementation of Wasserstein GAN with Gradient Penalty.

ⓘ This example uses Keras 3

View in Colab GitHub source


Wasserstein GAN (WGAN) with Gradient Penalty (GP)

The original Wasserstein GAN leverages the Wasserstein distance to produce a value function that has better theoretical properties than the value function used in the original GAN paper. WGAN requires that the discriminator (aka the critic) lie within the space of 1-Lipschitz functions. The authors proposed the idea of weight clipping to achieve this constraint. Though weight clipping works, it can be a problematic way to enforce 1-Lipschitz constraint and can cause undesirable behavior, e.g. a very deep WGAN discriminator (critic) often fails to converge.

The WGAN-GP method proposes an alternative to weight clipping to ensure smooth training. Instead of clipping the weights, the authors proposed a "gradient penalty" by adding a loss term that keeps the L2 norm of the discriminator gradients close to 1.


Setup

import os

os.environ["KERAS_BACKEND"] = "tensorflow"

import keras
import tensorflow as tf
from keras import layers

Prepare the Fashion-MNIST data

To demonstrate how to train WGAN-GP, we will be using the Fashion-MNIST dataset. Each sample in this dataset is a 28x28 grayscale image associated with a label from 10 classes (e.g. trouser, pullover, sneaker, etc.)

IMG_SHAPE = (28, 28, 1)
BATCH_SIZE = 512

# Size of the noise vector
noise_dim = 128

fashion_mnist = keras.datasets.fashion_mnist
(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()
print(f"Number of examples: {len(train_images)}")
print(f"Shape of the images in the dataset: {train_images.shape[1:]}")

# Reshape each sample to (28, 28, 1) and normalize the pixel values in the [-1, 1] range
train_images = train_images.reshape(train_images.shape[0], *IMG_SHAPE).astype("float32")
train_images = (train_images - 127.5) / 127.5
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-labels-idx1-ubyte.gz
 29515/29515 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-images-idx3-ubyte.gz
 26421880/26421880 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-labels-idx1-ubyte.gz
 5148/5148 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-images-idx3-ubyte.gz
 4422102/4422102 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
Number of examples: 60000
Shape of the images in the dataset: (28, 28)

Create the discriminator (the critic in the original WGAN)

The samples in the dataset have a (28, 28, 1) shape. Because we will be using strided convolutions, this can result in a shape with odd dimensions. For example, (28, 28) -> Conv_s2 -> (14, 14) -> Conv_s2 -> (7, 7) -> Conv_s2 ->(3, 3).

While performing upsampling in the generator part of the network, we won't get the same input shape as the original images if we aren't careful. To avoid this, we will do something much simpler: - In the discriminator: "zero pad" the input to change the shape to (32, 32, 1) for each sample; and - Ihe generator: crop the final output to match the shape with input shape.

def conv_block(
    x,
    filters,
    activation,
    kernel_size=(3, 3),
    strides=(1, 1),
    padding="same",
    use_bias=True,
    use_bn=False,
    use_dropout=False,
    drop_value=0.5,
):
    x = layers.Conv2D(
        filters, kernel_size, strides=strides, padding=padding, use_bias=use_bias
    )(x)
    if use_bn:
        x = layers.BatchNormalization()(x)
    x = activation(x)
    if use_dropout:
        x = layers.Dropout(drop_value)(x)
    return x


