Add Variable encoding so that functions with variable arguments can be saved to SavedModel.

PiperOrigin-RevId: 277826082
Change-Id: I38ab1cdf7990f449785271a0f37a10614efc7426

tensorflow/core/protobuf/struct.proto

@@ -1,10 +1,10 @@
 syntax = "proto3";
 
+package tensorflow;
+
 import "tensorflow/core/framework/tensor_shape.proto";
 import "tensorflow/core/framework/types.proto";
 
-package tensorflow;
-
 // `StructuredValue` represents a dynamically typed value representing various
 // data structures that are inspired by Python data structures typically used in
 // TensorFlow functions as inputs and outputs.
@@ -120,6 +120,7 @@ message TypeSpecProto {
     DATA_ITERATOR_SPEC = 6;   // IteratorSpec from data/ops/iterator_ops.py
     OPTIONAL_SPEC = 7;        // tf.OptionalSpec
     PER_REPLICA_SPEC = 8;     // PerReplicaSpec from distribute/values.py
+    VARIABLE_SPEC = 9;        // tf.VariableSpec
   }
   TypeSpecClass type_spec_class = 1;
 

tensorflow/python/BUILD

@@ -4672,6 +4672,18 @@ cuda_py_test(
     tags = ["no_windows_gpu"],
 )
 
+tf_py_test(
+    name = "variable_spec_test",
+    size = "small",
+    srcs = ["ops/variable_spec_test.py"],
+    additional_deps = [
+        ":framework_for_generated_wrappers",
+        ":framework_test_lib",
+        ":platform_test",
+        "//third_party/py/numpy",
+    ],
+)
+
 py_library(
     name = "training_lib",
     srcs = glob(

tensorflow/python/eager/function.py

@@ -143,9 +143,11 @@ def _flat_shape_list(*params):
   Returns:
     A list of entries containing either `None` or `TensorShape`.
   """
-  return [tensor_shape.TensorShape(x.shape)
+  return [
-          if isinstance(x, (ops.Tensor, tensor_spec.TensorSpec)) else None
+      tensor_shape.TensorShape(x.shape)
-          for x in nest.flatten(params, expand_composites=True)]
+      if isinstance(x, (ops.Tensor, tensor_spec.DenseSpec)) else None
+      for x in nest.flatten(params, expand_composites=True)
+  ]
 
 
 def _shape_less_specific_than(relaxed, to_check):
@@ -1651,7 +1653,7 @@ class ConcreteFunction(object):
                      self._func_graph.inputs[i].shape,
                      arg.shape))
       elif (self._signature is not None and
-            isinstance(self._signature[i], tensor_spec.TensorSpec)):
+            isinstance(self._signature[i], tensor_spec.DenseSpec)):
         tensor_inputs.append(
             ops.convert_to_tensor(arg, self._signature[i].dtype))
       else:
@@ -2208,7 +2210,8 @@ def _convert_inputs_to_signature(inputs, input_signature, flat_input_signature):
   need_packing = False
   for index, (value, spec) in enumerate(zip(flatten_inputs,
                                             flat_input_signature)):
-    if not _pywrap_utils.IsTensor(value):
+    if (isinstance(spec, tensor_spec.TensorSpec) and
+        not _pywrap_utils.IsTensor(value)):
       try:
         flatten_inputs[index] = ops.convert_to_tensor(
             value, dtype_hint=spec.dtype)
@@ -2392,11 +2395,12 @@ class Function(object):
           raise ValueError("Structure of Python function inputs does not match "
                            "input_signature.")
         flat_inputs = nest.flatten(args, expand_composites=True)
-        if any(not isinstance(arg, (ops.Tensor, tensor_spec.TensorSpec))
+        if any(not isinstance(arg, (ops.Tensor, tensor_spec.DenseSpec,
+                                    resource_variable_ops.BaseResourceVariable))
                for arg in flat_inputs):
           raise ValueError("When input_signature is provided, all inputs to "
-                           "the Python function must be Tensors or "
+                           "the Python function must be Tensors, Variables, "
-                           "tf.TensorSpec objects.")
+                           "tf.TensorSpec or tf.VariableSpec objects.")
         if any(not spec.is_compatible_with(other)
                for spec, other in zip(self.flat_input_signature, flat_inputs)):
           raise ValueError("Python inputs incompatible with input_signature: "
@@ -2701,7 +2705,7 @@ def register(func, *args, **kwargs):
 
 
 def validate_signature(signature):
-  if any(not isinstance(arg, tensor_spec.TensorSpec)
+  if any(not isinstance(arg, tensor_spec.DenseSpec)
          for arg in nest.flatten(signature, expand_composites=True)):
     raise TypeError("Invalid input_signature {}; input_signature must be "
                     "a possibly nested sequence of TensorSpec objects."

