mlpack IRC logs, 2020-02-23

Logs for the day 2020-02-23 (starts at 0:00 UTC) are shown below.

February 2020
--- Log opened Sun Feb 23 00:00:59 2020
04:09 -!- jenkins-mlpack2 [] has joined #mlpack
05:16 -!- ssk99 [31ce089a@] has joined #mlpack
05:22 -!- ssk99 [31ce089a@] has quit [Remote host closed the connection]
06:02 < metahost> Okay I was wrong regarding the file size. It is ~55MB to begin with. 300+ MB looked excessive, don't know why that came up :))
07:13 -!- Yihan [2d20f7f2@] has joined #mlpack
07:31 < metahost> Ah okay, building with -D DEBUG=ON produced that (348MB executable)!
08:01 -!- akash [~akash@] has joined #mlpack
08:02 -!- akash is now known as akashd
08:14 -!- ImQ009 [~ImQ009@unaffiliated/imq009] has joined #mlpack
08:23 < PrinceGuptaGitte> Hi everyone, does FFN has any way to get model summary like we can do in keras with`model.summary()`? I searched through it but couldn't find it. I think it is a necessary part FFN should have. So I'm thinking to implement it.
08:23 < PrinceGuptaGitte> I've searched myself too. I just want to get confirmation that it doesn't exist already because I was working on Residual Block for 2-3 days but now realized that it was already there in Sequential Layer (but I learned a lot of things while doing so because mlpack really is fairly complicated).
08:24 -!- Pranshu54 [76b9ec67@] has joined #mlpack
08:55 < Pranshu54> Hello everyone, I am Pranshu Vaish from India. Currently I am pursuing my UG in Mathematics and Computing from IIT(BHU). From last one year I am working on application of computer vision in biometrics and hyperspectral images for segmentation and classification. I am proficient with python and c++. I have gone through this link
08:55 < Pranshu54> and interested in contributing to mlpack under GSoC 2020. Would like to know how to proceed further.
09:02 -!- Pranshu54 [76b9ec67@] has quit [Ping timeout: 260 seconds]
09:40 < KhizirSiddiquiGi> @prince776 I don't think it has any.
09:42 < KhizirSiddiquiGi> Pranshu54: you can try going through also.
09:47 < PrinceGuptaGitte> @khizirsiddiqui thanks for the conformation. Then I'll start working on it.
10:12 < SaraanshTandonGi> > I am well versed with machine/deep learning and have also recently submitted a paper to IJCB 2020 which uses computer vision in biometrics. I use C++ regularly but not at development level. Although I am proficient in python and Pytorch.
10:12 < SaraanshTandonGi> > I was wondering how hard is it to get to know the C++ concepts mentioned in and if it is possible to learn them and participate this year.
10:12 < SaraanshTandonGi> Can someone pls help me with this?
10:31 < SriramSKGitter[m> @saraansh1999 : I don't think it is possible to objectively quantify how hard the concepts are, I can only speak for myself when I say that putting in a decent amount of effort, it took me a couple of months to get something non-trivial merged into the master branch. Secondly, I do believe it is possible to learn them and participate this summer :)
10:49 -!- Pranshu54 [76b9ec67@] has joined #mlpack
10:49 -!- Yihan [2d20f7f2@] has quit [Remote host closed the connection]
10:58 < PrinceGuptaGitte> @saraansh1999 I believe it's possible to understand the concepts of c++ required and participate this year. Concepts are fairly hard but if you use c++ regularly as you mentioned it should be fairly easy. I would suggest you try to solve some `good-first-issue` labeled issues. This will help you understand the code-base. When i started working on my first PR #2184, I only had basic understanding of c++ concepts like
10:58 < PrinceGuptaGitte> template and had never used c++ at development level.
11:26 < akashd> Hello Everyone, I am Akash Dhiman from India, currently in 2nd year of my B. E. course from Netaji Subhas University of Technology, Delhi. I have built few basic machine learning projects using keras, tensorflow and tensorflow-js, and have also worked with C/C++. Recently submitted a book chapter on "trends in deep learning application for biometrics" for review as well. I am interested for GSoC as well as contributing to open-source ml communities in general.
