08:27 < addy> Hey, I'm new to open source, I have some experience coding programs in c++ (Mostly competitive programming problems), can someone guide me where do I start? :P

08:41 < sameeran> hi

08:42 < sameeran> i need help where to start from

12:30 < sam___> hey anyone i am getting problem for this line mlpack_fastmks --help as ----segmentation fault

12:30 < sam___> please cananyyone help me

12:37 < rayl> I installed mlpack on my macOS from source code, but it cannot use command line. The error shows that "dyld: Library not loaded: @rpath/libhdf5.101.dylib Referenced from: /usr/local/lib/libarmadillo.8.dylib Reason: image not found Abort trap: 6"

14:09 < zoq> ImQ009: Good point, it's on my list to adjust the NLL layer to use 0 labels as input.

14:11 < zoq> ImQ009: Currently you have to create a datastream data, but adding a Predict that works with streaming data could be added, if that would makes things easier for you.

14:12 < zoq> vmg27: Good to know, I think boost will fix the issue in the upcoming version.

14:14 < zoq> dungeon_master: You could disable exlude the command line executables from the build process by adding -DBUILD_CLI_EXECUTABLES=OFF to the cmake command.

14:16 < zoq> dungeon_master: Not sure about the Clion IDE, maybe they do some extra steps in the backgroud? I think they use the same cmake file to configure the build process.

14:17 < zoq> addy: sameeran: http://www.mlpack.org/involved.html and mlpack.org/gsoc.html is a good point to start.

15:30 < ImQ009> zoq, I see, thank you for clarifying this. As for the data stream. I'm not entirely sure how it's handled underneath so I'd like to ask you the following

15:30 < ImQ009> Should the rho (and consequently, the amount of timesteps in streamed data) during prediction be the same as during training?

15:31 < ImQ009> Or can I set it to 1 for the purpose I explained earlier?

16:20 < zoq> Currently it has to be the same.

16:25 < ImQ009> Okay. That would explain unexpected results

16:25 < ImQ009> All clear now :)