GitLab

PiperOrigin-RevId: 413946342

PiperOrigin-RevId: 413928110

PiperOrigin-RevId: 413813975

With migration of form_utils_test, the reference context / executor will be dead code.

Further, there are outstanding bugs indicating that the reference context should be removed from the public API, and used in unit tests--however, it is not used for any unit tests any longer. Therefore, preferring to remove.

PiperOrigin-RevId: 413804412

Several sites in TFF are still passing in a context stack as the third argument to to_value; this changes ensures that this is no longer occurring.

PiperOrigin-RevId: 413784015

PiperOrigin-RevId: 413774989

* Added function to create a set of managers for a simulation to TFF.
* Fixed a bug in `CSVFileReleaseManager`; if you created an instance with a file that did not exist the CSVFileReleaseManager` would create the file for you, but if the folder did not exist it would not create the folder for you.
* Added test to cover bug in `CSVFileReleaseManager`.

PiperOrigin-RevId: 413759504

PiperOrigin-RevId: 413745813

PiperOrigin-RevId: 413722250

PiperOrigin-RevId: 413707584

This change promotes the capability to support user-defined classes from the concrete implementation of `FileProgramStateManager` to the abstract interface and makes it easier to write loops that contain program state containing user-defined classes.

PiperOrigin-RevId: 413696492

This API was adding complexity and it was difficult to reason about fault tolerance and initialization. This change is one step in reducing this complexity and make it easier to migrate to the Federated Program API.

PiperOrigin-RevId: 413486363

PiperOrigin-RevId: 413259738

Change the docstring of `federated_aggregate_keras_metric` as it can take either metrics or metric constructors now.

PiperOrigin-RevId: 413026905

PiperOrigin-RevId: 412953692

PiperOrigin-RevId: 412918171

PiperOrigin-RevId: 412219459

Pre-`main` instantiation of GRPC connections will no longer be supported in the future. This CL modifies the test annotation which accepted pre-initialized context harnesses to instead accept no-arg lambdas. This allows initialization of the test contexts to be delayed until the test itself is run.

PiperOrigin-RevId: 412120040

PiperOrigin-RevId: 411925855

Create a hierarchical histogram IterativeProcess that is compatible with tff.backends.mapreduce.MapReduceForm.

PiperOrigin-RevId: 411845363

PiperOrigin-RevId: 411842768

PiperOrigin-RevId: 411837027

Previously, we relied on numpy to provide helpful error messages in the case where the provided value did not match the required dtype. However, this resulted in somewhat unfriendly error messages such as the following (in the case where an int64 tensor was expected but a tf.data.Dataset was provided):

TypeError: Cannot cast scalar from dtype('O') to dtype('int64') according to the rule 'same_kind'

The additional context provided by this error message (the value being serialized, the Python type of the value being serialized, and the desired TFF TensorType) should help users find and correct errors more easily.

PiperOrigin-RevId: 411833481

PiperOrigin-RevId: 411828023

PiperOrigin-RevId: 411664233

PiperOrigin-RevId: 411629757

PiperOrigin-RevId: 411610540

PiperOrigin-RevId: 411173120

PiperOrigin-RevId: 411172320

This is a method to create a `tff.learning.models.FunctionalModel` from a `tf.keras.Model`.

PiperOrigin-RevId: 411169398

The CachingExecutor formerly played a critical role in the TFF execution stack, as it prevented multiple execution of the myriad duplicate computations in TFF's ASTs. Nowadays, TFF no longer generates these types of duplicate computations, and it is no longer to cache at this level to ensure performance.  Additionally, this type of caching plays poorly with nondeterministic computations.

Prior to this CL, the CachingExecutor was already unused outside of tests. This should be a no-op for TFF customers.

PiperOrigin-RevId: 411112532

PiperOrigin-RevId: 411090036

This change should enable the MacOS presubmit to succeed.

PiperOrigin-RevId: 410926931

In the tff.program library, there is usage of various `flatten` functions:

* `tf.nest.flatten`
* `tree.flatten`
* `structure_utils.flatten`, which is a wrapper around `tree.flatten_with_path`

This change does two things:

1. Consolidates usage of various `flatten` functions to the `tree` package, `tree` is used over `tf.nest` because it has more functionality.
2. Renames `structure_utils.flatten` to `structure_utils.flatten_with_name` to disambiguate this functionality.

PiperOrigin-RevId: 410910920

PiperOrigin-RevId: 410894245

PiperOrigin-RevId: 410631896

PiperOrigin-RevId: 410595976

PiperOrigin-RevId: 410587320

PiperOrigin-RevId: 410538549

PiperOrigin-RevId: 410538498