Create a tensor by calling a function in a loop in TensorFlow

Question

I need to create a tensor by calling some function fn over two other tensors and indices in a loop as follows:

tensor = [[fn(tensor1, tensor2, i, j) for i in range(3)] for j in range(4)]

Not sure how to approach this problem. Use tf.map_fn somehow?

Uncertainty about the function of your fn. You can also try tf.while_loop besides tf.map_fn. — giser_yugang
– giser_yugang, Commented Mar 26, 2019 at 1:43

Mohan Radhakrishnan · Accepted Answer · 2019-03-26 12:58:22Z

1

So for your simple case your code will execute as it is.

import tensorflow as tf


sess = tf.Session()

a = tf.constant([1,2,3])
b = tf.constant([3,4,5,6])

def fn( tensor1, tensor2, i, j ):
   return tensor1[i] * tensor2[j]

tensor = [[fn(a, b, i, j) for i in range(3)] for j in range(4)]

init = tf.global_variables_initializer()
sess.run(init)
print (sess.run(tensor))

[[3, 6, 9], [4, 8, 12], [5, 10, 15], [6, 12, 18]]

answered Mar 26, 2019 at 12:58

Mohan Radhakrishnan

3,2285 gold badges35 silver badges46 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Create a tensor by calling a function in a loop in TensorFlow

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related