Fix the Attribute Error in Differentially private CNN on MNIST example notebook

[rajasekharporeddy]

Currently the differentially_private_sgd.ipynb example notebook fails with an attribute error during the training while iterating over the train_loader_batched dataset since it does not have the attribute as_numpy_iterator().

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
/tmp/ipykernel_1514/3612508851.py in <cell line: 0>()
      2 
      3 for epoch in range(NUM_EPOCHS):
----> 4   for batch in train_loader_batched.as_numpy_iterator():
      5     params, opt_state = train_step(params, opt_state, batch)
      6 

AttributeError: 'BatchMapDataset' object has no attribute 'as_numpy_iterator'

[rdyro]

Thanks!