def get_discriminator_model():
    img_input = layers.Input(shape=IMG_SHAPE)
    # Zero pad the input to make the input images size to (32, 32, 1).
    x = layers.ZeroPadding2D((2, 2))(img_input)
    x = conv_block(
        x,
        64,
        kernel_size=(5, 5),
        strides=(2, 2),
        use_bn=False,
        use_bias=True,
        activation=layers.LeakyReLU(0.2),
        use_dropout=False,
        drop_value=0.3,
    )
    x = conv_block(
        x,
        128,
        kernel_size=(5, 5),
        strides=(2, 2),
        use_bn=False,
        activation=layers.LeakyReLU(0.2),
        use_bias=True,
        use_dropout=True,
        drop_value=0.3,
    )
    x = conv_block(
        x,
        256,
        kernel_size=(5, 5),
        strides=(2, 2),
        use_bn=False,
        activation=layers.LeakyReLU(0.2),
        use_bias=True,
        use_dropout=True,
        drop_value=0.3,
    )
    x = conv_block(
        x,
        512,
        kernel_size=(5, 5),
        strides=(2, 2),
        use_bn=False,
        activation=layers.LeakyReLU(0.2),
        use_bias=True,
        use_dropout=False,
        drop_value=0.3,
    )

    x = layers.Flatten()(x)
    x = layers.Dropout(0.2)(x)
    x = layers.Dense(1)(x)

    d_model = keras.models.Model(img_input, x, name="discriminator")
    return d_model


d_model = get_discriminator_model()
d_model.summary()
Model: "discriminator"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Layer (type)                     Output Shape                  Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ input_layer (InputLayer)        │ (None, 28, 28, 1)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ zero_padding2d (ZeroPadding2D)  │ (None, 32, 32, 1)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ conv2d (Conv2D)                 │ (None, 16, 16, 64)        │      1,664 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ leaky_re_lu (LeakyReLU)         │ (None, 16, 16, 64)        │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ conv2d_1 (Conv2D)               │ (None, 8, 8, 128)         │    204,928 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ leaky_re_lu_1 (LeakyReLU)       │ (None, 8, 8, 128)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ dropout (Dropout)               │ (None, 8, 8, 128)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ conv2d_2 (Conv2D)               │ (None, 4, 4, 256)         │    819,456 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ leaky_re_lu_2 (LeakyReLU)       │ (None, 4, 4, 256)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ dropout_1 (Dropout)             │ (None, 4, 4, 256)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ conv2d_3 (Conv2D)               │ (None, 2, 2, 512)         │  3,277,312 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ leaky_re_lu_3 (LeakyReLU)       │ (None, 2, 2, 512)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ flatten (Flatten)               │ (None, 2048)              │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ dropout_2 (Dropout)             │ (None, 2048)              │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ dense (Dense)                   │ (None, 1)                 │      2,049 │
└─────────────────────────────────┴───────────────────────────┴────────────┘
 Total params: 4,305,409 (16.42 MB)
 Trainable params: 4,305,409 (16.42 MB)
 Non-trainable params: 0 (0.00 B)

Create the generator

def upsample_block(
    x,
    filters,
    activation,
    kernel_size=(3, 3),
    strides=(1, 1),
    up_size=(2, 2),
    padding="same",
    use_bn=False,
    use_bias=True,
    use_dropout=False,
    drop_value=0.3,
):
    x = layers.UpSampling2D(up_size)(x)
    x = layers.Conv2D(
        filters, kernel_size, strides=strides, padding=padding, use_bias=use_bias
    )(x)

    if use_bn:
        x = layers.BatchNormalization()(x)

    if activation:
        x = activation(x)
    if use_dropout:
        x = layers.Dropout(drop_value)(x)
    return x


def get_generator_model():
    noise = layers.Input(shape=(noise_dim,))
    x = layers.Dense(4 * 4 * 256, use_bias=False)(noise)
    x = layers.BatchNormalization()(x)
    x = layers.LeakyReLU(0.2)(x)

    x = layers.Reshape((4, 4, 256))(x)
    x = upsample_block(
        x,
        128,
        layers.LeakyReLU(0.2),
        strides=(1, 1),
        use_bias=False,
        use_bn=True,
        padding="same",
        use_dropout=False,
    )
    x = upsample_block(
        x,
        64,
        layers.LeakyReLU(0.2),
        strides=(1, 1),
        use_bias=False,
        use_bn=True,
        padding="same",
        use_dropout=False,
    )
    x = upsample_block(
        x, 1, layers.Activation("tanh"), strides=(1, 1), use_bias=False, use_bn=True
    )
    # At this point, we have an output which has the same shape as the input, (32, 32, 1).
    # We will use a Cropping2D layer to make it (28, 28, 1).
    x = layers.Cropping2D((2, 2))(x)

