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/
blakjak888 is now known as BlakJak888
ImQ009 has joined #mlpack
< say4n> rcurtin, zoq: when reviewing PRs, do you do it on the website or checkout locally and then review, or use something like the github cli? :)
BlakJak888 is now known as BlakJak888[m]
BlakJak888[m] is now known as BlakJak888
BlakJak888 is now known as BlakJak888[m]
BlakJak888[m] is now known as BlakJak888
BlakJak888[m] has joined #mlpack
< HavshxjdnaggzGi4> How can I print the content and the leaves of the tree in KNN?
< HavshxjdnaggzGi4> Also, if KNN is a pairwise then why there is a tree ?
< HavshxjdnaggzGi4> Please, help is needed here.
< zoq> say4n: For me that depends on the PR, I usually start on the website, if the PR is complex or introduces a new feature I check it out locally build and test it. Gettting some intermediate output is often helpful to understand the logic as well, which I only get if I build it locally.
BlakJak888[m] has quit [Ping timeout: 256 seconds]
< say4n> Ah, I see.
< say4n> Also, I was thinking about introducing some basic linter checks as CI tasks in ensmallen too?
< zoq> say4n: Do you have anything special in mind? Because we have the Static Code Analysis Job, the Memory Job and I think the style check as well.
BlakJak888[m] has joined #mlpack
BlakJak888 has quit [Read error: Connection reset by peer]
blakjak888 has joined #mlpack
< say4n> Do we have the code style check enabled for ensmallen too? I think it is only enabled for the mlpack repo. I was thinking that maybe having checks for code style could help reduce some to and fro interaction in submitted PRs regarding code style. Or, better yet a local codes style check script before committing any changes (as a pre commit hook or as an explicit script manually executed before a commit).
< say4n> zoq: ^
< say4n> these are the checks I see on a PR in ensmallen.
< AakashkaushikGi4> Hey @say4n checkout https://github.com/mlpack/mlpack/issues/2585 I was thinking to work on this for mlpack, i think something similar could be implemented for ensmallen ?
< AakashkaushikGi4> Can we have a discussion about it ?
kgk has joined #mlpack
kgk has quit [Read error: Connection reset by peer]
< abernauer[m]> The bindings are installing right now and one of my immediate concerns were not an issue so that's a positive.
kgk has joined #mlpack
< abernauer[m]> yashwants19: rcurtin The R bindings didn't build on my system using the artifacts either.
< abernauer[m]> Might be an issue with my R installation.
kgk has quit [Read error: Connection reset by peer]
< abernauer[m]> Ok installed it looks like.
< rcurtin> shrit[m]: I thought we did VS15 and VS16, which corresponds to Visual Studio 2017 and Visual Studio 2019 respectively
< rcurtin> so I am not sure there is anything newer that we can upgrade to
< zoq> say4n: Let me active the style check job.
< rcurtin> if the only thing that we can do is drop VS15, I guess I am not opposed if it is the only choice
< zoq> say4n: You can already run cpplint on your local machine - https://github.com/mlpack/jenkins-conf/tree/master/linter
< rcurtin> zoq: any thoughts about dropping VS15 for cereal, since we can't seem to get it to build? or should we keep trying to figure out how? we would still have VS16 building successfully
< zoq> rcurtin: No not really, did you already tested to disable cotire?
< zoq> say4n: Not a huge fan of automated style jobs, but if there is a good soltution.
< abernauer[m]> zoq: Any guidelines or limitations on the packages I can use to rewrite examples for the R bindings?
< zoq> abernauer[m]: If you have to install a package just add it to - https://github.com/mlpack/examples/blob/master/binder/postBuild I would use conda to install it to make sure the deps are correct.
< zoq> abernauer[m]: I can install the R kernel if you need help there.
< zoq> abernauer[m]: Other than that you can use http://lab.mlpack.org/ to build/start your repo.
< zoq> abernauer[m]: Or https://mybinder.org/
< abernauer[m]> Ok I will use Binder as that seems like the most reproducible way for others to run the code. Yeah I have Jupyter on my system not sure if it has the R kernel installed.
< rcurtin> shrit[m]: maybe you can try disabling cotire as a last resort to see what that does?
< abernauer[m]> zoq: I can fork the examples repo or use lab.mlpack.org which ever makes it easiest. I was going to start with the movie lens example.
< zoq> abernauer[m]: Whatever works for you the best, you can write the example locally first and push it, and once it's there we can figure out the rest.
< abernauer[m]> ok sounds cool
< zoq> rcurtin: no problem running the accu benchmark on my two nvidia systems, will test AMD as well.
ImQ009 has quit [Quit: Leaving]
< abernauer[m]> zoq: Just noticed that random forest and cf are covered in the binding documentation on the mlpack website. I could expand on those examples e.g. write some additional code to visualize the data set like your Python example for cf.
< zoq> abernauer[m]: Sounds like a great idea to me.
< rcurtin> zoq: awesome, let's see what the AMD card will do :-D
< shrit[m]> rcurtin: I will try to disable cotire and see what it does. The issue is that if we merge cereal and VS15 is still failing, mlpack build for VS15 will be always failing
< shrit[m]> If I disable cotire and its work does that mean we need to disable cotire definitely?
< shrit[m]> Also I thought that VS15 is from 2015 :D
< abernauer[m]> Getting familiar with some of the NLP and text mining packages for this example in R.
< abernauer[m]> :)
< zoq> abernauer[m]: Excited to see the result, personally I really like to work on some examples.
< abernauer[m]> Yeah I like learning about theory and find it interesting, but actually getting to apply it on projects or data science examples is fun.