link_data_with_net.py

# Create tf dataset
with tf.name_scope('DataPipe'):
    filenames = tf.placeholder_with_default(get_files(FLAGS.data_path), shape=[None], name='filenames_tensor')
    dataset = load_and_process_data(filenames, batch_size=FLAGS.batch_size, shuffle=True)
    iterator = dataset.make_initializable_iterator()
    input_batch = iterator.get_next()

# Create model
vae = VAE(input_batch, FLAGS.latent_dim, FLAGS.learning_rate, )