tensorflow/python/eager/function_test.py

@@ -2055,6 +2055,27 @@ class FunctionTest(test.TestCase, parameterized.TestCase):
     with self.assertRaisesRegexp(ValueError, 'does not match'):
       defined(rt5)
 
+  def testInputSignatureWithVariableArgs(self):
+
+    def f(v):
+      v.assign_add(1)
+
+    signature = [
+        resource_variable_ops.VariableSpec(shape=[], dtype=dtypes.int32)
+    ]
+    defined = function.defun(f, input_signature=signature)
+
+    v1 = variables.Variable(0)
+    v2 = variables.Variable(0)
+
+    defined(v1)
+    self.assertEqual(v1.numpy(), 1)
+    self.assertEqual(v2.numpy(), 0)
+
+    defined(v=v2)
+    self.assertEqual(v1.numpy(), 1)
+    self.assertEqual(v2.numpy(), 1)
+
   def testTensorKeywordArguments(self):
 
     def foo(a, b):

tensorflow/python/framework/func_graph.py

@@ -99,6 +99,9 @@ def convert_structure_to_signature(structure, arg_names=None):
     if isinstance(arg, composite_tensor.CompositeTensor):
       # TODO(b/133606651) Do we need to inject arg_name?
       return arg._type_spec  # pylint: disable=protected-access
+    if isinstance(arg, resource_variable_ops.BaseResourceVariable):
+      name = "/".join([str(p) for p in path])
+      return resource_variable_ops.VariableSpec(arg.shape, arg.dtype, name)
     if isinstance(arg, (
         int,
         float,
@@ -292,7 +295,7 @@ class FuncGraph(ops.Graph):
     if key not in self._deferred_captures:
 
       def convert_to_placeholder(s):
-        if not isinstance(s, tensor_spec.TensorSpec):
+        if not isinstance(s, tensor_spec.DenseSpec):
           raise TypeError(
               "Expected a nest of `TypeSpec` objects, found %s of type %s." %
               (s, type(s)))
@@ -1177,7 +1180,7 @@ def _get_defun_inputs(args, names, structure, flat_shapes=None):
 
     flattened = nest.flatten(arg_value, expand_composites=True)
     tensor_specs = [
-        arg for arg in flattened if isinstance(arg, tensor_spec.TensorSpec)
+        arg for arg in flattened if isinstance(arg, tensor_spec.DenseSpec)
     ]
     specified_names = [arg.name for arg in tensor_specs if arg.name]
     if specified_names and len(specified_names) < len(tensor_specs):
@@ -1209,7 +1212,20 @@ def _get_defun_inputs(args, names, structure, flat_shapes=None):
               "_user_specified_name",
               attr_value_pb2.AttrValue(s=compat.as_bytes(requested_name)))
         function_inputs.append(placeholder)
-      elif isinstance(arg, resource_variable_ops.BaseResourceVariable):
+      elif isinstance(arg, (resource_variable_ops.BaseResourceVariable,
+                            resource_variable_ops.VariableSpec)):
+        if isinstance(arg, resource_variable_ops.VariableSpec):
+          name = arg.name or name
+          with func_graph.outer_graph.as_default():
+            placeholder = graph_placeholder(dtypes.resource, arg.shape,
+                                            name=name)
+
+            arg = resource_variable_ops.BaseResourceVariable(
+                name=name,
+                shape=arg.shape,
+                dtype=arg.dtype,
+                handle=placeholder,
+                handle_name=name)
         # Capture arg variables to create placeholders for them. These will be
         # removed as captures after the function is traced (since otherwise we'd
         # just add it back with a new placeholder when the variable was

tensorflow/python/framework/tensor_spec.py

@@ -29,16 +29,13 @@ from tensorflow.python.framework import type_spec
 from tensorflow.python.util.tf_export import tf_export
 
