marcoabreu commented on issue #11544: [MXNET-620]Fix flaky test batchnorm training URL: https://github.com/apache/incubator-mxnet/pull/11544#issuecomment-403058101 I see, that makes sense. Where does it fall back to if MKLDNN is not available? My concern here is that this fallback actually seems to be failing in some cases. This shouldn't happen, right? To me, it seems like we're now removing this problematic test although the problem might lie in whatever algorithm implementation we use in that fallback - and the test actually serving its purpose to reveal that problem.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