11:26 < akashd> I am getting familiar with the library for now and will update once I am done, It's nice to meet you all :)
11:28 < SaraanshTandonGi> Thanks all for the replies.I am trying to get mlpack working on my system. But when
11:29 < SaraanshTandonGi> I try to build it it is getting stuck at giving me an error I am really not able to understand
11:31 < SaraanshTandonGi>
11:31 < SaraanshTandonGi> this is the VERVOSE=1 make output
11:32 < SaraanshTandonGi> I found a couple of issues on github but am not able to get it to worl
11:32 < SaraanshTandonGi> work*
11:32 < SaraanshTandonGi>
11:33 -!- Pranshu54 [76b9ec67@] has quit [Ping timeout: 260 seconds]
11:42 < SaraanshTandonGi>
11:45 < akashd> SaraanshTandonGi which operating system are you using? It appears to be a dependency problem as far as I can tell
11:45 < akashd> try to see the dependencies listed here
11:47 < SriramSKGitter[m> @saraansh1999 : Looks like you're missing armadillo as far as I can see. Are you sure it's installed and present in `usr/lib` or `/usr/local/lib`?
11:50 < SaraanshTandonGi> I'll check
11:50 < SaraanshTandonGi> i installed armadillo first using the sudo apt-get option but that was installing a older version so i went ahead with armadillo's documentation
11:51 -!- akashd [~akash@] has quit [Quit: akashd]
11:55 < SaraanshTandonGi> when i do cmake
11:55 < SaraanshTandonGi> It reports Armadillo libraries to be in usr/lib
11:58 < SaraanshTandonGi> im on ubuntu 16.04
11:59 < SriramSKGitter[m> What do you mean by when?
12:06 * SaraanshTandonGi sent a long message: < >
12:07 < sreenik[m]> freenode_gitter_saraansh1999[m]: Your cython is out of date, although that might not be the exact problem here
12:08 < sreenik[m]> Do you have the hd5 library installed?
12:09 < sreenik[m]> *hdf5 I think
12:16 -!- TanayMehtaGitter [gittertana@gateway/shell/] has joined #mlpack
12:16 < TanayMehtaGitter> Hello everyone! I am a beginner here but have some experience in doing basic Machine Learning using Eigen in cpp. It would be very kind if anyone can show me the way to get started with Contributing on mlpack. I am reading the mlpack source code for a few weeks now, and it seems a bit overwhelming! Any help is greatly needed! Thanks!
12:17 < SriramSKGitter[m> @heytanay : Have a look at this -
12:17 < kartikdutt18Gitt> Hi @heytanay, you can try going through . There are some good first issues that you can look into after that.
12:18 < PrinceGuptaGitte> @saraansh1999 build files were generated successfully. What error are you facing?
12:18 < TanayMehtaGitter> @kartikdutt18 thanks!
12:19 < PrinceGuptaGitte> cython and doxygen shouldn't affect build process, I also didn't have them, it just won't install bindings
13:47 < SaraanshTandonGi> @prince776 im getting error when i do "make"
13:47 < SaraanshTandonGi>
13:58 < PrinceGuptaGitte> have you installed the correct version of armadillo (>=8.4) and uninstalled any previous version(s) installed using apt-get?
13:58 < SaraanshTandonGi> ill cehck
13:59 < PrinceGuptaGitte> Also make sure all these libraries are installed libboost-math-dev libboost-program-options-dev
13:59 < PrinceGuptaGitte> libboost-test-dev libboost-serialization-dev libensmallen-dev
13:59 < PrinceGuptaGitte> (edited) ... installed libboost-math-dev libboost-program-options-dev
13:59 < PrinceGuptaGitte> libboost-test-dev libboost-serialization-dev libensmallen-dev ... => ... installed `libboost-math-dev libboost-program-options-dev
13:59 < PrinceGuptaGitte> libboost-test-dev libboost-serialization-dev libensmallen-dev` ...
14:27 < PrinceGuptaGitte> Hi @zoq , Today I've updated my PRs #2192 and #2184 by adding those function layers in `layer_types.hpp` file so that now they can be added to FFN/RNN models. I didn't realize I'd to do this earlier. Please have a look at them when you have time. Thanks. I feel like they're almost done.
14:30 < SaraanshTandonGi> @prince776 apt-get is not able to fing ensmallen, should i just build it too.
14:31 < PrinceGuptaGitte> are you using `libensmallen-dev` to check it
14:33 < zoq> SaraanshTandon: ensmallen will be downloaded if it's not available, as part of the mlpack cmake step
14:39 < SaraanshTandonGi> is there any python version requirement for mlapck
14:39 < SaraanshTandonGi> I'm currently runing 3.5
14:39 < SaraanshTandonGi> running*
14:40 < zoq> araanshTandon: That works, also you only need python if you like to build the python bindings.
14:42 < zoq> SaraanshTandon: Looking at the past messages, I think your armadillo build isn't correct.
14:42 < zoq> Did you build armadillo manually?
14:44 -!- travis-ci [] has joined #mlpack
14:44 < travis-ci> shrit/models#4 (digit - aa9e453 : Omar Shrit): The build has errored.
