JuliaDiff / JuliaDiff/ChainRules.jl
rrule for multi-argument product
Nobody has claimed this yet.
- Dominant language
- Julia
- Stars
- 475
- Forks
- 98
- PR merge metrics
- No merged PRs in 30d
Description
From dfdx/Yota.jl#93:
A = rand(100, 100)
x = rand(100)
rrule(*, x', A, x) # ==> nothing
It's possible to binarize the operation on the AD engine side, but having a single rule that works automagically works for all ADs seems to be a better approach.
Putting it into my own ToDo list, but if somebody feels like taking over, please do.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start from the reported rrule(*, x', A, x) example and inspect the existing rrule definitions for multi-argument products in ChainRules.jl. Confirm how the rule should behave for the shown matrix and vector operands, then verify that one rule works across automatic-differentiation systems.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- julia
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100