Quiero usarfeedable diseño de iterador en la API de Dataset de tensorflow, para que pueda cambiar a datos de validación después de algunos pasos de entrenamiento. Pero si cambié a los datos de validación, finalizará toda la sesión.

El siguiente código demuestra lo que quiero hacer:

import tensorflow as tf


graph = tf.Graph()
with graph.as_default():
    training_ds = tf.data.Dataset.range(32).batch(4)
    validation_ds = tf.data.Dataset.range(8).batch(4)

    handle = tf.placeholder(tf.string, shape=[])
    iterator = tf.data.Iterator.from_string_handle(
        handle, training_ds.output_types, training_ds.output_shapes)
    next_element = iterator.get_next()

    training_iterator = training_ds.make_initializable_iterator()
    validation_iterator = validation_ds.make_initializable_iterator()


with graph.as_default():

    with tf.train.MonitoredTrainingSession() as sess:
        training_handle = sess.run(training_iterator.string_handle())
        validation_handle = sess.run(validation_iterator.string_handle())
        sess.run(training_iterator.initializer)
        count_training = 0
        while not sess.should_stop():
            x = sess.run(next_element, feed_dict={handle: training_handle})
            count_training += 1
            print('{} [training] {}'.format(count_training, x.shape))
            # print(x)

            # we do periodic validation
            if count_training % 4 == 0:
                sess.run(validation_iterator.initializer)
                count_validation = 0
                while not sess.should_stop():
                    y = sess.run(next_element, feed_dict={handle: validation_handle})
                    count_validation += 1
                    print('  {} [validation] {}'.format(count_validation, y.shape))
                    # print(y)

Los datos de entrenamiento tienen 32 elementos, agrupados con 4, por lo que obtuvimos 8 lotes que validamos cada 4 pasos, por lo que espero:

#  1 [training]
# 2 [training]
# 3 [training]
# 4 [training]
#      1 [validation]
#      2 [validation]
# 5 [training]
# 6 [training]
# 7 [training]
# 8 [training]
#      1 [validation]
#      2 [validation]

pero se detiene cuando se realiza la primera validación:

# 1 [training]
# 2 [training]
# 3 [training]
# 4 [training]
#      1 [validation]
#      2 [validation]

Entonces, cómo usar estofeedable iterador entf.MonitoredTrainingSession?