AttributeError: 'PrefetchDataset' object has no attribute 'make_one_shot_iterator'

[boobootoo2]

I get this error when running

/scripts/run_train.sh "./exp/my_experiment/" "0"

To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags. Traceback (most recent call last): File "/Users/jgs/Projects/python/musegan/scripts/../src/train.py", line 348, in <module> main() File "/Users/jgs/Projects/python/musegan/scripts/../src/train.py", line 172, in main train_x, _ = load_training_data(params, config) File "/Users/jgs/Projects/python/musegan/scripts/../src/train.py", line 105, in load_training_data train_x, train_y = dataset.make_one_shot_iterator().get_next(), None AttributeError: 'PrefetchDataset' object has no attribute 'make_one_shot_iterator' (jupyter) bash-3.2$

I read this issue thread for Tensorflow. Maybe this script needs refactoring for current version of tensorflow? Or maybe I need to use an older version of Tensorflow?
tensorflow/tensorflow#29252

[ruscot]

I had the same issue, to solve I replace the lane like this :
Last code :

    if params['is_conditional']:
        train_x, train_y = dataset.make_one_shot_iterator().get_next()
    else:
        train_x, train_y = dataset.make_one_shot_iterator().get_next(), None

New code :

    if params['is_conditional']:
        train_x, train_y = tf.compat.v1.data.make_one_shot_iterator(dataset).get_next()
    else:
        train_x, train_y = tf.compat.v1.data.make_one_shot_iterator(dataset).get_next(), None

[ruscot]

But it seems to have other issue due to tensorflow update.... A refactoring seems necessary no ?