tensorflow: Cannot create a stateful RNN with recurrent dropout

System information

• Have I written custom code (as opposed to using a stock example script provided in TensorFlow): Yes
• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): MacOSX 10.13.6
• Mobile device (e.g. iPhone 8, Pixel 2, Samsung Galaxy) if the issue happens on mobile device: N/A
• TensorFlow installed from (source or binary): binary
• TensorFlow version (use command below): tf.version.VERSION=2.0.0-dev20190413 tf.version.GIT_VERSION=v1.12.0-12481-gc7ce6f4cd9
• Python version: 3.6.8
• Bazel version (if compiling from source): N/A
• GCC/Compiler version (if compiling from source): N/A
• CUDA/cuDNN version: N/A
• GPU model and memory: N/A

Describe the current behavior I get an exception when trying to use recurrent_dropout in a stateful RNN:

.../tensorflow/python/ops/resource_variable_ops.py in __imul__(self, unused_other)
   1449
   1450   def __imul__(self, unused_other):
-> 1451     raise RuntimeError("Variable *= value not supported. Use "
   1452                        "`var.assign(var * value)` to modify the variable or "
   1453                        "`var = var * value` to get a new Tensor object.")

RuntimeError: Variable *= value not supported. Use `var.assign(var * value)` to modify the variable or `var = var * value` to get a new Tensor object.

The full stacktrace is below.

Describe the expected behavior No exception.

Code to reproduce the issue

from tensorflow import keras

model = keras.models.Sequential([
    keras.layers.GRU(128, return_sequences=True, stateful=True,
                     batch_input_shape=[32, None, 5],
                     recurrent_dropout=0.2)
])

Other info / logs Complete stacktrace:

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-1-3e98e7412ec2> in <module>
      4     keras.layers.GRU(128, return_sequences=True, stateful=True,
      5                      batch_input_shape=[32, None, 5],
----> 6                      recurrent_dropout=0.2)
      7 ])

.../tensorflow/python/training/tracking/base.py in _method_wrapper(self, *args, **kwargs)
    456     self._self_setattr_tracking = False  # pylint: disable=protected-access
    457     try:
--> 458       result = method(self, *args, **kwargs)
    459     finally:
    460       self._self_setattr_tracking = previous_value  # pylint: disable=protected-access

.../tensorflow/python/keras/engine/sequential.py in __init__(self, layers, name)
    106     if layers:
    107       for layer in layers:
--> 108         self.add(layer)
    109
    110   @property

.../tensorflow/python/training/tracking/base.py in _method_wrapper(self, *args, **kwargs)
    456     self._self_setattr_tracking = False  # pylint: disable=protected-access
    457     try:
--> 458       result = method(self, *args, **kwargs)
    459     finally:
    460       self._self_setattr_tracking = previous_value  # pylint: disable=protected-access

.../tensorflow/python/keras/engine/sequential.py in add(self, layer)
    167           # and create the node connecting the current layer
    168           # to the input layer we just created.
--> 169           layer(x)
    170           set_inputs = True
    171

.../tensorflow/python/keras/layers/recurrent.py in __call__(self, inputs, initial_state, constants, **kwargs)
    620
    621     if initial_state is None and constants is None:
--> 622       return super(RNN, self).__call__(inputs, **kwargs)
    623
    624     # If any of `initial_state` or `constants` are specified and are Keras

.../tensorflow/python/keras/engine/base_layer.py in __call__(self, inputs, *args, **kwargs)
    631                       base_layer_utils.AutoAddUpdates(self,
    632                                                       inputs)) as auto_updater:
--> 633                 outputs = call_fn(inputs, *args, **kwargs)
    634                 auto_updater.set_outputs(outputs)
    635

.../tensorflow/python/keras/layers/recurrent_v2.py in call(self, inputs, mask, training, initial_state)
    328           input_length=timesteps,
    329           time_major=self.time_major,
--> 330           zero_output_for_mask=self.zero_output_for_mask)
    331       # This is a dummy tensor for testing purpose.
    332       runtime = _runtime('unknown')

.../tensorflow/python/keras/backend.py in rnn(step_function, inputs, initial_states, go_backwards, mask, constants, unroll, input_length, time_major, zero_output_for_mask)
   3558     # the value is discarded.
   3559     output_time_zero, _ = step_function(input_time_zero,
-> 3560                                         initial_states + constants)
   3561     output_ta = tuple(
   3562         tensor_array_ops.TensorArray(

.../tensorflow/python/keras/layers/recurrent_v2.py in step(cell_inputs, cell_states)
    316
    317       def step(cell_inputs, cell_states):
--> 318         return self.cell.call(cell_inputs, cell_states, **kwargs)
    319
    320       last_output, outputs, states = K.rnn(

.../tensorflow/python/keras/layers/recurrent.py in call(self, inputs, states, training)
   1706
   1707       if 0. < self.recurrent_dropout < 1.:
-> 1708         h_tm1 *= rec_dp_mask[0]
   1709
   1710       if self.reset_after:

.../tensorflow/python/ops/resource_variable_ops.py in __imul__(self, unused_other)
   1449
   1450   def __imul__(self, unused_other):
-> 1451     raise RuntimeError("Variable *= value not supported. Use "
   1452                        "`var.assign(var * value)` to modify the variable or "
   1453                        "`var = var * value` to get a new Tensor object.")

RuntimeError: Variable *= value not supported. Use `var.assign(var * value)` to modify the variable or `var = var * value` to get a new Tensor object.