 
-@tf_export("TensorSpec")
+class DenseSpec(type_spec.TypeSpec):
-class TensorSpec(type_spec.BatchableTypeSpec):
+  """Describes a dense object with shape, dtype, and name."""
-  """Describes a tf.Tensor.
-
-  Metadata for describing the `tf.Tensor` objects accepted or returned
-  by some TensorFlow APIs.
-  """
 
   __slots__ = ["_shape", "_shape_tuple", "_dtype", "_name"]
 
+  _component_specs = property(lambda self: self)
+
   def __init__(self, shape, dtype=dtypes.float32, name=None):
     """Creates a TensorSpec.
 
@@ -63,15 +60,6 @@ class TensorSpec(type_spec.BatchableTypeSpec):
   def from_spec(cls, spec, name=None):
     return cls(spec.shape, spec.dtype, name or spec.name)
 
-  @classmethod
-  def from_tensor(cls, tensor, name=None):
-    if isinstance(tensor, ops.EagerTensor):
-      return TensorSpec(tensor.shape, tensor.dtype, name)
-    elif isinstance(tensor, ops.Tensor):
-      return TensorSpec(tensor.shape, tensor.dtype, name or tensor.op.name)
-    else:
-      raise ValueError("`tensor` should be a tf.Tensor")
-
   @property
   def shape(self):
     """Returns the `TensorShape` that represents the shape of the tensor."""
@@ -87,25 +75,14 @@ class TensorSpec(type_spec.BatchableTypeSpec):
     """Returns the (optionally provided) name of the described tensor."""
     return self._name
 
-  def is_compatible_with(self, spec_or_tensor):
+  def is_compatible_with(self, spec_or_value):
-    """Returns True if spec_or_tensor is compatible with this TensorSpec.
+    return (isinstance(spec_or_value, (type(self), self.value_type)) and
-
+            self._dtype.is_compatible_with(spec_or_value.dtype) and
-    Two tensors are considered compatible if they have the same dtype
+            self._shape.is_compatible_with(spec_or_value.shape))
-    and their shapes are compatible (see `tf.TensorShape.is_compatible_with`).
-
-    Args:
-      spec_or_tensor: A tf.TensorSpec or a tf.Tensor
-
-    Returns:
-      True if spec_or_tensor is compatible with self.
-    """
-    return (isinstance(spec_or_tensor, (TensorSpec, ops.Tensor)) and
-            self._dtype.is_compatible_with(spec_or_tensor.dtype) and
-            self._shape.is_compatible_with(spec_or_tensor.shape))
 
   def __repr__(self):
-    return "TensorSpec(shape={}, dtype={}, name={})".format(
+    return "{}(shape={}, dtype={}, name={})".format(
-        self.shape, repr(self.dtype), repr(self.name))
+        type(self).__name__, self.shape, repr(self.dtype), repr(self.name))
 
   def __hash__(self):
     return hash((self._shape_tuple, self.dtype))
@@ -120,19 +97,60 @@ class TensorSpec(type_spec.BatchableTypeSpec):
   def __ne__(self, other):
     return not self == other
 
-  value_type = property(lambda self: ops.Tensor)
-
   def most_specific_compatible_type(self, other):
     if (type(self) is not type(other)) or (self._dtype != other.dtype):
       raise ValueError("Types are not compatible: %r vs %r" % (self, other))
     shape = self._shape.most_specific_compatible_shape(other.shape)
     name = self._name if self._name == other.name else None
-    return TensorSpec(shape, self._dtype, name)
+    return type(self)(shape, self._dtype, name)
 
   def _serialize(self):
     return (self._shape, self._dtype, self._name)
 
