Allow variable reuse in function.

Change: 151807504
tensorflow · yifeif · Apr 4, 2017 · Mar 31, 2017 · Mar 31, 2017 · Mar 31, 2017
commit e4a8dc831dbf2894c79659d50aea73999c1ff173
diff --git a/tensorflow/python/framework/function.py b/tensorflow/python/framework/function.py
@@ -299,6 +299,7 @@ def getvar(self,
              shape=None,
              dtype=None,
              initializer=None,
+             reuse=None,
              trainable=True,
              collections=None,  # pylint: disable=redefined-outer-name
              use_resource=None,
@@ -319,6 +320,7 @@ def getvar(self,
           shape=shape,
           dtype=dtype,
           initializer=initializer,
+          reuse=reuse,
           trainable=trainable,
           collections=collections,
           use_resource=use_resource)
@@ -886,6 +888,11 @@ def foo(x, y):
   default graph. Because the addition of the function into the graph
   is deferred, the decorator can be used anywhere in the program.
 
+  Any variables created inside of the function are hoisted into the outer graph.
+  Note that the variables are created in the variable scope that was active
+  during the first call to the function. Subsequent function calls will refer to
+  the same set of variables.
+
   Definitions of functions are frozen in a graph as soon as the graph is used to
   create a session. Therefore, nodes using the function must be created in the
   graph before the corresponding session is created.

diff --git a/tensorflow/python/framework/function_test.py b/tensorflow/python/framework/function_test.py
@@ -722,6 +722,58 @@ def Bar(x):
       y = Bar(array_ops.zeros([1, 2, 3]))
       self.assertAllEqual(y.get_shape().as_list(), [1, 1, 2, 3])
 
+  def testVariableReuse(self):
+    def LinearWithReuse(input_tensor, reuse=None):
+      size = input_tensor.shape.dims[1]
+      with variable_scope.variable_scope("linear", reuse=reuse):
+        w = variable_scope.get_variable("w", shape=[size, size],
+                                        dtype=input_tensor.dtype)
+      return math_ops.matmul(input_tensor, w)
+
+    @function.Defun(dtypes.float32)
+    def Foo(inputs):
+      inputs = array_ops.reshape(inputs, [32, 100])
+      hidden = LinearWithReuse(inputs)
+      return LinearWithReuse(hidden, reuse=True)
+
+    input_op = array_ops.placeholder(shape=[32, 100], dtype=dtypes.float32)
+    output_op = Foo(input_op)
+
+    global_vars = variables.global_variables()
+    self.assertEqual(len(global_vars), 1)
+    self.assertEqual(global_vars[0].name, "linear/w:0")
+
+    with session.Session() as sess:
+      sess.run(variables.global_variables_initializer())
+      output_val = sess.run(output_op,
+                            feed_dict={input_op: np.random.rand(32, 100)})
+      self.assertEqual(output_val.shape, (32, 100))
+
+  def testFunctionCallInDifferentVariableScopes(self):
+    @function.Defun(dtypes.float32)
+    def Foo(inputs):
+      var = variable_scope.get_variable("var", shape=[10], dtype=dtypes.float32,
+                                        initializer=init_ops.ones_initializer())
+      return inputs + var
+
+    input_op = array_ops.placeholder(shape=[10], dtype=dtypes.float32)
+    with variable_scope.variable_scope("vs1"):
+      out1_op = Foo(input_op)
+
+    with variable_scope.variable_scope("vs2"):
+      out2_op = Foo(input_op)
+
+    global_vars = variables.global_variables()
+    self.assertEqual(len(global_vars), 1)
+    self.assertEqual(global_vars[0].name, "vs1/var:0")
+
+    with session.Session() as sess:
+      sess.run(variables.global_variables_initializer())
+      out1, out2 = sess.run([out1_op, out2_op],
+                            feed_dict={input_op: np.linspace(1, 10, 10)})
+      self.assertAllEqual(out1, np.linspace(2, 11, 10))
+      self.assertAllEqual(out2, np.linspace(2, 11, 10))
+
 
 class FunctionsFromProtos(test.TestCase):
 

diff --git a/tensorflow/python/ops/variable_scope.py b/tensorflow/python/ops/variable_scope.py
@@ -904,6 +904,7 @@ def get_variable(self,
                    dtype=None,
                    initializer=None,
                    regularizer=None,
+                   reuse=None,
                    trainable=True,
                    collections=None,
                    caching_device=None,
@@ -920,6 +921,8 @@ def get_variable(self,
       partitioner = self._partitioner
     if custom_getter is None:
       custom_getter = self._custom_getter
+    if reuse is None:
+      reuse = self._reuse
 
     full_name = self.name + "/" + name if self.name else name
     # Variable names only depend on variable_scope (full_name here),
@@ -942,7 +945,7 @@ def get_variable(self,
 
       return var_store.get_variable(
           full_name, shape=shape, dtype=dtype, initializer=initializer,
-          regularizer=regularizer, reuse=self.reuse, trainable=trainable,
+          regularizer=regularizer, reuse=reuse, trainable=trainable,
           collections=collections, caching_device=caching_device,
           partitioner=partitioner, validate_shape=validate_shape,
           use_resource=use_resource, custom_getter=custom_getter)