verne.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
< rcurtin> zoq: nice catch with the minibatch SGD test, thanks
< rcurtin> I'm still trying to reproduce the failure
< rcurtin> actually now that I look at it, I think that *was* the cause of the failure... testing on uninitialized data
< rcurtin> I guess on my system that was all uninitialized to 0s, and logistic regression didn't fail on that
< zoq> the travis build still fails
< zoq> ah no, wrong window
< rcurtin> yeah, I think it's still building the one with your changes
< zoq> Now I can use your test for the RMSprop optimizer :)
< rcurtin> :)
travis-ci has joined #mlpack
< travis-ci> mlpack/mlpack#535 (master - a1486eb : marcus): The build is still failing.
travis-ci has left #mlpack []
travis-ci has joined #mlpack
< travis-ci> mlpack/mlpack#536 (master - 2aedd27 : Ryan Curtin): The build is still failing.
travis-ci has left #mlpack []
< rcurtin> weird, looks like the fact that point 500 wasn't initialized was causing the issue
< zoq> that is weird ... another weird thing is that you can always count on travis to pick a bad seed or something like that ... I wonder why the RMSprop test works this time, but luckily there is another test that failed
< rcurtin> yeah, travis seems especially unlucky with picking seeds
< rcurtin> I guess that is good for testing though :)