spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yanbo Liang (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (SPARK-6496) Multinomial Logistic Regression failed when initialWeights is not null
Date Tue, 24 Mar 2015 12:08:53 GMT

    [ https://issues.apache.org/jira/browse/SPARK-6496?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14377759#comment-14377759
] 

Yanbo Liang edited comment on SPARK-6496 at 3/24/15 12:08 PM:
--------------------------------------------------------------

[~srowen] I have address this issue at github. I guess you have seen it.


was (Author: yanboliang):
[~srowen] I have address this issue at github.

> Multinomial Logistic Regression failed when initialWeights is not null
> ----------------------------------------------------------------------
>
>                 Key: SPARK-6496
>                 URL: https://issues.apache.org/jira/browse/SPARK-6496
>             Project: Spark
>          Issue Type: Bug
>          Components: MLlib
>    Affects Versions: 1.3.0
>            Reporter: Yanbo Liang
>
> This bug is easy to reproduce, when use Multinomial Logistic Regression to train multiclass
classification model with non-null initialWeights, it will throw an exception.
> When you run
> {code}
> val lr = new LogisticRegressionWithLBFGS().setNumClasses(3)
> val model = lr.run(input, initialWeights)
> {code}
> It will throw
> {code}
> requirement failed: LogisticRegressionModel.load with numClasses = 3 and numFeatures
= -1 expected weights of length -2 (without intercept) or 0 (with intercept), but was given
weights of length 10
> java.lang.IllegalArgumentException: requirement failed: LogisticRegressionModel.load
with numClasses = 3 and numFeatures = -1 expected weights of length -2 (without intercept)
or 0 (with intercept), but was given weights of length 10
> {code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

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


Mime
View raw message