Re: Failures for sklearn
Hi again,
Am Tue, Feb 15, 2022 at 12:15:38PM +0100 schrieb Andreas Tille:
> Dear Chiara,
>
> thanks a lot for working on this.
>
> Am Fri, Feb 11, 2022 at 02:33:07PM -1000 schrieb Chiara Marmo:
> > Dear scikit-learn maintainers,
> >
> > I have updated the scikit-learn version on salsa and modified some of the
> > EXCLUDE_TESTS variables in order to reflect the current status of the
> > scikit-learn issue tracker.
> > I was hoping to run continuous integration on salsa, but it is very time
> > consuming for scikit-learn, so I'm switching it on and off, when needed.
>
> Speaking about Salsa-CI: The last run is
>
> https://salsa.debian.org/science-team/scikit-learn/-/jobs/2454291
>
> connected to commit dba31326ec5b746ff3ba2ae067f4b49f4811c9eb. Would you
> confirm that your last commit excludes those two failures that are
> breaking the test?
Seems there is one failure for i386 remaining:
https://salsa.debian.org/science-team/scikit-learn/-/jobs/2469653
Given that it says:
E Not equal to tolerance rtol=0.001, atol=0
E
E Mismatched elements: 1 / 1024 (0.0977%)
E Max absolute difference: 1.64902041e-07
E Max relative difference: 0.00120168
I'll try relaxing tolerance limits a bit to solve this issue.
I'll let you know more later
Andreas.
--
http://fam-tille.de
Reply to: