TensorArray TensorArray_1_0: Could not read from TensorArray index 0 because it has not yet been written to

Question

I don't know how to use tensorarray. Here's the code. What's the bug in that?

import tensorflow as tf

aI=tf.TensorArray(tf.int32, 2)
aO=tf.TensorArray(tf.int32, 2)
aI=aI.unpack([[1,2],[1,2]])
def body(i,aI,aO):
    aO.write(i, aI.read(i)+1)
    return (i+1, aI, aO)
cond=lambda i, *_ : i<2
_, _, aO=tf.while_loop(cond, body, [0,aI,aO])
r=aO.pack()
with tf.Session() as sess:
    res=sess.run(r)
    print('done!')

Ziemo · Accepted Answer · 2018-04-11 18:14:37Z

8

I solved. It seems that inside the body of while_loop we should reassign the old TensorArray aO with the returned value of aO.write():

def body(i,aI,aO):
    aO=aO.write(i, aI.read(i)+1)
    return (i+1, aI, aO)

Whole code:

import tensorflow as tf

aI=tf.TensorArray(tf.int32, 2)
aO=tf.TensorArray(tf.int32, 2)
aI=aI.unpack([[1,2],[1,2]])
def body(i,aI,aO):
    aO=aO.write(i, aI.read(i)+1)
    return (i+1, aI, aO)
cond=lambda i, *_ : i<2
_, _, aO=tf.while_loop(cond, body, [0,aI,aO])
r=aO.pack()
with tf.Session() as sess:
    res=sess.run(r)
    print('done!')

edited Apr 11, 2018 at 18:14

Ziemo

9919 silver badges28 bronze badges

answered Feb 9, 2017 at 15:10

mxmxlwlw

3272 silver badges17 bronze badges

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

1 Comment

Ziemo Over a year ago

Note that currently (TF 1.4/1.5) TensorArray methods changed: unpack() -> unstack() and pack() -> stack()

Collectives™ on Stack Overflow

TensorArray TensorArray_1_0: Could not read from TensorArray index 0 because it has not yet been written to

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related