-  _component_specs = property(lambda self: self)
+  def _to_legacy_output_types(self):
+    return self._dtype
+
+  def _to_legacy_output_shapes(self):
+    return self._shape
+
+  def _to_legacy_output_classes(self):
+    return self.value_type
+
+
+@tf_export("TensorSpec")
+class TensorSpec(DenseSpec, type_spec.BatchableTypeSpec):
+  """Describes a tf.Tensor.
+
+  Metadata for describing the `tf.Tensor` objects accepted or returned
+  by some TensorFlow APIs.
+  """
+
+  __slots__ = []
+
+  def is_compatible_with(self, spec_or_tensor):  # pylint:disable=useless-super-delegation
+    """Returns True if spec_or_tensor is compatible with this TensorSpec.
+
+    Two tensors are considered compatible if they have the same dtype
+    and their shapes are compatible (see `tf.TensorShape.is_compatible_with`).
+
+    Args:
+      spec_or_tensor: A tf.TensorSpec or a tf.Tensor
+
+    Returns:
+      True if spec_or_tensor is compatible with self.
+    """
+    return super(TensorSpec, self).is_compatible_with(spec_or_tensor)
+
+  @classmethod
+  def from_tensor(cls, tensor, name=None):
+    if isinstance(tensor, ops.EagerTensor):
+      return TensorSpec(tensor.shape, tensor.dtype, name)
+    elif isinstance(tensor, ops.Tensor):
+      return TensorSpec(tensor.shape, tensor.dtype, name or tensor.op.name)
+    else:
+      raise ValueError("`tensor` should be a tf.Tensor")
+
+  value_type = property(lambda self: ops.Tensor)
 
   def _to_components(self, value):
     try:
@@ -174,15 +192,6 @@ class TensorSpec(type_spec.BatchableTypeSpec):
       raise ValueError("Unbatching a tensor is only supported for rank >= 1")
     return TensorSpec(self._shape[1:], self._dtype)
 
-  def _to_legacy_output_types(self):
-    return self._dtype
-
-  def _to_legacy_output_shapes(self):
-    return self._shape
-
-  def _to_legacy_output_classes(self):
-    return ops.Tensor
-
 
 # TODO(b/133606651): Should is_compatible_with should check min/max bounds?
 class BoundedTensorSpec(TensorSpec):

tensorflow/python/ops/resource_variable_ops.py

@@ -33,6 +33,7 @@ from tensorflow.python.framework import cpp_shape_inference_pb2
 from tensorflow.python.framework import dtypes
 from tensorflow.python.framework import ops
 from tensorflow.python.framework import tensor_shape
+from tensorflow.python.framework import tensor_spec
 from tensorflow.python.ops import array_ops
 from tensorflow.python.ops import gen_array_ops
 from tensorflow.python.ops import gen_logging_ops
@@ -1964,3 +1965,24 @@ def copy_to_graph_uninitialized(var):
 ops.NotDifferentiable("Assert")
 ops.NotDifferentiable("VarIsInitializedOp")
 ops.NotDifferentiable("VariableShape")
+
+
+class VariableSpec(tensor_spec.DenseSpec):
+  """Describes a tf.Variable."""
+
+  __slots__ = []
+
+  value_type = property(lambda self: BaseResourceVariable)
+
+  def _to_components(self, value):
+    raise NotImplementedError
+
+  def _from_components(self, components):
+    raise NotImplementedError
+
+  def _from_compatible_tensor_list(self, tensor_list):
+    assert len(tensor_list) == 1
+    return tensor_list[0]
+
+
+_pywrap_utils.RegisterType("VariableSpec", VariableSpec)

tensorflow/python/ops/variable_spec_test.py

@@ -0,0 +1,66 @@
+# Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+"""Tests for VariableSpec."""
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+from tensorflow.python.framework import constant_op
+from tensorflow.python.framework import dtypes
+from tensorflow.python.framework import tensor_shape
+from tensorflow.python.ops import resource_variable_ops
+from tensorflow.python.platform import test
+
+VariableSpec = resource_variable_ops.VariableSpec
+
+
+class VariableSpecTest(test.TestCase):
+
+  def test_properties(self):
+    spec = VariableSpec(shape=(1, 2, 3), dtype=dtypes.float64, name='vs')
+    self.assertEqual('vs', spec.name)
+    self.assertEqual(tensor_shape.TensorShape((1, 2, 3)), spec.shape)
+    self.assertEqual(dtypes.float64, spec.dtype)
+
+  def test_compatibility(self):
+    spec = VariableSpec(shape=None)
+    spec2 = VariableSpec(shape=[None, 15])
+    spec3 = VariableSpec(shape=[None])
+
+    self.assertTrue(spec.is_compatible_with(spec2))
+    self.assertFalse(spec2.is_compatible_with(spec3))
+
+    var = resource_variable_ops.UninitializedVariable(
+        shape=[3, 15], dtype=dtypes.float32)
+    var2 = resource_variable_ops.UninitializedVariable(
+        shape=[3], dtype=dtypes.int32)
+
+    self.assertTrue(spec2.is_compatible_with(var))
+    self.assertFalse(spec3.is_compatible_with(var2))
+
+    spec4 = VariableSpec(shape=None, dtype=dtypes.int32)
+    spec5 = VariableSpec(shape=[None], dtype=dtypes.int32)
+
+    self.assertFalse(spec.is_compatible_with(spec4))
+    self.assertTrue(spec4.is_compatible_with(spec5))
+    self.assertTrue(spec4.is_compatible_with(var2))
+
+    tensor = constant_op.constant([1, 2, 3])
+    self.assertFalse(spec4.is_compatible_with(tensor))
+
+
+if __name__ == '__main__':
+  test.main()

