Explicitly raise a (clearer) error message when models end up in invalid states due to interleaving graph and eager.

In rare cases code may have run w/o crashing when in these invalid states, but it's safer to error with an explanation rather than risk silent failures/fragile behavior.

PiperOrigin-RevId: 321192744
Change-Id: I9e97ac3b7cea27c9b389e5202de9f1c09a4aa2b8

tensorflow/python/keras/engine/base_layer_v1.py

@@ -252,6 +252,9 @@ class Layer(base_layer.Layer):
     # might want to turn it off, like Sequential model.
     self._auto_track_sub_layers = True
 
+    # Mark this layer as having been originally built as a tf1 layer/model
+    self._originally_built_as_v1 = True
+
   @trackable.no_automatic_dependency_tracking
   @generic_utils.default
   def build(self, input_shape):
@@ -651,6 +654,8 @@ class Layer(base_layer.Layer):
       ValueError: if the layer's `call` method returns None (an invalid value).
       RuntimeError: if `super().__init__()` was not called in the constructor.
     """
+    self._assert_built_as_v1()
+
     if not hasattr(self, '_thread_local'):
       raise RuntimeError(
           'You must call `super().__init__()` in the layer constructor.')
@@ -818,6 +823,20 @@ class Layer(base_layer.Layer):
 
     return outputs
 
+  def _assert_built_as_v1(self):
+    if not hasattr(self, '_originally_built_as_v1'):
+      raise ValueError(
+          'Your Layer or Model is in an invalid state. This can happen if you '
+          'are interleaving estimator/non-estimator models or '
+          'interleaving models/layers made in tf.compat.v1.Graph.as_default() '
+          'with models/layers created outside of it. '
+          'Converting a model to an estimator (via model_to_estimator) '
+          'invalidates all models/layers made before the conversion (even '
+          'if they were not the model converted to an estimator). '
+          'Similarly, making a layer or a model inside a '
+          'a tf.compat.v1.Graph invalidates all layers/models you previously '
+          'made outside of the graph.')
+
   @property
   def dtype(self):
     return self._dtype_policy.variable_dtype

tensorflow/python/keras/engine/training_v1.py

@@ -303,6 +303,7 @@ class Model(training_lib.Model):
         ValueError: In case of invalid arguments for
             `optimizer`, `loss`, `metrics` or `sample_weight_mode`.
     """
+    self._assert_built_as_v1()
     self._run_eagerly = kwargs.pop('run_eagerly', None)
     self._experimental_run_tf_function = kwargs.pop(
         'experimental_run_tf_function', True)
@@ -773,6 +774,7 @@ class Model(training_lib.Model):
         ValueError: In case of mismatch between the provided input data
             and what the model expects.
     """
+    self._assert_built_as_v1()
     _keras_api_gauge.get_cell('fit_v1').set(True)
     # Legacy support
     if 'nb_epoch' in kwargs:
@@ -893,6 +895,7 @@ class Model(training_lib.Model):
     Raises:
         ValueError: in case of invalid arguments.
     """
+    self._assert_built_as_v1()
     _keras_api_gauge.get_cell('evaluate_v1').set(True)
     self._assert_compile_was_called()
     self._check_call_args('evaluate')
@@ -972,6 +975,7 @@ class Model(training_lib.Model):
             or in case a stateful model receives a number of samples
             that is not a multiple of the batch size.
     """
+    self._assert_built_as_v1()
     _keras_api_gauge.get_cell('predict_v1').set(True)
     self._check_call_args('predict')