    g_model = keras.models.Model(noise, x, name="generator")
    return g_model


g_model = get_generator_model()
g_model.summary()
Model: "generator"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Layer (type)                     Output Shape                  Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ input_layer_1 (InputLayer)      │ (None, 128)               │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ dense_1 (Dense)                 │ (None, 4096)              │    524,288 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ batch_normalization             │ (None, 4096)              │     16,384 │
│ (BatchNormalization)            │                           │            │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ leaky_re_lu_4 (LeakyReLU)       │ (None, 4096)              │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ reshape (Reshape)               │ (None, 4, 4, 256)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ up_sampling2d (UpSampling2D)    │ (None, 8, 8, 256)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ conv2d_4 (Conv2D)               │ (None, 8, 8, 128)         │    294,912 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ batch_normalization_1           │ (None, 8, 8, 128)         │        512 │
│ (BatchNormalization)            │                           │            │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ leaky_re_lu_5 (LeakyReLU)       │ (None, 8, 8, 128)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ up_sampling2d_1 (UpSampling2D)  │ (None, 16, 16, 128)       │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ conv2d_5 (Conv2D)               │ (None, 16, 16, 64)        │     73,728 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ batch_normalization_2           │ (None, 16, 16, 64)        │        256 │
│ (BatchNormalization)            │                           │            │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ leaky_re_lu_6 (LeakyReLU)       │ (None, 16, 16, 64)        │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ up_sampling2d_2 (UpSampling2D)  │ (None, 32, 32, 64)        │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ conv2d_6 (Conv2D)               │ (None, 32, 32, 1)         │        576 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ batch_normalization_3           │ (None, 32, 32, 1)         │          4 │
│ (BatchNormalization)            │                           │            │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ activation (Activation)         │ (None, 32, 32, 1)         │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ cropping2d (Cropping2D)         │ (None, 28, 28, 1)         │          0 │
└─────────────────────────────────┴───────────────────────────┴────────────┘
 Total params: 910,660 (3.47 MB)
 Trainable params: 902,082 (3.44 MB)
 Non-trainable params: 8,578 (33.51 KB)

Create the WGAN-GP model

Now that we have defined our generator and discriminator, it's time to implement the WGAN-GP model. We will also override the train_step for training.

class WGAN(keras.Model):
    def __init__(
        self,
        discriminator,
        generator,
        latent_dim,
        discriminator_extra_steps=3,
        gp_weight=10.0,
    ):
        super().__init__()
        self.discriminator = discriminator
        self.generator = generator
        self.latent_dim = latent_dim
        self.d_steps = discriminator_extra_steps
        self.gp_weight = gp_weight

    def compile(self, d_optimizer, g_optimizer, d_loss_fn, g_loss_fn):
        super().compile()
        self.d_optimizer = d_optimizer
        self.g_optimizer = g_optimizer
        self.d_loss_fn = d_loss_fn
        self.g_loss_fn = g_loss_fn

    def gradient_penalty(self, batch_size, real_images, fake_images):
        """Calculates the gradient penalty.

        This loss is calculated on an interpolated image
        and added to the discriminator loss.
        """
        # Get the interpolated image
        alpha = tf.random.uniform([batch_size, 1, 1, 1], 0.0, 1.0)
        diff = fake_images - real_images
        interpolated = real_images + alpha * diff

        with tf.GradientTape() as gp_tape:
            gp_tape.watch(interpolated)
            # 1. Get the discriminator output for this interpolated image.
            pred = self.discriminator(interpolated, training=True)

