def _SoftmaxCrossEntropyWithLogitsGrad(op, grad_0, _):
  """Gradient function for SoftmaxCrossEntropyWithLogits."""
  # grad_0 is the backprop for cost, and we multiply it with the gradients
  # (which is output[1])
  # There is no gradient for the labels
  #
  # Currently there is no way to take the second derivative of this op
  # due to the fused implementation's interaction with tf.gradients(),
  # so we make sure we prevent silently incorrect results by raising
  # an error if the second derivative is requested via prevent_gradient.
  softmax_grad_without_gradient = array_ops.prevent_gradient(op.outputs[1])
  return _BroadcastMul(grad_0, softmax_grad_without_gradient), None

def _CTCLossGrad(op, grad_loss, _):
  """The derivative provided by CTC Loss.

  Args:
     op: the CTCLoss op.
     grad_loss: The backprop for cost.

  Returns:
     The CTC Loss gradient.
  """
  # Outputs are: loss, grad
  #
  # Currently there is no way to take the second derivative of this op
  # due to the fused implementation's interaction with tf.gradients(),
  # so we make sure we prevent silently incorrect results by raising
  # an error if the second derivative is requested via prevent_gradient.
  grad_without_gradient = array_ops.prevent_gradient(op.outputs[1])
  # Return gradient for inputs and None for
  # labels_indices, labels_values and sequence_length
  return [_BroadcastMul(grad_loss, grad_without_gradient), None, None, None]

 def testPreventGradient(self):
   with ops.Graph().as_default():
     inp = constant(1.0, shape=[100, 32], name="in")
     out = array_ops.prevent_gradient(inp)
     with self.assertRaisesRegexp(LookupError, "explicitly disabled"):
       _ = gradients.gradients(out, inp)

 def _FuzzyCTCLossGrad(op, grad_loss, _):
     grad_without_gradient = array_ops.prevent_gradient(
         op.outputs[1], message="Currently there is no way to take the second "
                                " derivative of ctc_loss due to the fused implementation's interaction "
                                " with tf.gradients()")
     return [_BroadcastMul(tf.expand_dims(grad_loss, -1), grad_without_gradient), None, None, None]