Skip to content

PEC-CSS/slowmokit

Repository files navigation

slowmokit logo

Slowmokit

ML kit in C++.

Header-only library which can be used for machine learning in C++.

Installation guide

To use this library locally

Prerequisites

Then following command will install the library

./scripts/install

If permission is required to execute, run the following command

chmod 705 ./scripts/install

Compile and run

To compile and run the code written using this library, you need to add the following flags along with the standard commands.

For example, you want to run hello.cpp and it has slowmokit dependencies.

g++ -std=c++17 -larmadillo hello.cpp -o hello && ./hello

Steps to contribute

Following are the steps to start contributing to our beautiful library:

  1. Install an IDE, preferably CLion.
  2. Fork this repo.
  3. Clone the forked repo, using git clone https://github.com/<your_username>/slowmokit.git.

After this,

You can directly use the executable file by running ./bin/nim in the root folder of library, to do the required configurations.

If it does not work then you need to configure everything by your own.

Configuring everything own

  1. Make the model directory you want to implement in the include/slowmokit/<model type>/<your_model>.

    • <model type> is basically the class of your model, for example linear_model or cluster.
  2. Create two files:

    • <model>.hpp : the header file for the model, create the class of model here and include all the stuff required for the model here.
      • Include core.hpp for all the basic functionalities already added.
      • Make sure to add doc comments above each function signature (refer other files for the format)
      • Variable and function names should be in camel case and class names should be in pascal case in all files.
    • <model>.cpp : implement the functions of the class here.
  3. Add the header file <model>.hpp in src/slowmokit.hpp under the documentation block as shown here. More on documentation block.

  4. Test the working by building in IDE.

This step is mandatory

  1. Format the files according to the convention. For simplicity, run ./scripts/clang-format-all include/ in gitbash(if on windows) or in terminal(if on linux/macos)

For this you need to have clang-format binaries installed in the machine.

To do that, run the following in the terminal

# with npm (windows)
npm install -g clang-format
# you have to restart terminal in case of windows

# with homebrew (macos)
brew install clang-format

# linux (Ubuntu)
sudo apt install clang-format

NOTE: Do NOT use #include <bits/stdc++.h> in any file, all the major header files are already included in core.hpp.]

Example

Once the algorithm is ready, run and test it. After testing, push it in the examples directory, with proper comments and instructions. With same directory structure as in include.

Documentation

Push the docs of the model in a <model>.md file in same directory structure in docs folder.

Pull Request (PR) format

After code is ready, you can make PR to the main branch. PR will be accepted only if:

  • Code pass the CI tests.
  • Example is pushed.
  • Code has proper comments and instructions.

GOOD LUCK!