Reed Wanderman-Milne 9765b5ab7f Improve mixed precision docstrings. ...

The docstrings have been reworded to make them more clear and concise. Some clarifying information is also added.

I removed some examples for uncommon use cases in order to shorten the docstrings, and rewrote or shortened other examples to make them easier and faster to read.

All references to experimental mixed precision APIs have been changed to use the non-experimental APIs. Examples now use the newly added attribute Layer.dtype_policy as well.

The section "How to use float64 in a Keras model" has been removed. Float64 can be enabled by setting floatx to float64 so I don't think its necessary to mention it in the policy section. Also, it's fairly obvious after reading the Policy docstring how to use float64 using policies: Just set the global policy to "float64".

I intend on cherrypicking this change into TF 2.4.

PiperOrigin-RevId: 339780420
Change-Id: I5f6ad44f54964114c398c306d2d7a4da39bb1c54

2020-11-04 11:22:31 -08:00

..

__init__.py

…

base_layer_test.py

Move mixed precision files out of experimental/ directory.

2020-10-21 15:19:21 -07:00

base_layer_utils_test.py

Make the sparse ops with Keras functional models error more verbose to aid debugging.

2020-07-17 12:22:12 -07:00

base_layer_utils.py

Refactor keras dependency to a common utility.

2020-10-20 16:06:05 -07:00

base_layer_v1.py

Move mixed precision files out of experimental/ directory.

2020-10-21 15:19:21 -07:00

base_layer.py

Improve mixed precision docstrings.

2020-11-04 11:22:31 -08:00

base_preprocessing_layer_test.py

In .adapt, don't freeze any shape values, only the number of dimensions.

2020-06-19 20:39:32 -07:00

base_preprocessing_layer_v1.py

Update TextVectorization to use internal layer adapt calls instead of its own combiner.

2020-05-19 09:26:19 -07:00

base_preprocessing_layer.py

Remove some of the usages of ops.EagerTensor from Keras. Specifically, the no-op usages (v1 keras code never runs when there are EagerTensors, so it never needs to be checked for v1 only code)

2020-09-28 12:19:39 -07:00

BUILD

Move mixed precision files out of experimental/ directory.

2020-10-21 15:19:21 -07:00

compile_utils_test.py

Update keras to use the public TF API for convert_to_tensor()

2020-08-27 20:57:16 -07:00

compile_utils.py

Replacing nest.is_sequence with nest.is_nested

2020-07-20 13:01:11 -07:00

control_flow_test.py

…

correctness_test.py

Remove passing experimental_run_tf_function in most tests.

2020-02-27 13:34:09 -08:00

data_adapter_test.py

Fix the data_adapter for dataset.Iterator.

2020-10-19 20:40:51 -07:00

data_adapter.py

Fix the data_adapter for dataset.Iterator.

2020-10-19 20:40:51 -07:00

deferred_sequential_test.py

Unifies the behaviors of "regular Sequential" (starts with an Input, wraps a Functional model) and "deferred Sequential" models (only gets built when it sees its input data for the first time).

2020-04-12 16:04:20 -07:00

feature_columns_integration_test.py

Update keras feature column test to explicitly use keras symbols.

2020-09-30 13:20:16 -07:00

functional_test.py

Fix input mapping logic for nested dictionaries in Functional Model.

2020-09-11 10:58:13 -07:00

functional.py

Serialize concrete function signature using structured_input_signature instead of function inputs.

2020-10-21 15:25:57 -07:00

input_layer_test.py

When a non-KerasTensor Tensor is passed to the tensor argument of keras.Input or InputLayer, make a KerasTensor directly from that tensor rather than erroring out.

2020-08-03 14:59:15 -07:00

input_layer.py

Switch all CompositeTensor instance checks in Keras to use a centralized tf_utils.is_extension_type util. This util will use the public ExtensionType api once it is in place.

2020-09-21 17:30:33 -07:00

input_spec_test.py

Move away from deprecated asserts

2020-06-30 16:10:22 -07:00

input_spec.py

Enable input spec checking for Functional models.

2020-08-03 10:12:01 -07:00

keras_tensor_test.py

Simplify string representation of KerasTensor.

2020-08-05 15:34:12 -07:00

keras_tensor.py

Add an additional NoneType check when converting a traced tensor to a KerasTensor.

2020-10-12 11:00:39 -07:00

node_test.py

Stops special-casing first argument in Keras's internal Node representation. This makes the special-casing-behavior in __call__ more explicit, and make it easier to reduce how much special casing of the first argument Keras does moving forward.

2020-04-25 13:47:50 -07:00

node.py

Fixed a bug with Functional model serialization when a layer that produces kwarg args of another layer had already been used to define a different functional model.

2020-08-04 18:38:55 -07:00

partial_batch_padding_handler.py

…

ragged_keras_tensor_test.py

Add support for RaggedTensor properties, instance methods, and class methods in the Keras functional API.

2020-10-16 12:11:06 -07:00

saving.py

…

sequential_test.py

Replace keras usages of private function.defun with tf.function

2020-09-15 11:06:13 -07:00

sequential.py

Fork the tf_inspect to keras, which use tf_decorator.unwrap as __internal__ API.

2020-09-14 17:26:35 -07:00

training_arrays_test.py

Remove passing experimental_run_tf_function in most tests.

2020-02-27 13:34:09 -08:00

training_arrays_v1.py

Remove some of the usages of ops.EagerTensor from Keras. Specifically, the no-op usages (v1 keras code never runs when there are EagerTensors, so it never needs to be checked for v1 only code)

2020-09-28 12:19:39 -07:00

training_dataset_test.py

Move away from deprecated asserts

2020-06-30 16:10:22 -07:00

training_distributed_v1.py

Add brackets to multi-line strings to make sure the entire string is captured.

2020-10-06 14:03:15 -07:00

training_eager_test.py

Stops special-casing first argument in Keras's internal Node representation. This makes the special-casing-behavior in __call__ more explicit, and make it easier to reduce how much special casing of the first argument Keras does moving forward.

2020-04-25 13:47:50 -07:00

training_eager_v1.py

Move mixed precision files out of experimental/ directory.

2020-10-21 15:19:21 -07:00

training_generator_test.py

Add v1 to the names of Keras files that are legacy-training-loop-specific.

2020-09-15 14:17:55 -07:00

training_generator_v1.py

Move v1-specific training_utils to training_utils_v1.py

2020-09-28 10:59:49 -07:00

training_gpu_test.py

Keras TF API usage cleanup: Copy test_util.use_gpu into Keras and replace the usage within Keras.

2020-07-21 14:30:11 -07:00

training_integration_test.py

Remove passing experimental_run_tf_function in most tests.

2020-02-27 13:34:09 -08:00

training_test.py

Update Keras instrumentation, and add tests to make sure it triggers. Specifically, make legacy TF layers no longer instrument as if they are Keras layers.

2020-10-15 15:58:53 -07:00

training_utils_v1_test.py

Move v1-specific training_utils to training_utils_v1.py

2020-09-28 10:59:49 -07:00

training_utils_v1.py

Remove some of the usages of ops.EagerTensor from Keras. Specifically, the no-op usages (v1 keras code never runs when there are EagerTensors, so it never needs to be checked for v1 only code)

2020-09-28 12:19:39 -07:00

training_utils.py

Move v1-specific training_utils to training_utils_v1.py

2020-09-28 10:59:49 -07:00

training_v1.py

Move mixed precision files out of experimental/ directory.

2020-10-21 15:19:21 -07:00

training.py

Move mixed precision files out of experimental/ directory.

2020-10-21 15:19:21 -07:00