PiperOrigin-RevId: 321785542

* Regex search replace.
* Removed unused imports.
* Removed BUILD deps.

PiperOrigin-RevId: 314969638

PiperOrigin-RevId: 313410789

Replace `import tensorflow.compat.v2 as tf` with `import tensorflow as tf` in the `learning` module.

PiperOrigin-RevId: 310572253

PiperOrigin-RevId: 310377468

Store the possible Python container of the result of `tff.federated_computation`s, for use in the reference executor.

PiperOrigin-RevId: 306885458

Removes hardwired training time metric from keras model. No longer used after removal of trainable model.

PiperOrigin-RevId: 304246695

PiperOrigin-RevId: 302944673

Step 1 of 3.

PiperOrigin-RevId: 297684545

-   Removes `tff.learning.from_compiled_keras_model()` from API.
-   Removes unused `tff.learning._TrainableKerasModel`.
-   Removes usage of `tf.keras.Model.output_names` and
    `tf.keras.Model.loss_functions` in `tff.learning._KerasModel` to be
    compliant with upcoming Keras refactoring.

As a side-effect of this change, TFF will no longer supported (though no known
usages):

1.   Specifying multiple loss functions or loss weights via dictionaries
keyed by model output name. Instead, using ordered sequences of
functions mapping to outputs is still possible.
2.   Sublcassed `tf.keras.Model` instances cannot be used in TFF.

PiperOrigin-RevId: 295824509

* This is required for parameterized test to work because decorators are parsed before the main function is called, if there is tf code in the parameters it may be not obvious why things are failing.

PiperOrigin-RevId: 294983486

Deprecated `tff.learning.from_compiled_keras_model`, point users to `tff.learning.from_keras_model`.

This is a small step into moving towards (only) using custom client training loops, rather than relying on Keras. This will also allow us to introspect Keras less, and give TFF libraries and users more flexibility and composibility.

PiperOrigin-RevId: 292603966

PiperOrigin-RevId: 292427614

PiperOrigin-RevId: 285801872

PiperOrigin-RevId: 285116495

de-duplication strategy in non-eager mode.

TF, under certain circumstances, prepends `_#` numbers to variables with the
same name. This appears to happen in non-eager mode and in some occasions with
Keras. TFF previously used these names for the type signature, which would
cause problems when the same model was created more than once in the same graph
context.

Instead, rely on the ordering of the sequence of variables contained in the
model and use the unnamed structure for types.

PiperOrigin-RevId: 285032395

PiperOrigin-RevId: 284254397

PiperOrigin-RevId: 284050731

PiperOrigin-RevId: 282830397

PiperOrigin-RevId: 281347566

PiperOrigin-RevId: 280784907

PiperOrigin-RevId: 280767713

PiperOrigin-RevId: 280689474

Previously tried calling shape on a nested structure of tensors, which will not
work. Instead flatten the structre and call shape on the first tensor (all
input tensors are expected to have the same first dimension size, the batch
size).

Updated example models and tests, ensured the test failed under the current
code and passes after this change.

Fixes #785

PiperOrigin-RevId: 280221692

The unittest adding the example to FedAvg tests currently does not run because
TFF doesn't have its hands on the lookup table initializer (see
#783). This test is disabled by
default (since it won't pass) and any fix should enable and ensure it passes.

Switch to the local executor for tests to make them faster, and decrease the
size to 'medium' to reflect this improvement.

PiperOrigin-RevId: 280083440

PiperOrigin-RevId: 266383040

Instead of checking for attribute, check that attribute value is not None.

PiperOrigin-RevId: 264409455

PiperOrigin-RevId: 263172828

Change calls from `tensor.numpy()` to `self.evaluate(tensor)`, which is compatible in graph and eager mode.

Additionally tighten up the assertions on nested structures in a few unittests.

PiperOrigin-RevId: 251246960

python/learning.

PiperOrigin-RevId: 250908176

* Raise errors with explicit parameters

* No warnings, throw error if param names don't match

* Add test case for `loss_weights` with wrong output keys

* Format and clean up files

* Update the documentation in `model.py` and `keras_utils.py`

* Add more test cases to `keras_utils_test.py` as sub-tests

* Change loss aggregation from average to sum of individual losses

PiperOrigin-RevId: 249909915

PiperOrigin-RevId: 249666998

Without this, input lists would be converted to single tensors which prevented
the use of Keras models that support multiple inputs.

PiperOrigin-RevId: 248348837

- Correctly use is_named_tuple to detect python namedtuple types.

PiperOrigin-RevId: 246586453

PiperOrigin-RevId: 246164724