tensorflow/python/saved_model/load_test.py

@@ -54,9 +54,9 @@ from tensorflow.python.lib.io import file_io
 from tensorflow.python.module import module
 from tensorflow.python.ops import array_ops
 from tensorflow.python.ops import cond_v2
-from tensorflow.python.ops import gen_resource_variable_ops
 from tensorflow.python.ops import lookup_ops
 from tensorflow.python.ops import math_ops
+from tensorflow.python.ops import resource_variable_ops
 from tensorflow.python.ops import variable_scope
 from tensorflow.python.ops import variables
 from tensorflow.python.ops.ragged import ragged_factory_ops
@@ -975,8 +975,6 @@ class LoadTest(test.TestCase, parameterized.TestCase):
                                     x=constant_op.constant(2.)).numpy())
 
   def test_concrete_function_variable_argument(self, cycles):
-    # TODO(allenl): Fix variables in input signatures.
-    self.skipTest("Need to fix encoding of variables in inputs signatures")
     capture = variables.Variable(0)
 
     @def_function.function
@@ -984,14 +982,29 @@ class LoadTest(test.TestCase, parameterized.TestCase):
       v.assign_add(1)
       capture.assign_sub(1)
 
+    @def_function.function(input_signature=[
+        resource_variable_ops.VariableSpec(shape=[], dtype=dtypes.int32)
+    ])
+    def func_with_input_signature(v):
+      v.assign_add(5)
+      capture.assign_sub(5)
+      return 1
+
     vsave = variables.Variable(1)
     root = tracking.AutoTrackable()
     root.f = func.get_concrete_function(vsave)
+    root.f_sig = func_with_input_signature.get_concrete_function()
     root.capture = capture
+
     self.assertEqual(1, vsave.numpy())
     root.f(vsave)
     self.assertEqual(2, vsave.numpy())
     self.assertEqual(-1, capture.numpy())
+
+    root.f_sig(vsave)
+    self.assertEqual(7, vsave.numpy())
+    self.assertEqual(-6, capture.numpy())
+
     imported = cycle(root, cycles)
 
     vload = variables.Variable(1)
@@ -999,8 +1012,13 @@ class LoadTest(test.TestCase, parameterized.TestCase):
     self.assertEqual(2, vload.numpy())
     imported.f(v=vload)
     self.assertEqual(3, vload.numpy())
-    self.assertEqual(-3, imported.capture.numpy())
+    self.assertEqual(-8, imported.capture.numpy())
-    self.assertEqual(-1, capture.numpy())
+
+    imported.f_sig(v=vload)
+    self.assertEqual(8, vload.numpy())
+    self.assertEqual(-13, imported.capture.numpy())
+
+    self.assertEqual(-6, capture.numpy())
 
   def test_function_and_component(self, cycles):
 