14:44 < travis-ci> Change view :
14:44 < travis-ci> Build details :
14:44 -!- travis-ci [] has left #mlpack []
14:50 -!- abinezer [b7521336@] has joined #mlpack
14:52 -!- abinezer [b7521336@] has quit [Remote host closed the connection]
14:58 -!- AbishaiEbenezerG [gitterabis@gateway/shell/] has joined #mlpack
14:58 < AbishaiEbenezerG> hi guys! I'm a total noob here. But can't wait to get started. I've begun learning and using mlpack and will do my best to contribute as much as i can!!
14:59 < zoq> AbishaiEbenezer: Welcome, let us know if we should clarify anything.
14:59 < PrinceGuptaGitte> Good luck with that @abishaiema .Also if you need some directions as to how to get started see and
14:59 < PrinceGuptaGitte> (edited) ... @abishaiema .Also if ... => ... @abishaiema . Also if ...
15:03 < AbishaiEbenezerG> sure sure!!
15:06 < Param-29Gitter[m> Hey @zoq how do I use to find bottlenecks of algorithms? Also can I know what compile time options to use when i try to compile any main.cpp file present in directories. (For example linear_svm_main.cpp)?
15:06 < kartikdutt18Gitt> Hi @zoq, If you get chance, I would love to hear your opinion on #2195 .
15:23 < LakshyaOjhaGitte> Hi @zoq can you please have a look at my 2 pending pr's soft shrink and hard shrink. and .
15:23 < LakshyaOjhaGitte> Do tell me what you think about the same.
15:23 < LakshyaOjhaGitte> Thanks.
15:24 < LakshyaOjhaGitte> I also updated the Hinge Embedding Loss Function it should work now.
15:24 < LakshyaOjhaGitte> Thanks.
15:48 -!- Yihan [2d20f7f2@] has joined #mlpack
16:08 -!- Yihan [2d20f7f2@] has quit [Remote host closed the connection]
16:44 -!- heisenbug [6ac1aafa@] has joined #mlpack
16:45 -!- heisenbug [6ac1aafa@] has quit [Remote host closed the connection]
16:47 -!- heisenbuug [6ac1aafa@] has joined #mlpack
17:26 < PrinceGuptaGitte> Hi, can anyone tell my why `boost::apply_visitor(outputParameterVisitor, network[i]).n_cols` is returning 0, when it's something like 42000
17:28 < PrinceGuptaGitte> Hi @zoq ,` OutputParameterVisitor`'s return type is `arma::mat` but it can also be `arma::cube` because `Forward` function takes that type as a template argument. Wouldn't it break the system when 3D matrix is being outputted by some layer.
17:29 < PrinceGuptaGitte> (edited) ... ,` OutputParameterVisitor`'s return ... => ... ,` OutputParameterVisitor` function return ...
17:29 < PrinceGuptaGitte> (edited) ... ,` OutputParameterVisitor` function return type is `arma::mat` but ... => ... ,` OutputParameterVisitor`when used returns`arma::mat` but ...
18:05 -!- heisenbuug [6ac1aafa@] has quit [Remote host closed the connection]
18:06 -!- heisenbuug [6ac1aafa@] has joined #mlpack
18:06 -!- heisenbuug [6ac1aafa@] has left #mlpack []
18:17 * SaraanshTandonGi sent a long message: < >
18:17 < SaraanshTandonGi> it shows no module named mlpack
18:17 < SaraanshTandonGi> i guess the python bindings are not proper?
18:19 * KhizirSiddiquiGi sent a long message: < >
19:34 -!- ImQ009 [~ImQ009@unaffiliated/imq009] has quit [Quit: Leaving]
19:55 * SaraanshTandonGi sent a long message: < >
21:10 -!- abhive19 [6753827c@] has joined #mlpack
21:19 -!- abhive19 [6753827c@] has quit [Remote host closed the connection]
22:01 -!- jonpsy[m] [jonpsymatr@gateway/shell/] has left #mlpack ["User left"]
23:02 -!- k3nz0 [~k3nz0@unaffiliated/k3nz0] has quit [Quit: Leaving]
23:51 < zoq> PrinceGupta: Right now, the visitor only supports: arma::mat -, but we should use a template here to support other types as well. But you are right we have to be carefull what type is used, not sure they have to be interchangeably (mix of arma::mat and arma::cube).
23:52 < zoq> SaraanshTandon: Nice that you got it working, let us know if we should clarify some points in the documentation to make it easier to avoid those issues.
--- Log closed Mon Feb 24 00:00:00 2020