Dopd opsd routing#6237
Draft
ucalyptus wants to merge 3 commits into
Draft
Conversation
Ports the OPSD trainer (huggingface#5990) as a base and adds a distillation_mode="dopd" option implementing the per-token routing from DOPD: Dual On-policy Distillation (arXiv:2606.30626). Each token is routed by advantage gap and teacher/student confidence into one of four regimes: light top-k reverse-KL, full-vocab JSD, a light stop-gradient student-consistency nudge, or a weak stop-gradient self-regularization fallback for ambiguous tokens. Adds pure-function CPU tests (no model/GPU) verifying routing correctness, mask exhaustiveness, and gradient behavior of the stop-gradient regimes.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
question for myself: https://github.com/huggingface/trl/pull/5990/changes does this PR need to be merged before i can proceed?