Fix ADAMW, and track the loss #46
Closed
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
ToucheSir
reviewed
Jan 30, 2022
| @@ -405,7 +405,7 @@ weight decay regularization. | |||
| (no need to change default) | |||
| """ | |||
| ADAMW(η = 1f-3, β = (9f-1, 9.99f-1), γ = 0, ϵ = eps(typeof(η))) = | |||
| OptimiserChain(ADAM{typeof(η)}(η, β, ϵ), WeightDecay(γ)) | |||
| OptimiserChain(ADAM{typeof(η)}(1, β, ϵ), WeightDecay{typeof(η)}(γ), Descent(η)) | |||
There was a problem hiding this comment.
Sorry, I just remembered a previous comment (maybe on Slack? Can't find it now) which noted that our implementation did not match the paper:
.
TL;DR is that (possibly because of an odd choice of variable names in the paper, they use α instead of η for the learning rate and η for something else), the "fixed" version is wrong and the original is correct. Apologies for the noise.
There was a problem hiding this comment.
oh damn, Brian is right, FluxML/Flux.jl#1612 should be reverted, I got confused by the nomenclature
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Outdated suggestions cannot be applied.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.
This makes the same changes as FluxML/Flux.jl#1612, xref #38.
It also alters the tests to keep track of the loss. Although it does not fix any of the convergence problems.