final_lr_ratio #745
Merged
final_lr_ratio #745
Conversation
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
hamishivi
approved these changes
Jul 2, 2025
hamishivi
approved these changes
Jul 2, 2025
finbarrtimbers
pushed a commit
that referenced
this pull request
Jul 2, 2025
finbarrtimbers
pushed a commit
that referenced
this pull request
Jul 2, 2025
Merged
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Outdated suggestions cannot be applied.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.
Currently, the SFT learning rate is run all the way down to zero by default in
finetune.py. This PR adds a--final_lr_ratioflag which allows the LR to run only down to a value oflearning_rate * final_lr_ratioover the course of SFT. Qwen, for instance, follows this strategy of only decaying the lr down by a fixed factrion.Only currently implemented for the linear scheduler.