[
https://issues.apache.org/jira/browse/OPENNLP-722?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14177409#comment-14177409
]
Joern Kottmann commented on OPENNLP-722:
----------------------------------------
I looked a bit further and came to the conclusion that it matters in which
order the active parameters in the PerceptronModel.evals method are added
together. It is probably a good idea to have a close look there to see if some
tricks should be applied to make the computation more accurate (e.g. sorting
the values to add values with a similar size first).
Anyway, I don't undertand why the parameters get sorted in the ComparableEvent.
I suggest that we don't sort them to not change the order in case the ids get
assigned differently accross Java versions.
Another option would be to make the id assignment stable, e.g. by using sorting
instead of relying on the sorting HashSet provides.
Any opinions?
> PerceptronPrepAttachTest fails only on Java 8
> ---------------------------------------------
>
> Key: OPENNLP-722
> URL: https://issues.apache.org/jira/browse/OPENNLP-722
> Project: OpenNLP
> Issue Type: Bug
> Components: Build, Packaging and Test
> Affects Versions: 1.6.0
> Reporter: Joern Kottmann
> Assignee: Joern Kottmann
> Priority: Minor
> Fix For: 1.6.0
>
>
> The test
> PerceptronPrepAttachTest.testPerceptronOnPrepAttachDataWithStepSizeDecrease
> fails if executed on Java 8.
> It would be really nice to track down the cause of that.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
