As it turns out, SystemML master computes correct results, whereas SystemML 0.14 compiled invalid instructions for the initialization of centroids. The details are in the jira and we can go ahead with the 0.15 release.
Regards, Matthias On Fri, Sep 1, 2017 at 11:40 PM, Matthias Boehm <mboe...@googlemail.com> wrote: > Arvind, regarding your question of open issues, there is one must-have > open issue: SYSTEMML-1882. Kmeans shows different convergence behavior > compared to the 0.14 release (and thus also substantially increased > execution times). Since this is potentially an issue of incorrect results, > I marked it as a blocker. > > Regards, > Matthias > > On Fri, Sep 1, 2017 at 7:37 PM, Matthias Boehm <mboe...@googlemail.com> > wrote: > >> +1 - I also think that a 0.15 release is a good idea. Many improvements >> and fixes have been made since 0.14 but we're not quite ready for a 1.0 >> release. >> >> Regards, >> Matthias >> >> On Fri, Sep 1, 2017 at 4:22 PM, Deron Eriksson <deroneriks...@gmail.com> >> wrote: >> >>> +1 Arvind, thank you for getting things moving with a 0.15 release. It's >>> very important with regards to the perceived health of the project. For >>> my >>> next project status report to the Apache Board of Directors, I look >>> forward >>> to reporting that we are about to publish a 0.15 release. >>> >>> Deron >>> >>> >>> On Fri, Sep 1, 2017 at 1:35 AM, Arvind Surve <ac...@yahoo.com.invalid> >>> wrote: >>> >>> > Hi, >>> > As we could not complete some of the key features targeted for SystemML >>> > 1.0, we will go next with release 0.15 instead of release 1.0. >>> > If you have any critical fixes to be added for master repository (now >>> > targeted for release 0.15) please let me know by EOD 09/01/2017 PST. >>> > I will plan on starting SystemML 0.15 Release Candidate build by >>> Tuesday >>> > 09/05/2017 morning. >>> > ThanksArvind--------Arvind Surve | Spark Technology Center | >>> > http://www.spark.tc/ >>> >> >> >
