fix: checkpoint saving with distributed optimizer + overlap param gather - #949
Merged
terrykong merged 1 commit intoAug 21, 2025
Conversation
âĶ for checkpoint saving Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com>
ananthsub
added
t-mcore
CI:L2
ashors1
approved these changes
Aug 20, 2025
ashors1
left a comment
ashors1
left a comment
Contributor
There was a problem hiding this comment.
Thank you for the fix @ananthsub!
terrykong
added
CI:docs
terrykong
enabled auto-merge
August 20, 2025 16:27
terrykong
added this pull request to the merge queue
Aug 20, 2025
ananthsub
deleted the
ananthsub/fix-ckpt-save-dist-opt-overlap-param-gather
branch
August 22, 2025 05:57
jveronvialard
pushed a commit
that referenced
this pull request
Aug 27, 2025
âĶher (#949) Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> Signed-off-by: Julien Veron Vialard <jveronvialar@nvidia.com>
soodoshll
pushed a commit
to soodoshll/RL
that referenced
this pull request
Aug 28, 2025
âĶher (NVIDIA-NeMo#949) Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> Signed-off-by: Qidong Su <qidongs@nvidia.com>
soodoshll
pushed a commit
to soodoshll/RL
that referenced
this pull request
Sep 4, 2025
âĶher (NVIDIA-NeMo#949) Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> Signed-off-by: Qidong Su <qidongs@nvidia.com>
PrinsYin
pushed a commit
to PrinsYin/RL
that referenced
this pull request
Nov 30, 2025
âĶher (NVIDIA-NeMo#949) Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does this PR do ?
Fixes checkpoint saving when distributed optimizer + overlap param gather are used together. We need to disable/re-enable the pre-forward hooks around checkpoint saving as done here to force the parameter sync to ensure consistency before checkpoint saving: https://github.com/NVIDIA-NeMo/Megatron-Bridge/blob/742aaf84d9a3c4108b675968cc5b5ed59e4e4921/src/megatron/bridge/training/train.py#L787-L805
Currently state: the loss spikes on resume without this:
With this PR:
Issues
List issues that this PR closes (syntax):
Usage
# Add a code snippet demonstrating how to use thisBefore your PR is "Ready for review"
Pre checks:
Additional Information