OussamaMater / Model-Builder

A Machine Learning Models Builder and Manager.

Project Setup Guide

Overview

A Machine Learning Models Builder and Manager built with Laravel, Livewire, Tensorflow and FastAPI, providing a set of features:

• A login system.
• The capability to create, manage, and utilize multiple datasets.
• The ability to train diverse models associated with various datasets asynchronously.
• Real-time notifications for ongoing model training or readiness for use.
• Instant notifications upon results becoming available.
• Comprehensive logs that showcase received results, along with search, filter, and deletion functionalities

Manual Setup

You can avoid this configurational step by using the Docker installation process.

Before setting up the project, make sure you have the required dependencies:

• Python >= 3.10
• PHP >= 8.1
• pip
• composer
• npm

Installing Dependencies

The following example works for Debian Based Linux Distributions.

Update the system

sudo apt update && sudo apt upgrade -y

Install Python

sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt install python3.10

Install pip

sudo apt install python3-pip

Install PHP

sudo apt get install -y php8.1-cli php8.1-common php8.1-mysql php8.1-zip php8.1-gd php8.1-mbstring php8.1-curl php8.1-xml php8.1-bcmath

Install composer

curl -sS https://getcomposer.org/installer -o /tmp/composer-setup.php

sudo php /tmp/composer-setup.php --install-dir=/usr/local/bin --filename=composer

Install npm

sudo apt install nodejs npm

Setting the Web Application

Make sure you are in the web-app directory

cd src/web-app

Install composer dependencies

composer install

Install npm dependencies and build the assets

npm install
npm run build

Create the .env file and generate the application encryption key

cp .env.example .env # Don't forget to change database credentials 
php artisan key:generate

Update the package.json file by swapping the scripts section with the one below

"scripts": {
    "dev": "vite",
    "build": "vite build"
},

Setting the Model

There are at least two ways to set up Python dependencies: Globally or in a Virtual Environment.

Via a Virtual Environment

Note

This method is recommended if you want to isolate the dependencies.

python3 -m venv venv

source venv/bin/activate

pip install --require-virtualenv --no-cache-dir --upgrade -r dockerfiles/ml-model/requirements.txt 

Globally

Note

This method is recommended if you are not using any other machine learning projects.

pip install --no-cache-dir --upgrade -r dockerfiles/ml-model/requirements.txt

Running the project

Start the model

cd src/ml-model

uvicorn home:app --host 127.0.0.1 --port 5000 --reload

Start the built-in web server

cd src/web-app

php artisan serve

If you want to expose the dev server to any other than localhost, you can use the following command

php artisan serve --host <your-ip> --port <your-port>

To expose the server to the world

Note

Requires port forwarding or firewall configuration.

php artisan serve --host 0.0.0.0 --port <your-port>

However Most of the features are available at /admin and to access it you must create a user using the following command:

php artisan make:filament-user --email yourmail --password yourpass --name you

Docker Setup

Make sure Docker is installed.

If you are using a Windows 10/11, you need to know that Windows uses WSL (Windows Subsystem for Linux), which is a layer between Windows and Linux, this makes Docker slow and browser requests may take 30-60 seconds to be completed.

Spin up the containers

docker-compose up -d --build app

Running the command will expose 3 services with the following ports:

• Nginx - :80
• PHP - :9000
• FastAPI - :5500

Install composer dependencies

docker-compose run --rm composer install

Install npm dependencies and build the assets

docker-compose run --rm npm install
docker-compose run --rm --service-ports npm run build

Create the .env file and generate the application encryption key

cp .env.example .env
docker-compose run --rm php artisan key:generate

You can now access the server at http://localhost.