[jira] [Resolved] (SPARK-57951) MLTest single-prediction checks should tolerate last-ULP FP differences (macOS arm64)

["Hyukjin Kwon (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/SPARK-57951?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Hyukjin Kwon resolved SPARK-57951.
----------------------------------
    Fix Version/s: 4.1.3
                   3.5.9
                   4.0.4
                   4.2.1
                   4.3.0
       Resolution: Fixed

Issue resolved by pull request 57032
[https://github.com/apache/spark/pull/57032]

> MLTest single-prediction checks should tolerate last-ULP FP differences (macOS arm64)
> -------------------------------------------------------------------------------------
>
>                 Key: SPARK-57951
>                 URL: https://issues.apache.org/jira/browse/SPARK-57951
>             Project: Spark
>          Issue Type: Bug
>          Components: ML, Tests
>    Affects Versions: 5.0.0
>            Reporter: Hyukjin Kwon
>            Assignee: Hyukjin Kwon
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 4.1.3, 3.5.9, 4.0.4, 4.2.1, 4.3.0
>
>
> MLTest.testClassificationModelSingleRawPrediction and testProbClassificationModelSingleProbPrediction compare the DataFrame transform output against the scalar predictRaw/predictProbability output with exact === equality. On arm64 macOS the two paths can differ in the last ULP (e.g. 1.543502002724983 vs 1.5435020027249835), failing MultilayerPerceptronClassifierSuite and others on the Build / Maven (JDK 21, MacOS-26) lane while Linux stays bit-identical. Compare with a tight absolute tolerance (1e-9) instead.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org