09:49 < travis-ci> ShikharJ/mlpack#42 (RBM - bcf8aca : Shikhar Jaiswal): The build has errored.

09:49 < travis-ci> Change view : https://github.com/ShikharJ/mlpack/compare/90e5827c17ba...bcf8aca07fe7

09:49 < travis-ci> Build details : https://travis-ci.org/ShikharJ/mlpack/builds/332690106

15:31 < manish7294> @rcurtin @zoq : I have trained two sparse coding models with same input parameters one after another and got different results for both dictionary and codes matrices. Is it normal?

15:32 < rcurtin> did you use the same random seed?

15:32 < rcurtin> I think you are talking about writing tests for them, in which case you shouldn't set the random seeds to the same value (instead use an initial dictionary), but if you are just testing e.g. from the command-line program, the --seed option will help :)

15:32 < manish7294> rcurtin: I haven't set seed so default value 0 is being used.

15:33 < rcurtin> right, if it's a command-line program seed=0 means 'std::time(NULL)' is used so it will be different every run

15:33 < rcurtin> but if you are in just one section of code and you train models sequentially, you will get a different result unless you reset the random seed (or use an initial dictionary)

15:34 < manish7294> rcurtin: Thanks I will try that.

15:35 < ImQ009> I seem to have a little trouble understanding the RNN architecture that's been implemented in mlpack. How can I specify the amount of neurons in the layers?

15:38 < rcurtin> ImQ009: these are specified in the parameters when you call the Add() function, I think the examples in recurrent_network_test.cpp should help clarify

15:39 < ImQ009> Oh, I see. Thank you

15:39 < ImQ009> Yes, this will be very helpful

15:45 < manish7294> rcurtin: Thanks, resetting seed worked.

15:48 < rcurtin> manish7294: good to know, but just FYI that's not a great idea in the tests so I'd recommend setting the initial dictionary instead, if you are looking into this for the purpose of writing tests :)

15:50 < manish7294> rcurtin: marked your words. Thanks for advice :)

16:08 < rcurtin> sure, happy to help

18:45 < nurdtechie98> Hey guys, I am new to Open Source.Looking forward to contribute..please help me get started.

18:46 < zoq> nurdtechie9: Hello, a good starting point is: https://github.com/mlpack/mlpack/blob/master/CONTRIBUTING.md

18:47 < nurdtechie98> Ohkk cool,thanks man.

20:30 < ImQ009> Anybody was able to set mlpack up under Windows? I've got issues with cmake not finding the boost libraries

20:31 < zoq> ImQ009: Are you able to use anaconda?

20:32 < ImQ009> Never used it. Though it appears the only Windows-built package of mlpack is quite outdated

20:33 < rcurtin> doesn't AppVeyor have pre-built Windows packages you could download? I am not sure how to do that though, I have never done it

20:33 < rcurtin> that would be the git master repo not 2.2.5 though

20:34 < rcurtin> https://ci.appveyor.com/project/mlpack/mlpack/build/%233780/job/uu6i7xlpaan6agac/artifacts

20:34 < zoq> if you just need the latest lib/bins AppVeyor is probably the best option

20:34 < rcurtin> like perhaps that might be useful? I am not sure how to use it though, I am not a good windows user

20:35 < ImQ009> Oh, interesting

20:35 < zoq> I have a conda script if you like to build it yourself

20:36 < zoq> I think conda is way easier to use as nuget

20:37 < ImQ009> I'll try the package that's on AppVeyor, thanks once again

20:37 < zoq> okay, make sure you get the package with all the libs

20:38 < ImQ009> I'm still rather puzzled by the boost issue. I'll try to investigate it more tommorow

20:38 < ImQ009> Because I was simply following this guide https://keon.io/mlpack-on-windows/

20:39 < zoq> right, that should work just fine

20:40 < ImQ009> Though, I'm using MSVC 2017, which might simply not work with this

20:42 < zoq> We are testing it for 2015 and 2017, so it should work.

20:42 < rcurtin> ImQ009: I gave the link for the 2015 build I think, but if you dig around you should be able to find the 2017 build too

20:43 < ImQ009> rcurtin, it's fine, 2015 should link just fine

20:45 < ImQ009> Maybe the boost version I'm using is the problem? I'm using Boost 1.66 for vc141 instead of 1.60 for vc140

20:45 < ImQ009> I see you're using the vc140 version for 2017 builds

23:56 < jenkins-mlpack> Project mlpack - git commit test build #2914: FAILURE in 26 min: http://masterblaster.mlpack.org/job/mlpack%20-%20git%20commit%20test/2914/

23:56 < jenkins-mlpack> * Ryan Curtin: Clean up CLI singleton implementation.

23:56 < jenkins-mlpack> * Ryan Curtin: singletons.hpp can also be removed.