        # 2. Calculate the gradients w.r.t to this interpolated image.
        grads = gp_tape.gradient(pred, [interpolated])[0]
        # 3. Calculate the norm of the gradients.
        norm = tf.sqrt(tf.reduce_sum(tf.square(grads), axis=[1, 2, 3]))
        gp = tf.reduce_mean((norm - 1.0) ** 2)
        return gp

    def train_step(self, real_images):
        if isinstance(real_images, tuple):
            real_images = real_images[0]

        # Get the batch size
        batch_size = tf.shape(real_images)[0]

        # For each batch, we are going to perform the
        # following steps as laid out in the original paper:
        # 1. Train the generator and get the generator loss
        # 2. Train the discriminator and get the discriminator loss
        # 3. Calculate the gradient penalty
        # 4. Multiply this gradient penalty with a constant weight factor
        # 5. Add the gradient penalty to the discriminator loss
        # 6. Return the generator and discriminator losses as a loss dictionary

        # Train the discriminator first. The original paper recommends training
        # the discriminator for `x` more steps (typically 5) as compared to
        # one step of the generator. Here we will train it for 3 extra steps
        # as compared to 5 to reduce the training time.
        for i in range(self.d_steps):
            # Get the latent vector
            random_latent_vectors = tf.random.normal(
                shape=(batch_size, self.latent_dim)
            )
            with tf.GradientTape() as tape:
                # Generate fake images from the latent vector
                fake_images = self.generator(random_latent_vectors, training=True)
                # Get the logits for the fake images
                fake_logits = self.discriminator(fake_images, training=True)
                # Get the logits for the real images
                real_logits = self.discriminator(real_images, training=True)

                # Calculate the discriminator loss using the fake and real image logits
                d_cost = self.d_loss_fn(real_img=real_logits, fake_img=fake_logits)
                # Calculate the gradient penalty
                gp = self.gradient_penalty(batch_size, real_images, fake_images)
                # Add the gradient penalty to the original discriminator loss
                d_loss = d_cost + gp * self.gp_weight

            # Get the gradients w.r.t the discriminator loss
            d_gradient = tape.gradient(d_loss, self.discriminator.trainable_variables)
            # Update the weights of the discriminator using the discriminator optimizer
            self.d_optimizer.apply_gradients(
                zip(d_gradient, self.discriminator.trainable_variables)
            )

        # Train the generator
        # Get the latent vector
        random_latent_vectors = tf.random.normal(shape=(batch_size, self.latent_dim))
        with tf.GradientTape() as tape:
            # Generate fake images using the generator
            generated_images = self.generator(random_latent_vectors, training=True)
            # Get the discriminator logits for fake images
            gen_img_logits = self.discriminator(generated_images, training=True)
            # Calculate the generator loss
            g_loss = self.g_loss_fn(gen_img_logits)

        # Get the gradients w.r.t the generator loss
        gen_gradient = tape.gradient(g_loss, self.generator.trainable_variables)
        # Update the weights of the generator using the generator optimizer
        self.g_optimizer.apply_gradients(
            zip(gen_gradient, self.generator.trainable_variables)
        )
        return {"d_loss": d_loss, "g_loss": g_loss}

Create a Keras callback that periodically saves generated images

class GANMonitor(keras.callbacks.Callback):
    def __init__(self, num_img=6, latent_dim=128):
        self.num_img = num_img
        self.latent_dim = latent_dim

    def on_epoch_end(self, epoch, logs=None):
        random_latent_vectors = tf.random.normal(shape=(self.num_img, self.latent_dim))
        generated_images = self.model.generator(random_latent_vectors)
        generated_images = (generated_images * 127.5) + 127.5

        for i in range(self.num_img):
            img = generated_images[i].numpy()
            img = keras.utils.array_to_img(img)
            img.save("generated_img_{i}_{epoch}.png".format(i=i, epoch=epoch))

Train the end-to-end model

# Instantiate the optimizer for both networks
# (learning_rate=0.0002, beta_1=0.5 are recommended)
generator_optimizer = keras.optimizers.Adam(
    learning_rate=0.0002, beta_1=0.5, beta_2=0.9
)
discriminator_optimizer = keras.optimizers.Adam(
    learning_rate=0.0002, beta_1=0.5, beta_2=0.9
)


