Recover missing state after collaborator restart #1268
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
Signed-off-by: Lerer, Eran <[email protected]>
cloudnoize
force-pushed
the
elerer/collaborator_restart.2
branch
from
47c2163 to
0222fd7
Compare
psfoley
reviewed
Jan 14, 2025
| else: | ||
| batch_size = 1 | ||
| # evaluation needed before metrics can be resolved | ||
| self.model.evaluate(self.data_loader.get_valid_loader(batch_size), verbose=1) |
There was a problem hiding this comment.
For other reviewers reference, this is needed for the scenario when the collaborator dies between the aggregated_model_validation and train task. If the collaborator restarts it will immediately be assigned train, and will not be able to resolve the necessary metrics. The names of needed metrics can be resolved by running the evaluate function on the model, but these metric values will not be sent to the aggregator.
psfoley
approved these changes
Jan 14, 2025
rahulga1
approved these changes
Jan 15, 2025
tanwarsh
approved these changes
Jan 15, 2025
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Outdated suggestions cannot be applied.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.
After restart the collaborator gets a list of tasks that it needs to finish as part of the current round, that list might be partial as it already executed some tasks prior the restart, previous tasks might have generated state that is needed for the future tasks of this round.
This PR recover or initialized the state of previous tasks enabling the collaborator to resume the execution of the current round.
https://jira.devtools.intel.com/browse/FEDAI-1229