I have read that, Pickle library is used to save trained model into a .pkl file to use it later. Also, we can save the weights of a trained model into a hdf5 format using
model.save()
and use those weights afterwards.
So why we use pickle instead of hdf5 files?