# Define the loss functions for the discriminator,
# which should be (fake_loss - real_loss).
# We will add the gradient penalty later to this loss function.
def discriminator_loss(real_img, fake_img):
    real_loss = tf.reduce_mean(real_img)
    fake_loss = tf.reduce_mean(fake_img)
    return fake_loss - real_loss


# Define the loss functions for the generator.
def generator_loss(fake_img):
    return -tf.reduce_mean(fake_img)


# Set the number of epochs for training.
epochs = 20

# Instantiate the customer `GANMonitor` Keras callback.
cbk = GANMonitor(num_img=3, latent_dim=noise_dim)

# Get the wgan model
wgan = WGAN(
    discriminator=d_model,
    generator=g_model,
    latent_dim=noise_dim,
    discriminator_extra_steps=3,
)

# Compile the wgan model
wgan.compile(
    d_optimizer=discriminator_optimizer,
    g_optimizer=generator_optimizer,
    g_loss_fn=generator_loss,
    d_loss_fn=discriminator_loss,
)

# Start training
wgan.fit(train_images, batch_size=BATCH_SIZE, epochs=epochs, callbacks=[cbk])
Epoch 1/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 79s 345ms/step - d_loss: -7.7597 - g_loss: -17.2858 - loss: 0.0000e+00
Epoch 2/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 118ms/step - d_loss: -7.0841 - g_loss: -13.8542 - loss: 0.0000e+00
Epoch 3/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 118ms/step - d_loss: -6.1011 - g_loss: -13.2763 - loss: 0.0000e+00
Epoch 4/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 119ms/step - d_loss: -5.5292 - g_loss: -13.3122 - loss: 0.0000e+00
Epoch 5/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 119ms/step - d_loss: -5.1012 - g_loss: -12.1395 - loss: 0.0000e+00
Epoch 6/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 119ms/step - d_loss: -4.7557 - g_loss: -11.2559 - loss: 0.0000e+00
Epoch 7/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 119ms/step - d_loss: -4.4727 - g_loss: -10.3075 - loss: 0.0000e+00
Epoch 8/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 119ms/step - d_loss: -4.2056 - g_loss: -10.0340 - loss: 0.0000e+00
Epoch 9/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -4.0116 - g_loss: -9.9283 - loss: 0.0000e+00
Epoch 10/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -3.8050 - g_loss: -9.7392 - loss: 0.0000e+00
Epoch 11/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -3.6608 - g_loss: -9.4686 - loss: 0.0000e+00
Epoch 12/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 121ms/step - d_loss: -3.4623 - g_loss: -8.9601 - loss: 0.0000e+00
Epoch 13/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -3.3659 - g_loss: -8.4620 - loss: 0.0000e+00
Epoch 14/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -3.2486 - g_loss: -7.9598 - loss: 0.0000e+00
Epoch 15/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -3.1436 - g_loss: -7.5392 - loss: 0.0000e+00
Epoch 16/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -3.0370 - g_loss: -7.3694 - loss: 0.0000e+00
Epoch 17/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -2.9256 - g_loss: -7.6105 - loss: 0.0000e+00
Epoch 18/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -2.8976 - g_loss: -6.5240 - loss: 0.0000e+00
Epoch 19/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -2.7944 - g_loss: -6.6281 - loss: 0.0000e+00
Epoch 20/20
 118/118 ━━━━━━━━━━━━━━━━━━━━ 14s 120ms/step - d_loss: -2.7175 - g_loss: -6.5900 - loss: 0.0000e+00

<keras.src.callbacks.history.History at 0x7fc763a8e950>

Display the last generated images:

from IPython.display import Image, display

display(Image("generated_img_0_19.png"))
display(Image("generated_img_1_19.png"))
display(Image("generated_img_2_19.png"))

png

png

png