02:32 < rcurtin> wow, I hadn't looked at the summer of code ideas wiki page in a while... but zoq's mlpack model zoo image made me audibly laugh :)

04:49 < jeffin143[m]> Toshal Agrawal: saw your idea on gsoc, i really do thing that is a great idea, and also agree with your point on avoiding a dependancy , but i guess developing a tool from scratch would take lot of time

04:50 < jeffin143[m]> Instead what we can do is transfer the data with a set format , may be json object or something and then since tensor board or other thing can render data, we just need to worry how to paas the data too front end

04:51 < jeffin143[m]> The whole point being , handle backend and the rendering part or other could be handled using existing. I am really not sure how would that pan out

04:51 < jeffin143[m]> But i am really excited to see the visualisation option in mlpack

04:53 < jeffin143[m]> While writing this , i again thought about it , an i guess writing frontend wouldn't be a big deal with all the library existing ,

04:53 < jeffin143[m]> May be i will come up with better idea :)

04:53 < jeffin143[m]> I just really felt happy to see that idea and come running here 😂

06:04 < metahost> zoq, rcurtin: For the "running mlpack on resource constrained devices" project, would running the ensmallen binary on the device be a valid sub-task?

06:05 < metahost> (because the ensmallen_tests binary in itself is pretty large in size too)

06:06 < Nakul[m]> Which might freez the browser..

06:06 < Nakul[m]> > (because the ensmallen_tests binary in itself is pretty large in size too)

06:06 < Nakul[m]> Yeah

08:32 < jenkins-mlpack2> Yippee, build fixed!

08:32 < jenkins-mlpack2> Project docker mlpack nightly build build #595: FIXED in 3 hr 18 min: http://ci.mlpack.org/job/docker%20mlpack%20nightly%20build/595/

13:51 < zoq> metahost: Keep in mind ensmallen is header only, so ensmallen_tests the binary isn't that interesting, but running some code that uses ensmallen on a resource constrained device could be an intersting first step, as it would envolve getting armadillo working as well.

14:46 < AryamanBhagatGit> Hi, I noticed that mlpack doesn't have any algorithms for multi-objective reinforcement learning. I am interested in implementing https://arxiv.org/pdf/1908.08342.pdf. Would this be something that would be useful?

14:53 < hemal[m]> Could anyone tell me how to reference external links in doxygen documentation ?

14:54 < hemal[m]> Say, I want to refer mlpack.org or google.com from doxygen documentation file sample.cpp,

14:54 < hemal[m]> I am not able to find the relevant tag.

15:08 < zoq> AryamanBhagat: Hello, interesting, but personally I'd like to see faster (more) single objective methods first.

15:08 < hemal[m]> Found it : simply, "<a href>xyz</a>"

15:59 < shuaibabh> quit

16:21 < Nakul[m]> https://github.com/mlpack/mlpack/pull/2075

16:22 < Nakul[m]> it is very starnge that test is failing at window vs14 even though i haven't changed anything on that file

18:26 < metahost> zoq: Thanks! So, getting armadillo -> ensmallen -> mlpack working on the H/W should be a desired progression right?

18:29 < metahost> Also, since ensmallen in header only, I think with optional includes, we can gain a lot easily.

18:30 < Nakul[m]> curl https://ftp.fau.de/macports/distfiles/armadillo/armadillo-8.400.0.tar.xz | tar -xvJ && cd armadillo*

18:30 < Nakul[m]> how could i get latest version of armadillo by this command since replacing the version number with latest one doesn't seems to work

18:32 < metahost> Nakul: You can install Armadillo using the package manager on your platform and keep it updated easily!

18:38 < Nakul[m]> i have latest on my local machine but build is falling on model repo due to this error Could NOT find Armadillo: Found unsuitable version "6.500.5", but required

18:38 < Nakul[m]> is at least "8.400.0" (found /usr/lib/libarmadillo.so)

20:40 < HimanshuPathakGi> Hey rcurtin while using this

20:40 < HimanshuPathakGi> while using this command

20:41 < HimanshuPathakGi> cmd.exe /C "cd . && C:\ProgramData\chocolatey\bin\c++.exe -Wall -Wextra -ftemplate-depth=1000 -Wno-unused-function -ffunction-sections -fdata-sections -Wa,-mbig-obj -O3 -fopenmp -O3 -DNDEBUG src/mlpack/methods/local_coordinate_coding/CMakeFiles/mlpack_local_coordinate_coding.dir/local_coordinate_coding_main.cpp.obj -o mlpack_local_coordinate_coding.exe -Wl,--out-implib,libmlpack_local_coordinate_coding.dll.a

20:41 < HimanshuPathakGi> -Wl,--major-image-version,0,--minor-image-version,0 -LD:/bld/mlpack_1580139447374/_h_env/Library/lib libmlpack.a D:/bld/mlpack_1580139447374/_h_env/Library/lib/armadillo.lib D:/bld/mlpack_1580139447374/_h_env/Library/lib/blas.lib D:/bld/mlpack_1580139447374/_h_env/Library/lib/lapack.lib D:/bld/mlpack_1580139447374/_h_env/Library/lib/libboost_program_options.lib

20:41 < HimanshuPathakGi> D:/bld/mlpack_1580139447374/_h_env/Library/lib/libboost_unit_test_framework.lib D:/bld/mlpack_1580139447374/_h_env/Library/lib/libboost_serialization.lib -lkernel32 -luser32 -lgdi32 -lwinspool -lshell32 -lole32 -loleaut32 -luuid -lcomdlg32 -ladvapi32 && cd .

20:41 < HimanshuPathakGi> do we need -L before every library

20:42 < HimanshuPathakGi> like this -LD:/bld/mlpack_1580139447374/_h_env/Library/lib/armadillo.lib

20:42 < HimanshuPathakGi> or not

20:42 < HimanshuPathakGi> (edited) or not => do we need -L before every address of library

20:52 < zoq> metahost: Agreed, also we have to look into some other dependencies as well e.g. boost, BLAS, but the pipeline for each looks probably similar.

20:52 < zoq> Nakul[m]: In this case we have to update the CI config, to install a newer armadillo version.

20:56 < zoq> HimanshuPathak: No -L, tells the compiler to look into a directory for library files.

20:57 < Nakul[m]> @zoq I tried adding different link to download armadillo in .tarvis.yaml but couldn't work.

20:58 < Nakul[m]> Specifically this one http://sourceforge.net/projects/arma/files/armadillo-9.800.4.tar.xz

21:00 < zoq> HimanshuPathak: Nakul[m]: Let me comment on the PR.

21:00 < Nakul[m]> Ok

21:01 < zoq> Nakul[m]: Let's see if that works.