| .. |
|
BUILD
|
Cleanup the build deps for base_layer.
|
2020-06-07 17:54:40 -07:00 |
|
__init__.py
|
…
|
|
|
base_layer.py
|
Relax the error about functional api construction w/ a mix of symbolic and non-symbolic tensors for built-in layers (such as layers.add and layers.multiply where using constants is a common user pattern)
|
2020-07-16 21:40:28 -07:00 |
|
base_layer_test.py
|
Fix a critical breakage in `training` argument default value in inference for
|
2020-06-25 15:26:51 -07:00 |
|
base_layer_utils.py
|
Add an explicit distribution_strategy attribute to TrackableHandler. Since it's treated as a weight by Keras, it needs to pass DistStrat checks during compile.
|
2020-06-16 07:46:19 -07:00 |
|
base_layer_utils_test.py
|
This change adds WIP support for using `KerasTensor` objects in Keras's functional API instead of symbolic graph tf.Tensors. It is controlled by an internal behavior flag that is disabled by default and is not yet exposed in TF's APIs.
|
2020-06-05 15:47:28 -07:00 |
|
base_layer_v1.py
|
Explicitly raise a (clearer) error message when models end up in invalid states due to interleaving graph and eager.
|
2020-07-14 11:45:20 -07:00 |
|
base_preprocessing_layer.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_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 |
|
compile_utils.py
|
Roll forward of cl/316247127: Make sure compiled metrics are accessible after loading from H5 or SavedModel.
|
2020-06-19 10:22:33 -07:00 |
|
compile_utils_test.py
|
Fix issue - 1. tf.keras.Model.compile metrics do not respect masking since TF 2.2
|
2020-06-05 11:26:28 -07:00 |
|
control_flow_test.py
|
…
|
|
|
correctness_test.py
|
…
|
|
|
data_adapter.py
|
Avoid deprecation warning related to the use of `collections.Sequence`. It will stop working in Python 3.8.
|
2020-06-17 14:35:15 -07:00 |
|
data_adapter_test.py
|
Do not wrap a single x input passed to Model.fit in a tuple
|
2020-06-09 00:09:26 -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
|
Rollback the previous change of feature_column test.
|
2020-06-05 08:44:53 -07:00 |
|
functional.py
|
Fix input mapping issue when model is constructed/tested with dict input tensor.
|
2020-06-22 18:28:15 -07:00 |
|
functional_test.py
|
Merge pull request #41483 from tomerk/cherrypicks_DC0YA
|
2020-07-17 12:07:45 -07:00 |
|
input_layer.py
|
Fix invalid keras tensor isinstance check
|
2020-06-09 13:59:23 -07:00 |
|
input_spec.py
|
…
|
|
|
input_spec_test.py
|
…
|
|
|
keras_tensor.py
|
Fix sparse kerastensors to maintain dense shape information after converting to a placeholder.
|
2020-06-15 14:01:44 -07:00 |
|
node.py
|
This change adds WIP support for using `KerasTensor` objects in Keras's functional API instead of symbolic graph tf.Tensors. It is controlled by an internal behavior flag that is disabled by default and is not yet exposed in TF's APIs.
|
2020-06-05 15:47:28 -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 |
|
partial_batch_padding_handler.py
|
…
|
|
|
saving.py
|
…
|
|
|
sequential.py
|
Reduce 1-layer Functional.__call__ overhead by ~10%
|
2020-06-01 16:51:20 -07:00 |
|
sequential_test.py
|
Simplify Layer.add_udpate in v2 and update version_selector to use v1 inside a
|
2020-06-17 11:01:44 -07:00 |
|
training.py
|
Roll forward of cl/316247127: Make sure compiled metrics are accessible after loading from H5 or SavedModel.
|
2020-06-19 10:22:33 -07:00 |
|
training_arrays.py
|
…
|
|
|
training_arrays_test.py
|
…
|
|
|
training_dataset_test.py
|
…
|
|
|
training_distributed.py
|
…
|
|
|
training_eager.py
|
…
|
|
|
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_generator.py
|
…
|
|
|
training_generator_test.py
|
…
|
|
|
training_gpu_test.py
|
…
|
|
|
training_integration_test.py
|
…
|
|
|
training_test.py
|
PR #40169: ensure model initialized on ANY trackable attr set
|
2020-06-16 13:24:58 -07:00 |
|
training_utils.py
|
Switches keras.backend.placeholder + keras.backend.function to build a keras model when running eagerly (instead of trying to directly lift ops out of a graph into a concretefunction). Allows us to strip most of EagerDefinedExecutionFunction from the keras backend.
|
2020-05-21 11:53:44 -07:00 |
|
training_utils_test.py
|
…
|
|
|
training_v1.py
|
Explicitly raise a (clearer) error message when models end up in invalid states due to interleaving graph and eager.
|
2020-07-14 11:45:20 -07:00 |
f923fa474b