Checkpointing and retrieval #13
Merged
Conversation
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
mkrima
reviewed
Apr 15, 2024
mkrima
reviewed
Apr 15, 2024
mkrima
reviewed
Apr 16, 2024
mkrima
reviewed
Apr 16, 2024
|
LGTM |
|
quick question: without having gone through the code in detail, does this also make sure the data sampling is deterministic (dataloader state is restored)? what i mean is, does the model see the same data when loading from a checkpoint vs when doing a full training run |
|
Good point, I have added those modifications in another PR |
|
Sorry I think I merged this too quickly. I forgot that we switched to using dataloader in this repo. The PR resets the rng state but our dataloader has its own sampler with its own generator. so we have to set the rng state for that too. I can do that in another PR in our hacakthon tomorrow: |
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Outdated suggestions cannot be applied.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.
Implemented checkpoint and retrieval for model, scheduler, and optimizer state_dict as well as random generator states for reproducibility.