Initializing tf.Variable from a Tensor Object (not a tf.Variable)

Question

I'm trying to initialize a tf.Variable from a tensor object. Normally, from another tf.Variable you simply just used the initialized_value() but if you make a tensor object this does not work:

a = tf.constant([4,5,6])
b = tf.Variable(a.initialized_value())

I have tried b = tf.Variable(a) but upon initialize_all_variables() being ran, tensorflow asks for a placeholder to be filled which is not correct.

The reason why I'm doing this in the first place is that @mrry has suggested to place a tf.Variable into the params argument in tf.gather to help speed things up.

Thanks for your feedback!

Yaroslav Bulatov · Accepted Answer · 2016-05-07 20:29:37Z

2

This seems to work

a = tf.constant([4,5,6])
b = tf.Variable(a)
sess = tf.InteractiveSession()
sess.run(tf.initialize_all_variables())
print b.eval()

answered May 7, 2016 at 20:29

Yaroslav Bulatov

58.1k23 gold badges147 silver badges210 bronze badges

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

1 Comment

LeavesBreathe Over a year ago

I ended up just using tf.assign after reading your suggestion. I think there was something wrong with how I was initializing the variables. Thanks!

Collectives™ on Stack Overflow

Initializing tf.Variable from a Tensor Object (not a tf.Variable)

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related