Draft
Conversation
Adds a new KL penalty mode that modifies advantages after group
normalization with a batch-centered relative KL signal:
advantage += coef * (avg_batch_KL - token_KL)
Tokens drifting more than the batch average from the reference get
penalized; tokens drifting less get a bonus. The sum is approximately
zero (variance-reducing). Avoids the gradient bias of KL-in-loss.
Configure with use_kl_in_advantages: true, kl_advantages_coef: 0.01.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
advantage += coef * (avg_batch_KL - token_KL)use_kl_in_advantages: true,kl_advantages_coef: 0.01Test plan
🤖 Generated with Claude Code