Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Different behavior tf.contrib.data.Dataset.unbatch() TF ver. 1.2.1 and ver 1.3.0-rc0 #11666

Closed
vbystricky opened this issue Jul 21, 2017 · 2 comments
Closed

Different behavior tf.contrib.data.Dataset.unbatch() TF ver. 1.2.1 and ver 1.3.0-rc0 #11666

vbystricky opened this issue Jul 21, 2017 · 2 comments
Assignees
@ebrevdo

Comments

@vbystricky
Copy link

vbystricky commented Jul 21, 2017 •
edited

Code for reproduce problem:

import tensorflow as tf
print(tf.__version__)
dataset = (tf.contrib.data.Dataset.range(10)
           .batch(5)
           .unbatch())
iterator = dataset.make_initializable_iterator()
print(iterator.get_next())

results in TF ver. 1.2.1:

1.2.1                                                  
Tensor("IteratorGetNext:0", shape=(), dtype=int64)

results in TF ver. 1.3.0-rc0:

1.3.0-rc0
(<tf.Tensor 'IteratorGetNext:0' shape=() dtype=int64>,)
@mrry
Copy link
Contributor

mrry commented Jul 21, 2017

@ebrevdo I've tried and failed to track down the commit that modified the behavior of Dataset.unbatch(), but it looks like we're inappropriately converting single-component datasets to tupled datasets. Can you please take a look?

@ebrevdo
Copy link
Contributor

ebrevdo commented Jul 21, 2017 via email

@caisq caisq closed this as completed in 307a2fc Jul 24, 2017
av8ramit pushed a commit to av8ramit/tensorflow that referenced this issue Jul 25, 2017
@ebrevdo
[tf.contrib.data] Fix unbatch scalar vs. tuple handling.
672c96c 
Fixes tensorflow#11666.

PiperOrigin-RevId: 162776187
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@ebrevdo ebrevdo

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
3 participants
@mrry @ebrevdo @vbystricky