def load_VAE():
model = VAE()
model.summary()
#model.load_weights('saved_model/VAE2_epochs_weights.h5')
model.load_weights('saved_model/VAE_weights.h5')
return model
history = model.fit(
x_train,
x_train,
initial_epoch=0,
epochs=epochs,
batch_size=batch_size,
callbacks=callbacks,
)
model.save_weights("model_weights_" + str(0) + ".h5")
return history
if __name__ == '__main__':
x_train, y_train, x_test, y_test = prepare_dataset()
model = VAE(img_shape, latent_dim)
model.built = True
weight_path = "model_weights.h5"
if os.path.exists(weight_path):
model.load_weights(weight_path)
optimizer = Adam()
history = train_model(
model,
x_train,
optimizer,
EPOCHS,
batch_size=BATCH_SIZE,
)
img_shape = (28, 28, 1)
EPOCHS = 2
BATCH_SIZE = 64
# load the mnist Dataset
x_train, y_train, x_test, y_test = prepare_dataset()
# load model
model = VAE(img_shape, latent_dim)
optimizer = Adam()
# Build the model
model.built = True
# load model weights
if os.path.exists("model_weights.h5"):
model.load_weights("model_weights.h5")
else:
train_model(
model,
x_train,
optimizer,
EPOCHS,
batch_size=BATCH_SIZE,
)
model.load_weights("model_weights_0.h5")
'''Visualize the Latent Space Distribution'''
def plot_clusters(data, labels):
# visualising the latent space
mean, _, _ = model.encoder.predict(data)
