120 практических задач - стр. 14

assert model.output_shape == (None, 8, 8, 256) # Убедитесь, что выходная форма такая

model.add(layers.Conv2DTranspose(128, (5, 5), strides=(2, 2), padding='same', use_bias=False))

model.add(layers.BatchNormalization())

model.add(layers.LeakyReLU())

assert model.output_shape == (None, 16, 16, 128)

model.add(layers.Conv2DTranspose(64, (5, 5), strides=(2, 2), padding='same', use_bias=False))

model.add(layers.BatchNormalization())

model.add(layers.LeakyReLU())

assert model.output_shape == (None, 32, 32, 64)

model.add(layers.Conv2DTranspose(3, (5, 5), strides=(2, 2), padding='same', use_bias=False, activation='tanh'))

assert model.output_shape == (None, 64, 64, 3)

return model

# Шаг 4: Построение дискриминатора

def build_discriminator():

model = models.Sequential()

model.add(layers.Conv2D(64, (5, 5), strides=(2, 2), padding='same', input_shape=[64, 64, 3]))

model.add(layers.LeakyReLU())

model.add(layers.Dropout(0.3))

model.add(layers.Conv2D(128, (5, 5), strides=(2, 2), padding='same'))

model.add(layers.LeakyReLU())

model.add(layers.Dropout(0.3))

model.add(layers.Conv2D(256, (5, 5), strides=(2, 2), padding='same'))

model.add(layers.LeakyReLU())

model.add(layers.Dropout(0.3))

model.add(layers.Flatten())

model.add(layers.Dense(1, activation='sigmoid'))

return model

# Построение генератора и дискриминатора

generator = build_generator()

discriminator = build_discriminator()

# Определение функции потерь и оптимизаторов

cross_entropy = tf.keras.losses.BinaryCrossentropy(from_logits=True)

def discriminator_loss(real_output, fake_output):

real_loss = cross_entropy(tf.ones_like(real_output), real_output)

fake_loss = cross_entropy(tf.zeros_like(fake_output), fake_output)

total_loss = real_loss + fake_loss

return total_loss

def generator_loss(fake_output):

return cross_entropy(tf.ones_like(fake_output), fake_output)

generator_optimizer = tf.keras.optimizers.Adam(1e-4)

discriminator_optimizer = tf.keras.optimizers.Adam(1e-4)

# Шаг 5: Построение и компиляция GAN

@tf.function

def train_step(images):

noise = tf.random.normal([BATCH_SIZE, 100])

with tf.GradientTape() as gen_tape, tf.GradientTape() as disc_tape:

generated_images = generator(noise, training=True)

real_output = discriminator(images, training=True)

fake_output = discriminator(generated_images, training=True)

gen_loss = generator_loss(fake_output)

disc_loss = discriminator_loss(real_output, fake_output)

gradients_of_generator = gen_tape.gradient(gen_loss, generator.trainable_variables)

gradients_of_discriminator = disc_tape.gradient(disc_loss, discriminator.trainable_variables)

generator_optimizer.apply_gradients(zip(gradients_of_generator, generator.trainable_variables))

discriminator_optimizer.apply_gradients(zip(gradients_of_discriminator, discriminator.trainable_variables))

def train(dataset, epochs):

for epoch in range(epochs):

for image_batch in dataset:

train_step(image_batch)

print(f'Эпоха {epoch + 1} завершена')

# Генерация изображений в конце каждой эпохи

if (epoch + 1) % 10 == 0:

noise = tf.random.normal([16, 100])

generate_and_save_images(generator, epoch + 1, noise)

# Шаг 6: Обучение GAN

EPOCHS = 100

train(train_dataset, EPOCHS)

# Шаг 7: Генерация изображений

def generate_and_save_images(model, epoch, test_input):

predictions = model(test_input, training=False)