Skip to content
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

fix: Make sure that checkpoints in PBT are removed once they are no longer needed #600

Merged
merged 1 commit into from
Mar 27, 2023

Conversation

mseeger
Copy link
Collaborator

@mseeger mseeger commented Mar 15, 2023

… needed

This should solve the issue about checkpoints not being removed. I just return STOP instead of PAUSE. See the code comment why I think this is fine.
I also implemented early checkpoint removal for the synchronous schedulers, which requires the scheduler to flag paused trials for which the checkpoint is no longer needed.


By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

@mseeger mseeger requested a review from aaronkl March 15, 2023 15:36
@mseeger mseeger force-pushed the remove_checkpoints branch from 1bb58c5 to c489c8f Compare March 15, 2023 15:41
@mseeger
Copy link
Collaborator Author

mseeger commented Mar 16, 2023

I should also check what happens for synchronous multi-fidelity methods.
The potential issues with promo-based ASHA remain, because checkpoints of paused trials are never removed before the end of the experiment.

@mseeger mseeger marked this pull request as draft March 16, 2023 10:46
@mseeger
Copy link
Collaborator Author

mseeger commented Mar 16, 2023

Converting this to draft. I'd still like to resoved the situation for synchronous MF schedulers

@codecov
Copy link

codecov bot commented Mar 16, 2023

Codecov Report

Patch coverage: 97.50% and project coverage change: +0.11 🎉

Comparison is base (b4d4ed0) 64.98% compared to head (b514dda) 65.09%.

Additional details and impacted files
@@            Coverage Diff             @@
##             main     #600      +/-   ##
==========================================
+ Coverage   64.98%   65.09%   +0.11%     
==========================================
  Files         353      354       +1     
  Lines       26238    26279      +41     
==========================================
+ Hits        17051    17107      +56     
+ Misses       9187     9172      -15     
Impacted Files Coverage Δ
syne_tune/optimizer/schedulers/pbt.py 91.66% <77.77%> (+1.44%) ⬆️
syne_tune/callbacks/remove_checkpoints_callback.py 100.00% <100.00%> (ø)
syne_tune/optimizer/scheduler.py 89.55% <100.00%> (+0.32%) ⬆️
...e/optimizer/schedulers/synchronous/dehb_bracket.py 100.00% <100.00%> (ø)
...tune/optimizer/schedulers/synchronous/hyperband.py 96.75% <100.00%> (+0.20%) ⬆️
...imizer/schedulers/synchronous/hyperband_bracket.py 97.08% <100.00%> (+0.11%) ⬆️
...chedulers/synchronous/hyperband_bracket_manager.py 100.00% <100.00%> (ø)
syne_tune/tuner.py 85.43% <100.00%> (+0.53%) ⬆️
tst/schedulers/test_hyperband_sychronous.py 99.00% <100.00%> (+0.08%) ⬆️
tst/test_pbt.py 90.74% <100.00%> (ø)

... and 6 files with indirect coverage changes

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

@mseeger mseeger marked this pull request as ready for review March 17, 2023 08:07
@mseeger
Copy link
Collaborator Author

mseeger commented Mar 17, 2023

OK, this should go in, since it fixed the checkpoint situation for PBT and the synchronous schedulers. Pre-emptive removal for ASHA promotion is more complicated, I'll do it later.

@mseeger mseeger force-pushed the remove_checkpoints branch from 40deb1e to c9d404a Compare March 17, 2023 08:37
@mseeger mseeger force-pushed the remove_checkpoints branch from c9d404a to b514dda Compare March 18, 2023 14:09
@mseeger mseeger merged commit 6e883f8 into main Mar 27, 2023
@mseeger mseeger deleted the remove_checkpoints branch March 27, 2023 09:21
@wesk wesk changed the title Make sure that checkpoints in PBT are removed once they are no longer… fix: Make sure that checkpoints in PBT are removed once they are no longer… Apr 11, 2023
@wesk wesk added the fix label Apr 11, 2023
@wesk wesk changed the title fix: Make sure that checkpoints in PBT are removed once they are no longer… fix: Make sure that checkpoints in PBT are removed once they are no longer needed Apr 11, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants