Repository navigation
Conversation
This branch has not been deployed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Noam, warmup and cosine schedules change the actual adaptive update rate while decoupled weight decay still uses the group maximum learning rate. Pass the current step rate only to decay, preserving group schedule parameters and the existing three-argument decay extension API.
Independent Adam recurrences check moments and parameter trajectories across schedules, both update formulas, coupled/absolute controls, and dynamic rate hooks with parameter groups and missing gradients. All four optimizer overrides are fixed as one scheduling defect.
Validation: the same 32-case regression file gives 14 failures and 18 passes on upstream, then 32 passes with this change. Tested with Python 3.12.10, PyTorch 2.14.1 CPU, pytest 9.1.1 and official labml 0.4.168. No skips or collection errors. The recorded upstream tree has no formal original unit suite; GPU and full-model training are untested.