Update PFI tests so they configure sensitive learners to use a single thread. #1926
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
TomFinley
requested review from
yaeldekel,
artidoro,
shmoradims,
sfilipi,
rogancarr,
montebhoover and
Zruty0
TomFinley
commented
Dec 19, 2018
| var model = ML.Clustering.Trainers.KMeans("Features", clustersCount: 5, | ||
| advancedSettings: args =>{ args.NormalizeFeatures = NormalizeOption.No;}) | ||
| var model = ML.Clustering.Trainers.KMeans("Features", clustersCount: 5, | ||
| advancedSettings: args =>{ args.NormalizeFeatures = NormalizeOption.No; args.NumThreads = 1; }) |
There was a problem hiding this comment.
This is what I am hoping will solve the issue. But, again, it's mostly speculative on my part. I have TBH only seen the spurious failure happen on the build servers for MacOS debug, for which my ability to repro is limited.
There was a problem hiding this comment.
I am also not sure what is this NormalizeOption.No. I do not think Fit will auto-normalize in any event anyway?
There was a problem hiding this comment.
Yep, I think this argument is only use by the entry point.
In reply to: 242977200 [](ancestors = 242977200)
artidoro
approved these changes
Dec 19, 2018
ghost
locked as resolved and limited conversation to collaborators
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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.
Fixes #1925 .
While I was at it I also set the number of threads to logistic regression to 1 as a better practice, since even though it's less likely to trigger a failure here, better safe than sorry. As mentioned in the issue since FastTree's results don't depend on number of threads I did not set it there.
Also incidentally started using
Assert.Equals(a, b)instead ofAssert.True(b == a), so as to provide slightly more helpful error messages for when test failures happen, but that should not affect the running of the test.As with most spurious test failures, due to the indeterminacy of the effects I am not actually 100% sure this will solve the issue, but it was just one obvious mistake I saw when reviewing the test code, so I figured this might help improve the spurious failures on account of this issue.