I am getting a tfrecord corrupted error while training. The tfrecord was created by threading using joblib. See the code below.
Parallel(n_jobs=-1, require=\'
