Creating an array of tensors in tensorflow?

Question

Suppose I have two tensor variables each of size 2x4:

v1 = tf.logical_and(a1, b1)
v2 = tf.logical_and(a2, b2)

Instead, I want to store these in an array called v which is of size 2x2x4. How do I do this in Tensorflow? The idea would be something like this:

for i in range(2):
  v[i] = tf.logical_and(a[i],b[i])

How do I initialize v? I tried initializing v as a numpy array which did not work. I also tried initializing it as a tensorflow variable, ie. tf.Variable(tf.zeros([2])) but that does not work either.

Note, a and b are dynamic inputs, ie. they are tf.placeholder variables.

MMN · Accepted Answer · 2016-10-15 23:45:09Z

2

tf.pack() is probably what you are looking for.

answered Oct 15, 2016 at 23:45

MMN

6766 silver badges7 bronze badges

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

2 Comments

jlcv Over a year ago

How would I store each tensor v[i] because I want to run it through a for loop and there might be more than 2 tensors.

MMN Over a year ago

put the v[i]s in a normal list, pass the list as the first arg to tf.pack(myListOfTensors)

Matthew Feickert · Accepted Answer · 2017-11-19 13:16:51Z

0

As a potentially relevant update, in TensorFlow 1.0 tf.pack() was renamed to tf.stack().

answered Nov 19, 2017 at 13:16

Matthew Feickert

9079 silver badges28 bronze badges

Collectives™ on Stack Overflow

Creating an array of tensors in tensorflow?

2 Answers 2

2 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related