Open
Conversation
sfc-gh-lmerrick
requested review from
sfc-gh-caxu,
sfc-gh-jrasley,
sfc-gh-mwyatt,
sfc-gh-sbekman,
sfc-gh-srajbhandari and
sfc-gh-zhyao
as code owners
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Warmup-stable-decay (WSD) is a common learning rate schedule which is parameterized slightly differently than the default linear decay schedule (canonically by step count rather than fraction of training steps, at least in
transformers). This PR adds support for WSD to the scheduler factory, moving the warmup ratio out of the base scheduler config and into theHFSchedulerConfigclass.It may also make sense to rename
HFScheduler*configs and classes because in truth the config and factory are specific to linear schedules and do not have the right parameters to support other schedules like WSD.