Fix the Attribute Error in gradient_accumulation_and_microbatching.ipynb example notebook

[rajasekharporeddy]

Currently the gradient_accumulation_and_microbatching.ipynb example notebook fails with an attribute error at final cell due to the usage ".get_value()" on JAX Array object.

--------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
/tmp/ipykernel_2474/647667932.py in <cell line: 0>()
      1 # This shows the norm for the gradient of the embedding layer per example.
      2 # High uniformity of the norm values is encouraging.
----> 3 aux.metrics['embedding']['embedding'].get_value()

/usr/local/lib/python3.12/dist-packages/flax/nnx/variablelib.py in __getattr__(self, name)
    279     if name in object.__getattribute__(self, '_var_metadata'):
    280       return self._var_metadata[name]
--> 281     return getattr(self.raw_value, name)
    282
    283   def __setattr__(self, name: str, value: tp.Any):

AttributeError: 'jaxlib._jax.ArrayImpl' object has no attribute 'get_value'

This is fixed by using .value instead of .get_value() in the final cell to display the out.

Repro Colab notebook: gist

Working notebook: gist