ChanServ changed the topic of #mlpack to: "mlpack: a fast, flexible machine learning library :: We don't always respond instantly, but we will respond; please be patient :: Logs at http://www.mlpack.org/irc/
< GauravSinghGitte> Hi, @zoq if you have aome time kindly have a look at #2238. Thanks.
< kartikdutt18Gitt> Great, I'll make the changes for the same then.
richk449 has joined #mlpack
< jenkins-mlpack2> Project docker mlpack nightly build build #632: STILL FAILING in 2 hr 46 min: http://ci.mlpack.org/job/docker%20mlpack%20nightly%20build/632/
arham has joined #mlpack
arham has quit [Remote host closed the connection]
zoq has quit [Ping timeout: 265 seconds]
zoq has joined #mlpack
Sreekanth has joined #mlpack
< Sreekanth> Hi all,
< zoq> Sreekanth: Hello :)
< Sreekanth> I am Sreekanth, a second year Computer Science pursuing B.Tech at Indian Institute of Technology (IIT), Bhubaneswar. I want to be part of GSoC 2020, under the organization mlpack. I went through the ideas list proposed by the mlpack organization and found the idea "Reinforcement Learning" really interesting. Can someone help me out in proceeding to
< Sreekanth> the further steps..
< Sreekanth> *Computer Science under graduate
< zoq> Sreekanth: Thanks for the interest, https://www.mlpack.org/community.html and https://www.mlpack.org/gsoc.html should you help get started.
< Sreekanth> Can I get some more information regarding this idea 'Reinforcement Learning' :] . Currently I am getting familiar with its source code https://github.com/mlpack/mlpack/tree/master/src/mlpack/methods/reinforcement_learning. There are no relevant tickets or issues provided for this idea in the ideas list. Is there something else that I can do apart
< Sreekanth> from getting familiar with the mlpack source code and the tests ? At the present moment, should I need to go through the research papers provided as the references ?
togo has joined #mlpack
< togo> do you apply ML on the IRC chat?
< zoq> Sreekanth: Yeah, you can go over the papers as well; that said maybe you can find something that could be imporved or extended in the RL codebase?
< zoq> togo: What do you mean?
Sreekanth has quit [Remote host closed the connection]
richk449 has quit [Remote host closed the connection]
ImQ009 has joined #mlpack
BisakhMondal[m] has joined #mlpack
< pickle-rick[m]> What version of gcc do you guys generally use? I'm having trouble building mlpack from source. Its throwing a seg fault due to "internal compiler error". I haven't had this problem before. (Ubuntu 18.04, gcc=7.4.0)
< zoq> pickle-rick[m]: Most probably you run out of memory.
< pickle-rick[m]> Oh
< zoq> pickle-rick[m]: do you build with multiple cores?
< pickle-rick[m]> I tried make alone, and with -j option. Both failed.
< pickle-rick[m]> And I think it's failing at different points each time I try to build
< zoq> I see, how much memory do you have?
< pickle-rick[m]> 16 gigs. That's why I was kind of surprised, when you said I might've run out of memory
< zoq> okay, yeah that is more than enough
< zoq> can you track the memory when you build mlpack?
< pickle-rick[m]> Is it something about my local build environment that's going wrong?
< rcurtin> usually if it's an internal compiler error, there will be some other output near there that might say something like "Killed" if it's really running out of memory
< rcurtin> I'm on debian sid, so gcc --version for me reports 9.2.1
< rcurtin> I don't know if upgrading will help though
< rcurtin> maybe a lot of that 16GB is in use by a browser or something?
< pickle-rick[m]> I'll try the memory tracking and upgrading gcc...
< zoq> I also use 9.2.1-21.
< pickle-rick[m]> Although I'm not sure on how I might do memory tracking... I'll look it up
< zoq> just open top or htop
< pickle-rick[m]> <zoq "just open top or htop"> okay thanks. will do
< rcurtin> /usr/include/c++/7/system_error:232:44: internal compiler error: Segmentation fault
< rcurtin> :-O
< rcurtin> that doesn't look like an out of memory error
< zoq> yeah
< rcurtin> uhh, maybe upgrading gcc will work, but if it doesn't my suggestions past that would be
< rcurtin> - try clang instead?
< zoq> I think Conrad said something about that particular version, not sure.
< rcurtin> - ...this is a long shot, but maybe your RAM is bad? maybe run a memtest? I'd only suggest that if you're occasionally seeing other strangenesses and segfaults in random things
< rcurtin> zoq: I think you might be right, I just can't remember any of the details
< pickle-rick[m]> <rcurtin " - ...this is a long shot, but m"> This could actually be the case... I've seen some strangeness in random things for sure :)
< pickle-rick[m]> but i'll try upgrading gcc in the mean time
< rcurtin> pickle-rick[m]: it certainly couldn't hurt to try, but it's probably a long shot
< rcurtin> yeah, definitely a gcc upgrade first is the easiest thing to try I think :)
< zoq> the gcc-7.4.0 build looks fine
< rcurtin> hmm, it does, for all versions of boost and armadillo that we're testing too
< rcurtin> I need to get some lunch, I'll be back later :)
< zoq> but yeah, easy to upgrade gcc
< BisakhMondal[m]> Hello community, myself Bisakh, a GSoC aspirant, want to contribute to the idea 'Essential Deep Learning Modules'. Should I contact the potential mentors?
< zoq> BisakhMondal[m]: Hello, we like to discuss that sort of things on the public channels.
< togo> zoq: do you apply ML on the IRC chat? Its because I notice there is even a log of the chat, so you have a data set you can train with and a live testbed?
< zoq> togo: We don't, but I think this is a great idea, maybe some sort of ml support bot for some simple questions :)
< pickle-rick[m]> <rcurtin "pickle-rick: it certainly couldn"> looks like you're right... The memtest ran fine for 100M, 200M and 300M. But for any tests for 400M and above, the whole thing just goes crazy and throws a bunch of failures.
< togo> It would be just the learning experience if i could observe, how would one go about it?
< togo> zoq: I am at zero with ML
< togo> however I see the limitation of this user interface allowing only text output
zoso_floyd has joined #mlpack
zoso_floyd has quit [Client Quit]
naruarjun[m] has joined #mlpack
< BisakhMondal[m]> Hello zoq. isn’t it a public channel, I thought so..
< zoq> BisakhMondal[m]: yes, this is a public channel I thought you like to contact mentors per mail or something like that
< zoq> togo: Well, I think this is a great idea for a get started tutorial or something like that.
< pickle-rick[m]> <pickle-rick[m] "looks like you're right... The m"> I would need to replace my ram yeah?
kartik18 has joined #mlpack
kartik18 has left #mlpack []
< togo> zoq: and what if the ml learns how to improve itself, takes over the channel and eventually the whole world ;)
ahmad242 has joined #mlpack
< ahmad242> hii ,
ahmad242 has quit [Remote host closed the connection]
< zoq> togo: maybe it takes over the channel :)
M_slack_mlpack_7 has joined #mlpack
ImQ009 has quit [Quit: Leaving]
< rcurtin> pickle-rick[m]: wow, I'm really surprised, I didn't really think it would be bad memory
< rcurtin> and yeah, you would probably need to replace the RAM to resolve that issue, at least, unfortunately
< pickle-rick[m]> Yeah... bummer... But on the bright side, we know what the issue is. I can hopefully get this fixed. Thanks @rcurtin @zoq
< rcurtin> sure, happy to help :) sorry it turned out this way though. did a gcc upgrade help, by any chance?
togo has quit [Quit: Leaving]