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/
lozhnikov_ has joined #mlpack
lozhnikov has quit [Quit: ZNC 1.7.5 - https://znc.in]
jenkins-mlpack2 has quit [Ping timeout: 272 seconds]
jenkins-mlpack2 has joined #mlpack
< HimanshuPathakGi> Hey zoq rcurtin do we have implementation flatten layer in ann module
< HimanshuPathakGi> I try to find it but not there in layer
< kartikdutt18Gitt> Hi @himanshupathak21061998, I don't think we have flatten but its simply resizing the output cube/mat to given number of channels. I think its more related to armadillo than ann.
< kartikdutt18Gitt> Refer the conversation above or pytorch documentation.
< HimanshuPathakGi> Yes it is more related to armadillo but may be we should include this but it is quite simple to change shape
< kartikdutt18Gitt> Yes armadillo has .resize() to simply change the number of channels.
< HimanshuPathakGi> It is lot easier so we don't need it
UmarJ has quit [Ping timeout: 268 seconds]
kpdamania has joined #mlpack
< jenkins-mlpack2> Project docker mlpack nightly build build #605: UNSTABLE in 3 hr 1 min: http://ci.mlpack.org/job/docker%20mlpack%20nightly%20build/605/
UmarJ has joined #mlpack
kpdamania has quit [Remote host closed the connection]
UmarJ has quit [Ping timeout: 260 seconds]
< zoq> HimanshuPathak: Internally, everything is just continuous memory, so if you need to flatten the input/output you could always do something like: arma::Mat(input.memptr(), 1, input.n_elem()), and that will give you a flattern version of the input whitout any speed penalty.
UmarJ has joined #mlpack
UmarJ has quit [Ping timeout: 260 seconds]
< GauravSinghGitte> Hi everyone, during the past few days I was getting acquainted with the mlpack repository and to contribute it by implementing the GELU activation function. Should I go ahead and open an issue and start working on it?
< zoq> GauravSingh: Hello, sounds good please feel free to open an issue and a PR.
< Nakul[m]> @zoq I was thinking to start working on mlpack cmake
< zoq> Nakul[m]: If you are down for a challenge, the bindings part might be tricky.
< kartikdutt18Gitt> Hi @zoq, if you get the chance kindly have a look at PR #38 in models repo, I made the changes you requested in the same.Thanks.
< zoq> kartikdutt18: will do
< Nakul[m]> Here someone doing something for visualization ? Just curious
< Nakul[m]> > Nakul: If you are down for a challenge, the bindings part might be tricky.
< Nakul[m]> Okk
< kartikdutt18Gitt> @zoq, Could you help me out with this https://github.com/mlpack/mlpack/pull/2172/files#r375798353, I can't seem to get testing right
< kartikdutt18Gitt> @zoq, Could you help me out with this https://github.com/mlpack/mlpack/pull/2172/files#r375798353, I can't seem to get testing right
UmarJ has joined #mlpack
UmarJ has quit [Ping timeout: 260 seconds]
ImQ009 has joined #mlpack
< HimanshuPathakGi> Thanks zoq I will try this arma::Mat(input.memptr(), 1, input.n_elem()) next time.
< PrinceGuptaGitte> Hi, I have a question regarding activation functions.
< PrinceGuptaGitte> Currently the deriv() function takes input y then calculates f'(x) where f(x) = y
UmarJ has joined #mlpack
< zoq> PrinceGupta: right
< zoq> PrinceGupta: f'(f(x) )
< zoq> PrinceGupta: I see the comment is confusing.
< kartikdutt18Gitt> Hi @prince776, I think you are right, that's a typo.
< PrinceGuptaGitte> Hi, I implemented ISRU function for issue #2181 but when I built it, it didn't build the new create file isru_function.hpp, am I missing something?
< kartikdutt18Gitt> Hi I think you didn't add in cmakeFile.txt
< kartikdutt18Gitt> src/mlpack/methods/ann/activation_functions/CMakeLists.txt
< PrinceGuptaGitte> oh i totally forgot thanks a lot
< kartikdutt18Gitt> Not an issue, could you also leave a comment on #2181 so that other people don't end up working on the same issue.
< kartikdutt18Gitt> Thanks.
< metahost> Hi everyone, the meeting's live! Here's the link: https://zoom.us/j/3820896170
< kartikdutt18Gitt> Hi @zoq, Could you help me out with this https://github.com/mlpack/mlpack/pull/2172/files#r375798353, I can't seem to get testing right (Sorry to keep pestering you)
< zoq> kartikdutt18: Yes, once I have a chance I will look into it and leave a comment.
AbhinavvermaGitt has joined #mlpack
< AbhinavvermaGitt> libstb-dev showing error while I m trying to install the package through command sudo apt-get install libstb-dev in my 32 bit Linux mint. Do we any alternative for the same.
< kartikdutt18Gitt> Hi @abhiverma1924, could you paste the error log? Thanks.
travis-ci has joined #mlpack
< travis-ci> mlpack/models#219 (master - 384fc5c : Ryan Birmingham): The build has errored.
travis-ci has left #mlpack []
< AbhinavvermaGitt> Error : Unable to locate package libstb-dev
< AbhinavvermaGitt> @kartikdutt18
< zoq> Abhinavverma: Not sure there is a libstb-dev package, isn't it header only? Anyway, if you buld mlpack using cmake, it will download the file for you.
< kartikdutt18Gitt> I think libstb-dev is only available in (very) recent versions of ubuntu (I referred this https://pkgs.org/download/libstb-dev )
< PrinceGuptaGitte> @zoq yea it really confused me about how it's working. I had to look into other activation function files and base_layer to confirm.
< AbhinavvermaGitt> @kartikdutt18 I have already checked this link before there is no version for 32 bit.
travis-ci has joined #mlpack
< travis-ci> mlpack/models#221 (master - 8aeb6a1 : Ryan Birmingham): The build passed.
travis-ci has left #mlpack []
< zoq> PrinceGupta: Yeah, I see, will clrafiy that, or if you like to open a PR, please feel free, but don't feel obligated.
< sreenik[m]> zoq: rcurtin Sorry for the late response. Thanks for the input on the gsoc idea, I'll keep them in mind and modify as soon as I get a chance. There's a presentation tomorrow at college and well, currently I am in the middle of a house party :)
< zoq> sreenik[m]: Sounds good, still a lot of time to do changes; also have fun at the party :)
< sreenik[m]> zoq: Haha yes :)
< rcurtin> :) sounds good to me
< rcurtin> nice to talk to everyone on the video meetup, thanks for joining :)
< pickle-rick[m]> Yeah, fun and informative! Thanks :)
< PrinceGuptaGitte> I have added ISRU activation function. Somone kindly review it and suggest any changes if required. Issue #2181
< metahost> rcurtin: It was fun! :)
ImQ009 has quit [Quit: Leaving]
< HimanshuPathakGi> I was not able to join video meet-up Due to poor internet connection next time I will try join
dendre has joined #mlpack