Looks like Tensorflow suggests to use its data pipeline.
I\'m reading https://www.tensorflow.org/guide/data#consuming_sets_of_files
My code is like the followi
