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
<SimeetNayan[m]> Has this page been updated https://github.com/mlpack/mlpack/wiki/SummerOfCodeIdeas?
<SimeetNayan[m]> I see 2023, but some projects are from 2022 too
<SimeetNayan[m]> Like Example Zoo
<zoq[m]> We are updating the ideas page over the coming year, that said, some of the projects are open-ended, the Example Zoo project is one of them.
<SimeetNayan[m]> So Zoo project would be this year too? What would be the additional thing we would be required to do? Using same dataset on another algorithm?
<zoq[m]> We don't have an example for every method that mlpack implements, so closing this gap is one thing, another thing is to write another example for another method or combination of methods.
<SimeetNayan[m]> mlpack contains deep learning modules too right?
<zoq[m]> Correct, we don't have a Python API for the neural network code, so you have to stick to C++, but we do have some neural network C++ examples.
<zoq[m]> The examples we can add for the neural network codebase are basically endless.
<SimeetNayan[m]> I have extensive experience with both C++ and Machine Learning but I have not used ML in C++ till now. I have been doing ML in Python. I use C++ for development and making Open Source contributions mainly. Do you guys observe any difference while training models in C++ and Python?
<SimeetNayan[m]> Like reduced time for C++? or something like that?
<zoq[m]> Almost all the big nn frameworks use C++ in the backend, PyTorch is slowly transitioning to a Python only codebase. I guess main difference is the interface.
<IWNMWEIWNMWE[m]> I have been running into an error which says that libopenblas.dll not found even though the .dll.a file is there and has been added as an additional dependency
<rcurtin[m]> is this a compilation error or a runtime error?
<IWNMWEIWNMWE[m]> its a compilation error
<IWNMWEIWNMWE[m]> it happens when I try to use the linearreg or decisiontree for training
<rcurtin[m]> okay, maybe libopenblas.dll.a has not been compiled correctly? honestly I am not sure, I am not a regular Windows user...
<zoq[m]> The easiest solution is probably to switch to the windows linux subsystem, if that is a possibility.
<rcurtin[m]> you might also consider taking a look at what we do for CI: https://github.com/mlpack/mlpack/blob/master/.ci/windows-steps.yaml
<rcurtin[m]> maybe there is some subtle difference between the process of how we link against OpenBLAS there and in your setup
<rcurtin[m]> note that if you can build and run mlpack_test (or any command-line program or anything else), those link against OpenBLAS too
<IWNMWEIWNMWE[m]> Ah I forgot to add all the .dll from the to the directory containing the projects .exe as a post build event
<IWNMWEIWNMWE[m]> Ah I forgot to add all the .dll files from openblas native bin to project directory containing the .exe file as a post build cmd in visual studio