03:43 < AakashkaushikGit> Thanks @zoq

07:50 < kunalsingh2002Gi> Hello Mentors , I'm KUNAL SINGH 2nd year computer science undergraduate from IIT Kharagpur, India , I'm very much interested in contributing to mlpack but I'm unable to find a starting point , can anyone help me with this ?

07:50 < SuryamArnavKalra> Hi developers !

07:50 < SuryamArnavKalra> I am Suryam Arnav Kalra , second year undergraduate student in Computer Science and Engineering department at IIT Kharagpur , India.

07:50 < SuryamArnavKalra> I am a beginner in open source and i am interested in contributing to mlpack. It would be really great if you could help me get a starting point to begin my journey

09:05 < NippunSharmaGitt> Hi @suryam35 @kunalsingh2002 , you can look at the community page as a start [https://www.mlpack.org/community.html ]( https://www.mlpack.org/community.html), [https://www.mlpack.org/gsoc.html ]( https://www.mlpack.org/gsoc.html ) might be helpful

09:28 < RishabhGarg108Gi> @suryam35 @kunalsingh2002 , taking up a "good first issue" would be a good entry point. If you don't find anything interesting to work, then you can always open an issue for something that you would like to work and could be added to mlpack. Feel free to post your queries here :)

09:56 < Samyak> Hi, I am working on parallelization using MPI(message passing interface) library. Has anyone worked before in this area? Are there any challenges you faced for any particular algorithm?

13:00 < NippunSharmaGitt> Hi I was looking at the source code of the automatic bindings of mlpack. Can someone tell what will be a good place to start understanding it ?

17:13 < jeffin143[m]> @yashwants19:matrix.org: are you there?

20:32 < VivekTalwarGitte> Hello Everyone , I'm Vivek Talwar 1st year computer science graduate at IIIT Hyderabad, India , I'm very much interested in contributing to mlpack

20:39 < zoq> VivekTalwarGitte: Hello there, in case you search for a starting point, checkout our community page - https://www.mlpack.org/community.html

20:44 < VivekTalwarGitte> thanks

21:05 < VivekTalwarGitte> Is it compulsory to contribute in C++ only or we can contribute in python bindings as well

21:05 < VivekTalwarGitte> ?

21:06 < VivekTalwarGitte> https://www.mlpack.org/doc/stable/python_documentation.html

21:07 < VivekTalwarGitte> I am referring this to get started in python

21:08 < zoq> VivekTalwarGitte: If you like to contribute a new method or extend an exsisting method, you need to do that in C++; the Python bindings always use the C++ implementation.

21:10 < VivekTalwarGitte> let me get my hands dirty with some starting modules and work around with basics and than think of how I can contribute to the current methods

21:10 < VivekTalwarGitte> or propose a new one altogether in C++

21:11 < zoq> Sounds good, let us konw if there is anything we can clarify.

21:48 < zoq> AakashkaushikGit: Got the simple network working, I'll create a bigger network and compare it with the current implementation.

22:04 < shrit[m]> vivektalwar13071999 you can add some examples in Python in the example repository

22:30 < zoq> A first test shows that the boost:variant approach is slightly -> boost::variant 70.1904s : vtable 70.2215s

22:30 < zoq> faster

22:44 < rcurtin> zoq: interesting; is that the average over three trials?

22:45 < zoq> 5, I'll open a PR with some results tomorrow.

22:49 < zoq> If the difference is that marginal, I guess we don't have to stick with the variant approach.

22:50 < rcurtin> awesome, interested to see the results

22:51 < rcurtin> I guess the other part of the picture is compile time and memory usage

23:19 < shrit[m]> that is good to here, the difference is very marginal 100 ms seconds for boost variant,

23:20 < shrit[m]> zoq thanks for creating the results, also a big thanks for aakash-kaushik for working on the experiment

23:22 < shrit[m]> Also I think we are going to gain a lot of compile time, however I am not sure about memory usage, I do not see any reason that the memory usage increases with vtable

23:29 < rcurtin[m]> oh, I meant memory usage during compilation. I doubt boost::visitor usage has any runtime memory issues