Make BatchNorm twice-differentiable #2255
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Solves the BatchNorm not being twice-differentiable problem first reported at https://discourse.julialang.org/t/compilation-error-in-zygote-flux-and-cuda-interaction/92571 and made into issue FluxML#2154.
|
Nice! Can you add a cuda test? |
ToucheSir
requested changes
May 9, 2023
| function batchnorm_pullback(Δ) | ||
| grad = ∇batchnorm(g, b, x, unthunk(Δ), running_mean, running_var, momentum; kw...) | ||
| (NoTangent(), grad..., NoTangent(), NoTangent(), NoTangent()) | ||
| function batchnorm_pullback(Δ, σ²Δ) |
There was a problem hiding this comment.
Where is this second parameter coming from? You can see on the CI log that this raises a MethodError. The pullback should only take one argument (Δ) and destructure it from a tuple into its component parts if need be.
|
I think we'd also want a test for double-differentiability on GPU. Make sure you're looking at Buildkite and not GitHub actions because the latter doesn't run any of the GPU code :) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Solves the 'BatchNorm not being twice-differentiable' problem first reported at https://discourse.julialang.org/t/compilation-error-in-zygote-flux-and-cuda-interaction/92571 and made into issue #2154. Relevant documentation update would be made at https://fluxml.ai/Flux.jl/stable/models/layers/#Flux.BatchNorm, however this is unnecessary.
PR Checklist