-
Notifications
You must be signed in to change notification settings - Fork 3.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Enable saving and loading stateful DataLoaders in Trainer #19361
Merged
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
awaelchli
added
feature
Is an improvement or enhancement
data handling
Generic data-related topic
trainer
fun
Staff contributions outside working hours - to differentiate from the "community" label
and removed
pl
Generic label for PyTorch Lightning package
labels
Jan 29, 2024
awaelchli
force-pushed
the
feature/stateful-dataloader
branch
from
January 29, 2024 02:45
1d75450
to
5317545
Compare
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
awaelchli
changed the title
Save
Enable saving and loading stateful DataLoaders in Trainer
Jan 30, 2024
state_dict
for stateful DataLoadersfor more information, see https://pre-commit.ci
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## master #19361 +/- ##
==========================================
- Coverage 84% 49% -35%
==========================================
Files 448 440 -8
Lines 37887 37762 -125
==========================================
- Hits 31649 18392 -13257
- Misses 6238 19370 +13132 |
Borda
approved these changes
Jan 30, 2024
tchaton
approved these changes
Jan 30, 2024
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice ;)
carmocca
reviewed
Jan 31, 2024
carmocca
reviewed
Jan 31, 2024
for more information, see https://pre-commit.ci
…/lightning into feature/stateful-dataloader
for more information, see https://pre-commit.ci
…/lightning into feature/stateful-dataloader
carmocca
approved these changes
Feb 1, 2024
mergify
bot
added
ready
PRs ready to be merged
and removed
has conflicts
ready
PRs ready to be merged
labels
Feb 1, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
data handling
Generic data-related topic
fault tolerance
feature
Is an improvement or enhancement
fun
Staff contributions outside working hours - to differentiate from the "community" label
pl
Generic label for PyTorch Lightning package
ready
PRs ready to be merged
trainer
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?
Saves the
state_dict
of training dataloaders into the checkpoint, and enables that state to be loaded when resuming from a checkpoint. An example of a stateful dataloader islightning.data.StreamingDataLoader
.The implementation collects the state of all iterables under the CombinedLoader that follow the stateful interface. The states are collected over the flattened view (CombinedLoader.flattened) and stored in a last. They get restored in the same manner via the flattened view. This means the number of iterables and the order in which they are given must be exactly the same as when the checkpoint was saved, otherwise the loading will fail.
Fixes #17105
Closes #17543
cc @Borda @justusschock @awaelchli @carmocca