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.

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

Feature request: Allow for custom hooks in Slim's evaluate_once to support TFDBG #13444

Closed
rjbruin opened this issue Oct 2, 2017 · 2 comments
Assignees
Labels
stat:contribution welcome Status - Contributions welcome type:feature Feature requests

Comments

@rjbruin
Copy link

rjbruin commented Oct 2, 2017

System information

  • TensorFlow version: v1.3.0-rc2-20-g0787eee

Describe the problem

The evaluation functions provided by contrib.slim.python.slim are wrappers around methods from contrib.training.python.training. contrib.slim.python.slim.evaluation_loop provides argument hooks which attaches custom hooks to the evaluation loop. This can be used to hook the TFDBG debugger into evaluation. contrib.slim.python.slim.evaluation_once however does not provide the argument, even though the underlying contrib.training.python.training.evaluate_once does support the argument. The code to extend the hooks with custom hooks is already there in contrib.slim.python.slim.evaluation_loop but that fix somehow was not applied to contrib.slim.python.slim.evaluation_once.

The request is to implement the addition of custom hooks to contrib.slim.python.slim.evaluation_once so that TFDBG can be used with this method. This is a really easy fix that adds a lot of functionality.

@caisq caisq self-assigned this Oct 3, 2017
@caisq caisq added the type:feature Feature requests label Oct 3, 2017
@rjbruin
Copy link
Author

rjbruin commented Nov 8, 2017

@caisq Sorry to bother you, but could you give an indication of when you could get to this? It would really help me to have this in TensorFlow. :)

@caisq
Copy link
Contributor

caisq commented Nov 8, 2017

@rjbruin Sorry about the delay.

@sguada @rjbruin's FR makes sense to me. The change should be small and straightforward. I'll send you a change list soon.

@sguada sguada added the stat:contribution welcome Status - Contributions welcome label Nov 8, 2017
@yifeif yifeif closed this as completed in 1c79da7 Nov 10, 2017
aselle pushed a commit to aselle/tensorflow that referenced this issue Nov 11, 2017
to enable TFDBG debugging of slim.evaluation.evaluate_once()

Fixes: tensorflow#13444
PiperOrigin-RevId: 175101022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
stat:contribution welcome Status - Contributions welcome type:feature Feature requests
Projects
None yet
Development

No branches or pull requests

3 participants