modifying dataset on the fly with TFRecordDataset type

Question

I am iterating a TFRecordDataset object. When I do the following;

import numpy

print (numpy.int32(dataset.shape).tolist())

it gives me the s