rcurtin_irc changed the topic of #mlpack to: mlpack: a scalable machine learning library (https://www.mlpack.org/) -- channel logs: https://libera.irclog.whitequark.org/mlpack -- NOTE: messages sent here might not be seen by bridged users on matrix, gitter, or slack
<ShubhamAgrawal[m> <zoq[m]> "then activate the conda env, and..." <- I have already installed mlpack
<ShubhamAgrawal[m> Do I need to install it again in environment or just include it in xeus-cling file?
<zoq[m]> <ShubhamAgrawal[m> "Do I need to install it again in..." <- I would recommend to install it again, to make sure it links against the right dependencies.
<ShubhamAgrawal[m> <zoq[m]> "I would recommend to install..." <- And how do i point cling load to correct location?
<zoq[m]> https://github.com/mlpack/examples/blob/master/binder/xeus-cling.hpp is included at the top of each notebook, you can adjust the path in the file itself, or copy the lines into the notebook.
<ShubhamAgrawal[m> What will I put in this? DCEREAL_INCLUDE_DIR=/usr/include/
<zoq[m]> You can install cereal from conda-forge -- https://anaconda.org/conda-forge/cereal
<zoq[m]> Which is autocratically detected by CMake.
<ShubhamAgrawal[m> ok
<ShubhamAgrawal[m> cp: cannot create regular file '/root/miniconda3/envs/mlpack/include/mlpack/': Not a directory
<ShubhamAgrawal[m> Error coming up in bash file
<zoq[m]> Looks like you installed minconda inside the root folder, so you might need to use `sudo` or switch the user that has permissions to write into `root/`.
<ShubhamAgrawal[m> zoq[m]: I think issue is I removed mlpack as dep
<ShubhamAgrawal[m> Then I tried to do this step
<ShubhamAgrawal[m> So I think I need to first install mlpack
<ShubhamAgrawal[m> Then try this step
<ShubhamAgrawal[m> zoq: I am unable to use mlpack
<ShubhamAgrawal[m> And other libraries don't seems to work
<zoq[m]> If you can help me a little but here that would be great, like does the xeus-cling kernel work in general?
<zoq[m]> Do you get any error message.
<ShubhamAgrawal[m> zoq[m]: jupyter lab opens
<ShubhamAgrawal[m> But I am unable to include iostream too
<ShubhamAgrawal[m> Showing some error with armadillo
<zoq[m]> so you are saying `cout << 1;` does not work either?
<ShubhamAgrawal[m> zoq[m]: #include <iostream> shows error
<zoq[m]> If you can post the error message, I guess that will show us something.
<zoq[m]> s/show/tell/
<ShubhamAgrawal[m> But I need to open xeus-cling in docker for exact error
<zoq[m]> Okay, so this looks like it's not using the C++ libs from the conda env.
<ShubhamAgrawal[m> But I am using docker container and it is not possible to install from other sources as far as I know
<zoq[m]> yeah, but you still use conda inside the docker?
<ShubhamAgrawal[m> zoq[m]: For xeus-cling, you need conda right?
<ShubhamAgrawal[m> Or is there another way?
<zoq[m]> You don't need it, but it would make things easier.
<zoq[m]> You have your system c++ libs and your conda ones.
<zoq[m]> So I think what happend here is that it picked up the system ones.
<ShubhamAgrawal[m> zoq[m]: Is there any way to use xeus-cling with system compiler
<ShubhamAgrawal[m> Can you tell me how to use xeus-cling in docker
<zoq[m]> It's not a good idea to mix them.
<zoq[m]> What docker are you using?
<ShubhamAgrawal[m> zoq[m]: Linux container
<zoq[m]> Right, base image is what I meant.
<ShubhamAgrawal[m> ubuntu:latest
<zoq[m]> Okay, I'll take a look into the docker setup later today.
<ShubhamAgrawal[m> Ok thanks
<rcurtin[m]> Shubham Agrawal: out of curiosity, are you planning to use a notebook interface for your development this summer? or are you just going to use it for testing / running existing examples / etc.?
<ShubhamAgrawal[m> rcurtin[m]: Right now, I have 2 type of env
<ShubhamAgrawal[m> 1st is dev container with mlpack installed and using vscode inside
<ShubhamAgrawal[m> 2nd is xeus-cling notebook
<rcurtin[m]> if you'll do actual development inside the notebook (like writing library functions, etc.), let us know how it works out... I don't know if many people are doing that, so it would be interesting to know if you have any issues, or if it works well
<ShubhamAgrawal[m> rcurtin: Can you see the Concat layer and update if anything changed from last time?
<rcurtin[m]> I have a couple local changes but didn't manage to debug everything over the weekend... let me push what I have
<rcurtin[m]> pushed; `Forward()` and `Backward()` appear to work, but I am still debugging `Gradient()`
<ShubhamAgrawal[m> And should I typedef MultiLayer to Sequential?
<ShubhamAgrawal[m> And I am also thinking to convert LambdaMapReduce to Residual layer
<ShubhamAgrawal[m> As Concat is already being developed by you
<rcurtin[m]> we should remove references to `Sequential` entirely instead of using a typedef
<rcurtin[m]> yeah, that should work fine
<ShubhamAgrawal[m> rcurtin[m]: Ok
<ShubhamAgrawal[m> I will just use MultiLayer as it is
<rcurtin[m]> up to you if you want to implement some concatenation functionality in `LambdaMapReduce`; like I suggested before, these two layers are probably temporary anyway
<ShubhamAgrawal[m> ?
<ShubhamAgrawal[m> rcurtin[m]: Ok