[go: nahoru, domu]

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

C API gradient tape support #47748

Open
rnett opened this issue Mar 12, 2021 · 2 comments
Open

C API gradient tape support #47748

rnett opened this issue Mar 12, 2021 · 2 comments
Assignees
Labels
comp:apis Highlevel API related issues stat:awaiting tensorflower Status - Awaiting response from tensorflower type:feature Feature requests

Comments

@rnett
Copy link
Contributor
rnett commented Mar 12, 2021

System information

  • TensorFlow version: 2.4.1
  • Are you willing to contribute it: Probably not

Describe the feature and the current behavior/state.

Gradients are being ported to C++, with the tape API in c/eager/gradients.h.

However, there is no equivalent C API, and as per tensorflow/community#335 it does not seem there are current plans to add one.

I assume this is on the radar somewhere, but I don't see any tracking issues, so I'm making this as tracking (and a request).

Will this change the current api? How?

It will expose the existing gradient tape api to the C API. Gradient registration should be exposed as well to allow for registering custom gradients (it seems to be tape specific?).

Who will benefit with this feature?

It will enable eager mode to support gradients for projects using the C API bindings, including the official JVM bindings (which is what prompted this issue) and those that depend on it.

@rnett rnett added the type:feature Feature requests label Mar 12, 2021
@amahendrakar amahendrakar added the comp:apis Highlevel API related issues label Mar 12, 2021
@amahendrakar amahendrakar assigned ymodak and unassigned amahendrakar Mar 12, 2021
@ymodak ymodak added the stat:awaiting tensorflower Status - Awaiting response from tensorflower label Mar 12, 2021
@ymodak ymodak assigned saxenasaurabh and unassigned ymodak Mar 12, 2021
@bhack
Copy link
Contributor
bhack commented Apr 24, 2021

/cc @joanafilipa

@bhack
Copy link
Contributor
bhack commented Apr 24, 2021

For JVM, are they working directly with C++ interfaces tensorflow/java#283?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
comp:apis Highlevel API related issues stat:awaiting tensorflower Status - Awaiting response from tensorflower type:feature Feature requests
Projects
None yet
Development

No branches or pull requests

6 participants