@@ -1644,7 +1662,7 @@ class LoadTest(test.TestCase, parameterized.TestCase):
   def test_destroy_resource(self, cycles):
 
     def get_handle():
-      return gen_resource_variable_ops.var_handle_op(
+      return resource_variable_ops.var_handle_op(
           shape=tensor_shape.as_shape([]),
           dtype=dtypes.float32,
           shared_name="my_var_name",
@@ -1655,7 +1673,7 @@ class LoadTest(test.TestCase, parameterized.TestCase):
 
       def destroy_resource(self):
         handle = get_handle()
-        gen_resource_variable_ops.destroy_resource_op(
+        resource_variable_ops.destroy_resource_op(
             handle, ignore_lookup_error=True)
 
     class MyResource(tracking.TrackableResource):
@@ -1669,7 +1687,7 @@ class LoadTest(test.TestCase, parameterized.TestCase):
         return get_handle()
 
       def _initialize(self):
-        gen_resource_variable_ops.assign_variable_op(
+        resource_variable_ops.assign_variable_op(
             self.resource_handle, 1.0, name="assign")
 
     class MyModel(tracking.AutoTrackable):
@@ -1681,10 +1699,9 @@ class LoadTest(test.TestCase, parameterized.TestCase):
       @def_function.function(input_signature=[])
       def increase(self):
         handle = self.resource.resource_handle
-        gen_resource_variable_ops.assign_add_variable_op(
+        resource_variable_ops.assign_add_variable_op(
             handle, 10.0, name="assign_add")
-        return gen_resource_variable_ops.read_variable_op(
+        return resource_variable_ops.read_variable_op(handle, dtypes.float32)
-            handle, dtypes.float32)
 
     root = MyModel()
     imported = cycle(root, cycles)
@@ -1699,7 +1716,7 @@ class LoadTest(test.TestCase, parameterized.TestCase):
     # Try to destroy the resource again, should fail.
     with self.assertRaisesRegexp(errors.NotFoundError,
                                  r"Resource .* does not exist."):
-      gen_resource_variable_ops.destroy_resource_op(
+      resource_variable_ops.destroy_resource_op(
           handle, ignore_lookup_error=False)
 
   def test_function_called_as_operation(self, cycles):

tensorflow/python/saved_model/nested_structure_coder.py

@@ -44,6 +44,7 @@ from tensorflow.python.framework import indexed_slices
 from tensorflow.python.framework import sparse_tensor
 from tensorflow.python.framework import tensor_shape
 from tensorflow.python.framework import tensor_spec
+from tensorflow.python.ops import resource_variable_ops
 from tensorflow.python.ops import tensor_array_ops
 from tensorflow.python.ops.ragged import ragged_tensor
 from tensorflow.python.util import compat
@@ -467,6 +468,8 @@ class _TypeSpecCodec(object):
           optional_ops.OptionalSpec,
       struct_pb2.TypeSpecProto.PER_REPLICA_SPEC:
           values.PerReplicaSpec,
+      struct_pb2.TypeSpecProto.VARIABLE_SPEC:
+          resource_variable_ops.VariableSpec,
   }
 
   # Mapping from type (TypeSpec subclass) to enum value.

tensorflow/python/saved_model/save_test.py

@@ -45,6 +45,7 @@ from tensorflow.python.module import module
 from tensorflow.python.ops import array_ops
 from tensorflow.python.ops import lookup_ops
 from tensorflow.python.ops import math_ops
+from tensorflow.python.ops import resource_variable_ops
 from tensorflow.python.ops import variables
 from tensorflow.python.saved_model import loader
 from tensorflow.python.saved_model import loader_impl
@@ -429,6 +430,18 @@ class SaveTest(test.TestCase):
     self.assertAllClose({"output_0": 3 * (1 + 4 + 9 + 16)},
                         _import_and_infer(save_dir, {"x": 3}))
 
+  def test_variable_args_cannot_be_used_as_signature(self):
+    @def_function.function(input_signature=[
+        resource_variable_ops.VariableSpec(shape=[], dtype=dtypes.int32)])
+    def f(unused_v):
+      return 1
+    root = tracking.AutoTrackable()
+    root.f = f.get_concrete_function()
+    with self.assertRaisesRegexp(ValueError,
+                                 "tf.Variable inputs cannot be exported"):
+      save.save(root, os.path.join(self.get_temp_dir(), "saved_model"),
+                signatures=root.f)
+
 
 class SavingOptionsTest(test.TestCase):
 

tensorflow/python/saved_model/signature_serialization.py

@@ -22,6 +22,7 @@ from tensorflow.python.eager import def_function
 from tensorflow.python.eager import function as defun
 from tensorflow.python.framework import ops
 from tensorflow.python.framework import tensor_spec
+from tensorflow.python.ops import resource_variable_ops
 from tensorflow.python.saved_model import revived_types
 from tensorflow.python.saved_model import signature_constants
 from tensorflow.python.training.tracking import base
@@ -51,12 +52,21 @@ def _valid_signature(concrete_function):
     # 1.x style.
     return False
   try:
+    _validate_inputs(concrete_function)
     _normalize_outputs(concrete_function.structured_outputs, "unused", "unused")
   except ValueError:
     return False
   return True
 
