[Tune] Fix best trial in ProgressReporter with nan #31276
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: Antoni Baum <[email protected]>
justinvyu
approved these changes
Dec 21, 2022
|
Inf values will be handled correctly by the comparison, so I think it is fine to keep them. |
AmeerHajAli
pushed a commit
that referenced
this pull request
Jan 12, 2023
If a trial reports a metric value of `np.nan` or similar (which is not vanilla Python `None`), any comparison made between that value and any other value will return false, thus leading to the `nan` value being considered the best value for the purpose of determining best trial in `TuneReporterBase` if it was reported first. ``` Trial 1 -> np.nan Trial 2 -> 1 1 > np.nan == False -> np.nan is the best trial ``` To guard against that issue, we use `pd.isnull` check which is more general than the `is None` check used previously. Signed-off-by: Antoni Baum <[email protected]>
tamohannes
pushed a commit
to ju2ez/ray
that referenced
this pull request
Jan 25, 2023
If a trial reports a metric value of `np.nan` or similar (which is not vanilla Python `None`), any comparison made between that value and any other value will return false, thus leading to the `nan` value being considered the best value for the purpose of determining best trial in `TuneReporterBase` if it was reported first. ``` Trial 1 -> np.nan Trial 2 -> 1 1 > np.nan == False -> np.nan is the best trial ``` To guard against that issue, we use `pd.isnull` check which is more general than the `is None` check used previously. Signed-off-by: Antoni Baum <[email protected]> Signed-off-by: tmynn <[email protected]>
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.
Signed-off-by: Antoni Baum [email protected]
Why are these changes needed?
If a trial reports a metric value of
np.nanor similar (which is not vanilla PythonNone), any comparison made between that value and any other value will return false, thus leading to thenanvalue being considered the best value for the purpose of determining best trial inTuneReporterBaseif it was reported first.To guard against that issue, we use
pd.isnullcheck which is more general than theis Nonecheck used previously.Related issue number
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.