FluxML/Zygote.jl

gradient of keyword argument?

Open

#446 opened on Jan 6, 2020

View on GitHub
 (10 comments) (0 reactions) (0 assignees)Julia (1,568 stars) (220 forks)batch import
discussionhelp wantedperformance

Description

Is there any ways to compute the backward gradient of a keyword argument? For example, the var function can pass a pre-computed mean value as a keyword argument, but right now it seems like we can't get the gradient of it. This will have an effect on something like Flux.normalise.

Contributor guide

Tech stack
None
Domain
machine learning
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
3-5 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
needs maintainer response
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
JuliaAD basics
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
25
Research direction
Review the discussion in issue #446 to understand the request for gradient of keyword arguments. Examine the Zygote.jl codebase, particularly the parts handling `var` and `normalise`, to see where keyword arguments are processed. Check for any existing PRs or issues that may touch upon this feature. Propose a plan to extend the gradient computation to support keyword arguments while maintaining backward compatibility.