Checkpoint connector bugfixes #10647
Open
Conversation
This file contains 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
jstjohn
commented
Sep 26, 2024
jstjohn
force-pushed
the
jstjohn/nemo_checkpoint_connector_fixes
branch
3 times, most recently
from
eb4bca5
to
6069101
Compare
Signed-off-by: John St John <jstjohn@nvidia.com>
jstjohn
force-pushed
the
jstjohn/nemo_checkpoint_connector_fixes
branch
from
6069101
to
a4ad2d7
Compare
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| @@ -170,16 +171,20 @@ | |||
| trainer (pl.Trainer): The trainer with the strategy to save the model. | |||
| dump_io (bool): If True, the IO configuration will be saved to the output path. | |||
| """ | |||
| # Import here to avoid circular import | |||
| from nemo.lightning.pytorch.callbacks.model_checkpoint import ModelCheckpoint | |||
Check notice
Code scanning / CodeQL
Cyclic import Note
There was a problem hiding this comment.
@jstjohn any way to avoid this?
akoumpa
reviewed
Sep 27, 2024
| ) | ||
|
|
||
| _trainer.state.fn = TrainerFn.FITTING # needed for proper save. |
There was a problem hiding this comment.
@jstjohn please add what do you mean here, what would be missing if fn was not set to fitting?
akoumpa
reviewed
Sep 27, 2024
| @@ -277,7 +278,7 @@ def on_train_end(self, trainer, pl_module): | |||
| else: | |||
| super()._save_last_checkpoint(trainer, monitor_candidates) | |||
| if self.save_context_on_train_end and not self.always_save_context and is_global_rank_zero(): | |||
| TrainerContext.from_trainer(trainer).io_dump(ckpt_to_dir(self.last_model_path) / "context") | |||
| TrainerContext.from_trainer(trainer).io_dump(ckpt_to_dir(self.last_model_path) / self.CONTEXT_PATH) | |||
There was a problem hiding this comment.
@jstjohn ideally I'd like to avoid changing the checkpoint structure but if we have to do it, let's add a comment giving an example for the use-case and cherry-pick this PR to make it to the 24.09 release.
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.
What does this PR do ?
Get checkpoint connector working for bionemo (see https://github.com/NVIDIA/bionemo-fw-ea/pull/180)
Changelog
/weightsand/contextsubdirectory scheme so checkpoint loaders work properly with checkpoints created by this method.