Assume I have some stream of input data.
How to create a pytorch dataset to resemble that stream?
How to also do it to support multiprocessing?
It is ob
