Skip to content

Add addons to top-level modules #718

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

Closed
fsx950223 opened this issue Nov 25, 2019 · 3 comments
Closed

Add addons to top-level modules #718

fsx950223 opened this issue Nov 25, 2019 · 3 comments

Comments

@fsx950223
Copy link
Member

Describe the feature and the current behavior/state.
Add addons to top-level modules just like estimator for convenient
Relevant information

  • Are you willing to contribute it (yes/no):
  • Are you willing to maintain it going forward? (yes/no):
  • Is there a relevant academic paper? (if so, where):
  • Is there already an implementation in another framework? (if so, where):
  • Was it part of tf.contrib? (if so, where):

Which API type would this fall under (layer, metric, optimizer, etc.)
tensorflow.addons
Who will benefit with this feature?
addons users
Any other info.

@seanpmorgan
Copy link
Member

Hi @fsx950223 could you further explain what you're requesting? Are you asking for Addons to be a dependency of TF-Core just as Estimator is?

If so it is highly unlikely to ever be a dependency as there is a major push for modular tensorflow:
tensorflow/community#77

@fsx950223
Copy link
Member Author

fsx950223 commented Nov 26, 2019

Hi @fsx950223 could you further explain what you're requesting? Are you asking for Addons to be a dependency of TF-Core just as Estimator is?

If so it is highly unlikely to ever be a dependency as there is a major push for modular tensorflow:
tensorflow/community#77

For example:
If I install tensorflow_addons, I could use addons by

import tensorflow as tf
image=tf.addons.image.rotate(image)

rather than I have to use

import tensorflow_addons as tfa
image=tfa.image.rotate(image)

Tensorflow_addons dynampic mounts to tensorflow.

@seanpmorgan
Copy link
Member

Understood. While I agree this would be a nice change for the end-user; that strategy has created a bit of a tangled mess for packaging TF. See this new RFC:
tensorflow/community#182

It creates a circular dependency because TFA depends on TF and TF depends on TFA. Going forward it looks as though every component of the TF ecosystem will be it's own import (see RFC)

Closing, but thanks for the suggestion and we're always looking for end-user improvements!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants