[
https://issues.apache.org/jira/browse/SPARK-51118?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17925872#comment-17925872
]
Ruifeng Zheng commented on SPARK-51118:
---------------------------------------
The ml side just uses the public APIs of UDF, this seems to be an issue in UDT
fallback
> Use Arrow python UDF for ml internal code
> -----------------------------------------
>
> Key: SPARK-51118
> URL: https://issues.apache.org/jira/browse/SPARK-51118
> Project: Spark
> Issue Type: Sub-task
> Components: ML, PySpark
> Affects Versions: 4.0.0
> Reporter: Xinrong Meng
> Priority: Major
>
> When enabling Arrow optimization for Python UDF, areas below error out with
> ValueError: 'tolist' is not in list.
> - CrossValidator Fit
> The issue occurs in test_save_load_nested_estimator, when executing
> cv.fit(dataset).
> Full stacktrace see
> [https://github.com/xinrong-meng/spark/actions/runs/13167027085/job/36749584932]
> - OneVsRestModel _transform
> Full stacktrace see
> [https://github.com/xinrong-meng/spark/actions/runs/13188569330/job/36819938746]
> - UnaryTransformer _transform
> Now Arrow optimization is disabled in those places explicitly, but we should
> enable those in the near future.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
