Training Error

[isadoramsouza]

I am trying to reproduce you implementation, but i got this error:

LookupError: No gradient defined for operation 'pool4' (op type: MaxPoolWithArgmax)

How can I solve?

thanks

[okapi1125]

I also have got the same error.
How can I solve it?
(In the past recent version, this problem didn't occur?)
Thank you.

[isadoramsouza]

add in model.py:

from tensorflow.python.framework import ops
from tensorflow.python.ops import gen_nn_ops
@ops.RegisterGradient("MaxPoolWithArgmax")
def _MaxPoolWithArgmaxGrad(op, grad, unused_argmax_grad):
  return gen_nn_ops._max_pool_grad(op.inputs[0],
                                   op.outputs[0],
                                   grad,
                                   op.get_attr("ksize"),
                                   op.get_attr("strides"),
                                   padding=op.get_attr("padding"),
                                   data_format='NHWC')

[boyob]

add in model.py:

from tensorflow.python.framework import ops
from tensorflow.python.ops import gen_nn_ops
@ops.RegisterGradient("MaxPoolWithArgmax")
def _MaxPoolWithArgmaxGrad(op, grad, unused_argmax_grad):
  return gen_nn_ops._max_pool_grad(op.inputs[0],
                                   op.outputs[0],
                                   grad,
                                   op.get_attr("ksize"),
                                   op.get_attr("strides"),
                                   padding=op.get_attr("padding"),
                                   data_format='NHWC')

I did as you said, but I got this error, what should i do?
InvalidArgumentError (see above for traceback): No OpKernel was registered to support Op 'MaxPoolWithArgmax' with these attrs. Registered devices: [CPU], Registered kernels:
device='GPU'; Targmax in [DT_INT64]; T in [DT_HALF]
device='GPU'; Targmax in [DT_INT64]; T in [DT_FLOAT]

 [[Node: pool1 = MaxPoolWithArgmax[T=DT_FLOAT, Targmax=DT_INT64, ksize=[1, 2, 2, 1], padding="SAME", strides=[1, 2, 2, 1]](conv1/Relu)]]