refactor(trainer): remove deprecated light-megatron backend path#585
Merged
Xiaoming-AMD merged 3 commits intomainfrom Mar 23, 2026
Merged
refactor(trainer): remove deprecated light-megatron backend path#585Xiaoming-AMD merged 3 commits intomainfrom
Xiaoming-AMD merged 3 commits intomainfrom
Conversation
- Update third_party/Megatron-LM to core_v0.16.0.
- Adapt muon optimizer patch signatures/call paths to Megatron 0.16 (config_overrides, optimizer arg forwarding).
- Relax turbo TE spec provider gate to <0.17 for DeepSeek compatibility on v0.16.
- Fix DDP/FSDP grad-sync compatibility by allowing _BaseDataParallel.{start,finish}_grad_sync to accept extra kwargs (e.g. force_all_reduce).
f8a4952 to
ed4533e
Compare
Xiaoming-AMD
approved these changes
Mar 23, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Remove the legacy light-megatron trainer implementation and clean up all related framework aliases. This change deletes the lightmegatron trainer modules and removes light-megatron routing from parser and hook dispatchers (train pretrain and projection performance), so framework resolution now follows megatron directly without a compatibility alias.