tensorflow - Write Custom Python-Based Gradient Function for an Operation? (without C++ Implementation) -

i'm trying write custom gradient function 'my_op' sake of example contains call tf.identity() (ideally, graph).

import tensorflow tf tensorflow.python.framework import function   def my_op_grad(x):     return [tf.sigmoid(x)]   @function.defun(a=tf.float32, python_grad_func=my_op_grad) def my_op(a):     return tf.identity(a)   = tf.variable(tf.constant([5., 4., 3., 2., 1.], dtype=tf.float32))  sess = tf.session() sess.run(tf.initialize_all_variables())  grad = tf.gradients(my_op(a), [a])[0]  result = sess.run(grad)  print(result)  sess.close() 

unfortunately following error:

traceback (most recent call last):   file "custom_op.py", line 19, in <module>     grad = tf.gradients(my_op(a), [a])[0]   file "/users/njk/tfm/lib/python3.5/site-packages/tensorflow/python/framework/function.py", line 528, in __call__     return call_function(self._definition, *args, **kwargs)   file "/users/njk/tfm/lib/python3.5/site-packages/tensorflow/python/framework/function.py", line 267, in call_function     compute_shapes=false)   file "/users/njk/tfm/lib/python3.5/site-packages/tensorflow/python/framework/ops.py", line 2285, in create_op     raise typeerror("input #%d not tensor: %s" % (idx, a)) typeerror: input #0 not tensor: <tensorflow.python.ops.variables.variable object @ 0x1080d2710> 

i know possible create custom c++ operation, in case need write custom gradient function can written in python using standard tensorflow operations, avoid writing unnecessary c++ code.

also, i'm using upstream version of tensorflow github.

note python_grad_func needs same interface ops.registergradient (https://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/framework/function.py#l349).

here modified code example:

def my_op_grad(op, grad): ### instead of my_op_grad(x)                                                       return tf.sigmoid(op.inputs[0])                                                @function.defun(a=tf.float32, python_grad_func=my_op_grad)                        def my_op(a):                                                                         return tf.identity(a)                                                          def main(unused_argv):                                                               = tf.variable(tf.constant([-5., 4., -3., 2., 1.], dtype=tf.float32))            sess = tf.session()                                                               sess.run(tf.initialize_all_variables())                                            = tf.identity(a) #workaround bug github.com/tensorflow/tensorflow/issues/3710    grad = tf.gradients(my_op(a), [a])[0]                                             result = sess.run(grad)                                                            print(result)                                                                      sess.close()      

output:

[ 0.00669286  0.98201376  0.04742587  0.88079709  0.7310586 ]