Improve Ludwig resilience to Ray Tune issues #1660
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
justinxzhao
reviewed
Jan 7, 2022
| collect_overall_stats=True, | ||
| return_type='dict', | ||
| debug=debug | ||
| if best_model_path is not None: |
There was a problem hiding this comment.
Not totally necessary for this PR, but the level of nesting here is a bit intimating. It might be worth abstracting out some of this logic into helper methods, or adding comments to describe stages of the overall flow.
There was a problem hiding this comment.
Good call out, thanks! PR updated.
justinxzhao
approved these changes
Jan 8, 2022
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Outdated suggestions cannot be applied.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.
*) A trial can intermittently fail in Ray Tune, e.g., ray-project/ray#21458
In general, it seems more resilient to specify the option to have Ray Tune retry failed trials once.
*) Ray Tune can intermittently fail to upload checkpoints, e.g., ray-project/ray#21469
In general, it seems more resilient to have the Ludwig post-search evaluation process warn on & skip
trials with missing checkpoints, to provide value for the completed search. Note that unless
the checkpoints from the best trial are missing, the lack of other checkpoints is not critical.