Share one implementation between adam and adamw - #86
Merged
jessegrabowski merged 2 commits intoAug 12, 2026
Conversation
The step counter, moment buffers, bias corrections and amsgrad handling were duplicated verbatim; pymc-devs#73 was a bug caused by exactly that pair drifting. The two arithmetic groupings are kept as separate branches rather than folded, since folding moves a third of float32 adam results in the last bits.
Seven positional arguments, three of them adjacent floats, so transposing beta1 and beta2 at a call site would have been silent.
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #86 +/- ##
==========================================
- Coverage 95.48% 95.46% -0.03%
==========================================
Files 45 45
Lines 1818 1807 -11
==========================================
- Hits 1736 1725 -11
Misses 82 82 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The step counter, moment buffers, bias corrections and amsgrad handling were duplicated verbatim between the two rules, and #73 was a bug caused by exactly that pair drifting apart. Now one private helper, with the state namespace as an argument so the two keep separate buffers.
The two arithmetic groupings stay as separate branches rather than folded into one. They agree in exact arithmetic and differ in the last bits — folding them moves about a third of float32 adam results — so this is bit-exact against the old code for adam, adamw, either with amsgrad, and adamw with a mask.