3 examples of 'keras plot model' in Python

46
def save_model_graph(self):
47
    plot_model(self.Discriminator, to_file='/data/Discriminator_Model.png')

25
def visualize_model():
26
    model = load_model('shallowCNN_model.h5') #See 'How to export keras models?' to generate this file before loading it.
27
    # plot_model(model, to_file='shallowCNN_model.png')
28
    model.summary()

131
def train(self, x_data):
132
    x_data, x_data = super().train(x_data, x_data)
133
134
    x_mu_std = np.vstack((np.median(x_data), np.std(x_data)))
135
    np.save(self.fullfilepath + 'meanstd_x.npy', x_mu_std)
136
    x_data -= x_mu_std[0]
137
    x_data /= x_mu_std[1]
138
139
    csv_logger = CSVLogger(self.fullfilepath + 'log.csv', append=True, separator=',')
140
141
    if self.task == 'classification':
142
        raise RuntimeError('astroNN VAE does not support classification task')
143
144
    reduce_lr = ReduceLROnPlateau(monitor='loss', factor=0.5, epsilon=self.reduce_lr_epsilon,
145
                                  patience=self.reduce_lr_patience, min_lr=self.reduce_lr_min, mode='min',
146
                                  verbose=2)
147
    self.compile()
148
    self.plot_model()
149
150
    training_generator = DataGenerator(x_data.shape[1], self.batch_size).generate(x_data)
151
152
    self.keras_model.fit_generator(generator=training_generator, steps_per_epoch=x_data.shape[0] // self.batch_size,
153
                                   epochs=self.max_epochs, max_queue_size=20, verbose=2, workers=os.cpu_count(),
154
                                   callbacks=[reduce_lr, csv_logger])
155
156
    astronn_model = 'model_weights.h5'
157
    self.keras_vae.save_weights(self.fullfilepath + astronn_model)
158
    print(astronn_model + ' saved to {}'.format(self.fullfilepath + astronn_model))
159
160
    K.clear_session()
161
162
    return None