IAL32
November 25, 2021, 8:16am
1
Hi, community!
For research purposes, I started looking into tf.autodiff.ForwardProp. However, I have also noticed that supporting batches of tangents inside the function are not yet fully implemented . I saw that there has been some awesome work on this topic from a GSOC student last year, and I would like to bring it to an end or bring it forward at least!
However, it is worth noting that I am also a student, and I would need some support on the matter.
Thanks!
Bhack
November 26, 2021, 12:52am
2
Welcome to the TF forum,
/cc @markdaoust
IAL32
November 26, 2021, 12:11pm
4
I have reached out to Allen directly, and unfortunately he said that he is not working on TF anymore.
IAL32
December 8, 2021, 11:02pm
7
Another friendly bump.
The Pytorch community has been doing very fast steps lately on this very topic , and perhaps there is something that we could learn from their approach.
Bhack
December 9, 2021, 7:49pm
9
@markdaoust Do you know someone in this are that could join in the thread?
IAL32
January 15, 2022, 11:06am
11
Hi! @Bhack @markdaoust any new developments?
Bhack
January 15, 2022, 1:38pm
12
I don’t see any new team member Active on this file:
Do you have an idea of the scope of the work that’s necessary here?
There is a process for design-reviews in TensorFlow:
# TensorFlow Request for Comments (TF-RFC)
The purpose of a TensorFlow RFC is to engage the TensorFlow community in
development, by getting feedback from stakeholders and experts, and
communicating design changes broadly.
## Who is involved?
Any **community member** may help by providing feedback on whether the RFC will
meet their needs.
An **RFC author** is one or more community member who writes an RFC and is
committed to championing it through the process.
An **RFC sponsor** is any maintainer who sponsors the RFC and will shepherd it
through the RFC review process.
A **review committee** is a group of maintainers who have the responsibility of
recommending the adoption of the RFC.
show original
# tf-design-reviews criteria
## Overview
The TensorFlow team has run internal and public design reviews for a while
now. This document tries to capture what type of questions get asked and
concerns get addressed in TF design reviews. It is intended to be used by design
authors as a way of spot checking whether a design review will be useful for
them and by design sponsors as a way of making sure a design proposal clears the
bar for review (ideally every topic in this document should be addressed by the
design proposal for it to be considered).
The main goal of tf-design-reviews is to socialize big changes to TF, document
them, and ensure all stakeholders get a chance to comment on planned
improvements before they’re implemented. Any time a change is made to TF that
will affect multiple aspects of its design or user interface, we should solicit
a design review. TF design reviews themselves are not binding: final approval
rests with whoever has the authority to approve the required code changes, and
the design review is a tool to get consensus and feedback on big changes before
actual approval.
show original
But I guess that without a sponsor from the core team you’re somewhat stuck at the “Recruit a sponsor from the maintainers of the project”.
Bhack
January 18, 2022, 3:34pm
15
Yes and I suppose the forum is the best place where we could collect an early feedback from a maintainer before preparing an RFC with its own overhead (/cc @thea ).
But as our CODOWNERS on github is partially maintained who is maintaining this component now? Can we mention him in this thread?
P.s. I’ve already made another ping on the github issue.
