Error from PlaidML when running DDPG

PlaidMLFailsDDPG.log

$ python ddpg_pendulum.py 
/Users/fak/.virtualenvs/control/lib/python2.7/site-packages/gym/__init__.py:15: UserWarning: gym.undo_logger_setup is deprecated. gym no longer modifies the global logging configuration
  warnings.warn("gym.undo_logger_setup is deprecated. gym no longer modifies the global logging configuration")
WARN: gym.spaces.Box autodetected dtype as <type 'numpy.float32'>. Please provide explicit dtype.
WARN: gym.spaces.Box autodetected dtype as <type 'numpy.float32'>. Please provide explicit dtype.
INFO:plaidml:Opening device "opencl_amd_amd_radeon_pro_vega_56_compute_engine.0"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
flatten_1 (Flatten)          (None, 3)                 0         
_________________________________________________________________
dense_1 (Dense)              (None, 16)                64        
_________________________________________________________________
activation_1 (Activation)    (None, 16L)               0         
_________________________________________________________________
dense_2 (Dense)              (None, 16)                272       
_________________________________________________________________
activation_2 (Activation)    (None, 16L)               0         
_________________________________________________________________
dense_3 (Dense)              (None, 16)                272       
_________________________________________________________________
activation_3 (Activation)    (None, 16L)               0         
_________________________________________________________________
dense_4 (Dense)              (None, 1)                 17        
_________________________________________________________________
activation_4 (Activation)    (None, 1)                 0         
=================================================================
Total params: 625
Trainable params: 625
Non-trainable params: 0
_________________________________________________________________
None
____________________________________________________________________________________________________
Layer (type)                     Output Shape          Param #     Connected to                     
====================================================================================================
observation_input (InputLayer)   (None, 1, 3)          0                                            
____________________________________________________________________________________________________
action_input (InputLayer)        (None, 1)             0                                            
____________________________________________________________________________________________________
flatten_2 (Flatten)              (None, 3)             0           observation_input[0][0]          
____________________________________________________________________________________________________
concatenate_1 (Concatenate)      (None, 4)             0           action_input[0][0]               
                                                                   flatten_2[0][0]                  
____________________________________________________________________________________________________
dense_5 (Dense)                  (None, 32)            160         concatenate_1[0][0]              
____________________________________________________________________________________________________
activation_5 (Activation)        (None, 32L)           0           dense_5[0][0]                    
____________________________________________________________________________________________________
dense_6 (Dense)                  (None, 32)            1056        activation_5[0][0]               
____________________________________________________________________________________________________
activation_6 (Activation)        (None, 32L)           0           dense_6[0][0]                    
____________________________________________________________________________________________________
dense_7 (Dense)                  (None, 32)            1056        activation_6[0][0]               
____________________________________________________________________________________________________
activation_7 (Activation)        (None, 32L)           0           dense_7[0][0]                    
____________________________________________________________________________________________________
dense_8 (Dense)                  (None, 1)             33          activation_7[0][0]               
____________________________________________________________________________________________________
activation_8 (Activation)        (None, 1)             0           dense_8[0][0]                    
====================================================================================================
Total params: 2,305
Trainable params: 2,305
Non-trainable params: 0
____________________________________________________________________________________________________
None
Traceback (most recent call last):
  File "ddpg_pendulum.py", line 62, in <module>
    agent.compile(Adam(lr=.001, clipnorm=1.), metrics=['mae'])
  File "/Users/fak/.virtualenvs/control/lib/python2.7/site-packages/rl/agents/ddpg.py", line 153, in compile
    self.actor_train_fn = K.function(critic_inputs, [self.actor(critic_inputs)], updates=updates)
  File "/Users/fak/.virtualenvs/control/lib/python2.7/site-packages/plaidml/keras/backend.py", line 735, in function
    return _Function(inputs, outputs, updates, name)
  File "/Users/fak/.virtualenvs/control/lib/python2.7/site-packages/plaidml/keras/backend.py", line 126, in __init__
    name=name)
  File "/Users/fak/.virtualenvs/control/lib/python2.7/site-packages/plaidml/tile.py", line 1049, in compose
    bindings.insert(op, op.bind(bindings))
  File "/Users/fak/.virtualenvs/control/lib/python2.7/site-packages/plaidml/tile.py", line 279, in bind
    applier.add_input(input_name, input_value.bind(bindings))
  File "/Users/fak/.virtualenvs/control/lib/python2.7/site-packages/plaidml/__init__.py", line 1366, in add_input
    _lib().plaidml_apply_add_input(self, name.encode(), _as_plaidml_var(value))
  File "/Users/fak/.virtualenvs/control/lib/python2.7/site-packages/plaidml/__init__.py", line 716, in _check_err
    self.raise_last_status()
  File "/Users/fak/.virtualenvs/control/lib/python2.7/site-packages/plaidml/library.py", line 131, in raise_last_status
    raise self.last_status()
plaidml.exceptions.Unknown: Applying function, tensor with mismatching dimensionality: I, expected=1, got=2