 
+def _validate_inputs(concrete_function):
+  if any(isinstance(inp, resource_variable_ops.VariableSpec)
+         for inp in nest.flatten(
+             concrete_function.structured_input_signature)):
+    raise ValueError(("Functions that expect tf.Variable inputs cannot be "
+                      "exported as signatures."))
+
+
 def find_function_to_export(saveable_view):
   """Function to export, None if no suitable function was found."""
   # If the user did not specify signatures, check the root object for a function
@@ -98,6 +108,8 @@ def canonicalize_signatures(signatures):
            "got {}. Only `tf.functions` with an input signature or "
            "concrete functions can be used as a signature.").format(function))
 
+    _validate_inputs(signature_function)
+
     # Re-wrap the function so that it returns a dictionary of Tensors. This
     # matches the format of 1.x-style signatures.
     # pylint: disable=cell-var-from-loop

tensorflow/python/util/util.cc

@@ -512,21 +512,23 @@ bool IsCompositeTensorHelper(PyObject* o) {
   return check_cache->CachedLookup(o);
 }
 
-// Returns 1 if `o` is an instance of TypeSpec, but is not TensorSpec.
+// Returns 1 if `o` is an instance of TypeSpec, but is not TensorSpec or
+// VariableSpec.
 // Returns 0 otherwise.
 // Returns -1 if an error occurred.
 bool IsTypeSpecHelper(PyObject* o) {
   static auto* const check_cache = new CachedTypeCheck([](PyObject* to_check) {
     int is_type_spec = IsInstanceOfRegisteredType(to_check, "TypeSpec");
-    int is_tensor_spec = IsInstanceOfRegisteredType(to_check, "TensorSpec");
+    int is_dense_spec = (IsInstanceOfRegisteredType(to_check, "TensorSpec") ||
-    if ((is_type_spec == -1) || (is_tensor_spec == -1)) return -1;
+                         IsInstanceOfRegisteredType(to_check, "VariableSpec"));
-    return static_cast<int>(is_type_spec && !is_tensor_spec);
+    if ((is_type_spec == -1) || (is_dense_spec == -1)) return -1;
+    return static_cast<int>(is_type_spec && !is_dense_spec);
   });
   return check_cache->CachedLookup(o);
 }
 
 // Returns 1 if `o` is a (non-string) sequence or CompositeTensor or
-// (non-TensorSpec) TypeSpec.
+// (non-TensorSpec and non-VariableSpec) TypeSpec.
 // Returns 0 otherwise.
 // Returns -1 if an error occurred.
 int IsSequenceOrCompositeHelper(PyObject* o) {

tensorflow/tools/api/golden/v1/tensorflow.-tensor-spec.pbtxt

@@ -1,6 +1,7 @@
 path: "tensorflow.TensorSpec"
 tf_class {
   is_instance: "<class \'tensorflow.python.framework.tensor_spec.TensorSpec\'>"
+  is_instance: "<class \'tensorflow.python.framework.tensor_spec.DenseSpec\'>"
   is_instance: "<class \'tensorflow.python.framework.type_spec.BatchableTypeSpec\'>"
   is_instance: "<class \'tensorflow.python.framework.type_spec.TypeSpec\'>"
   is_instance: "<type \'object\'>"

tensorflow/tools/api/golden/v2/tensorflow.-tensor-spec.pbtxt

@@ -1,6 +1,7 @@
 path: "tensorflow.TensorSpec"
 tf_class {
   is_instance: "<class \'tensorflow.python.framework.tensor_spec.TensorSpec\'>"
+  is_instance: "<class \'tensorflow.python.framework.tensor_spec.DenseSpec\'>"
   is_instance: "<class \'tensorflow.python.framework.type_spec.BatchableTypeSpec\'>"
   is_instance: "<class \'tensorflow.python.framework.type_spec.TypeSpec\'>"
   is_instance: "<type \'object\'>"