Remove the cloning argument to compile().

Keras models are distributed by cloning in graph mode and without cloning in eager mode as of the change # 258652546.

PiperOrigin-RevId: 258893048

tensorflow/python/keras/distribute/distributed_training_utils.py

@@ -638,9 +638,7 @@ def _prepare_feed_values(model, inputs, targets, sample_weights, mode):
 def is_distributing_by_cloning(model):
   """Decide whether this model is going to be distributed via cloning.
 
-  We are going to distribute the model by cloning if the user has signaled
-  that intent by setting `cloning=True` in `Model.compile()` unless we are in
-  graph mode.
+  We are going to distribute the model by cloning in graph mode.
 
   Args:
     model: Keras model to distribute.

tensorflow/python/keras/engine/training.py

@@ -245,7 +245,6 @@ class Model(network.Network):
     if ((sample_weight_mode is not None)
         or (target_tensors is not None)
         or (weighted_metrics is not None)
-        or (kwargs.get('cloning', False))
         or not context.executing_eagerly()):
       # Fallback out of things that aren't supported with v2 loops
       self._run_distributed = False
@@ -269,12 +268,6 @@ class Model(network.Network):
           self._distribution_strategy = (
               distribution_strategy_context.get_strategy())
 
-    # Check whether the experimental feature of distributing the Model without
-    # cloning is requested.
-    # TODO(b/124517980, b/124377929): Remove this temporary undocumented way
-    # of enabling the feature and graduate it to the main distributed code path.
-    self._cloning = kwargs.pop('cloning', False)
-
     if not self._run_distributed:
       self._validate_compile_param_for_distribution_strategy(self.run_eagerly,
                                                              sample_weight_mode,
@@ -479,8 +472,7 @@ class Model(network.Network):
         # TODO(scottzhu): Finish getting sequences working with the v2 loops.
         and not isinstance(inputs, (data_utils.Sequence))
         and not distributed_training_utils.is_tpu_strategy(
-            self._distribution_strategy)
-        and not getattr(self, '_cloning', False)):
+            self._distribution_strategy)):
       return training_v2.Loop()
 
     # Case 1: distribution strategy.
@@ -2417,8 +2409,7 @@ class Model(network.Network):
           loss_weights=self.loss_weights,
           target_tensors=target_tensors,
           run_eagerly=self.run_eagerly,
-          run_distributed=self._run_distributed,
-          cloning=self._cloning)
+          run_distributed=self._run_distributed)
 
     # In graph mode, if we had just set inputs and targets as symbolic tensors
     # by invoking build and compile on the model respectively, we do not have to

tensorflow/python/keras/engine/training_arrays.py

@@ -283,8 +283,6 @@ def model_iteration(model,
         # Get outputs.
         try:
           # `ins` can be callable in tf.distribute.Strategy + eager case.
-          # TODO(b/134179782):  Simplify this condition when cloning never
-          # happens.
           if not callable(ins) or (
               model._distribution_strategy and
               not distributed_training_utils.is_distributing_by_cloning(model)):