support loading checkpoints from a given step - #1040
Merged
Conversation
Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com>
Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com>
Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com>
Contributor
Author
|
/ok to test 2d65c06 |
Contributor
Author
|
/ok to test deaaa85 |
chtruong814
pushed a commit
that referenced
this pull request
Oct 23, 2025
* support loading checkpoints from a given step Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> * support loading checkpoints from a given step Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> * add config test Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> * fix tests Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> --------- Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> Signed-off-by: NeMo Bot <nemo-bot@nvidia.com>
ananthsub
added a commit
that referenced
this pull request
Oct 23, 2025
* support loading checkpoints from a given step * support loading checkpoints from a given step * add config test * fix tests --------- Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> Signed-off-by: NeMo Bot <nemo-bot@nvidia.com> Co-authored-by: Ananth Subramaniam <ansubramania@nvidia.com>
paul-gibbons
pushed a commit
to paul-gibbons/Megatron-Bridge
that referenced
this pull request
Oct 29, 2025
* support loading checkpoints from a given step Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> * support loading checkpoints from a given step Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> * add config test Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> * fix tests Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> --------- Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> Signed-off-by: Paul Gibbons <pgibbons@nvidia.com>
nv-mollys
pushed a commit
that referenced
this pull request
Oct 31, 2025
* support loading checkpoints from a given step Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> * support loading checkpoints from a given step Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> * add config test Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> * fix tests Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> --------- Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> Signed-off-by: mollys <mollys@mollys.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
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.
Add support for
ckpt_stepin the checkpoint config.This allows users to specify a specific checkpoint directory to resume from based on the iteration. This is an opt-in feature. The default is still to resume from the latest checkpoint saved.
Special handling is needed for PEFT where there are 2 checkpoint loads:
pretrained_checkpointloaddirectory in case of resuming the adapter states within a training run.ckpt_stepis intended to only take effect forloaddirectories aspretrained_checkpointshould be unchanged over the course of training.