andgate / mnist

Handwritten digit recognizer built with pytorch and solid js

MNIST Demo

Simple MNIST demo for machine learning practice.

Model is built with pytorch, and UI is build with solid-js.

Building

Create a python virtual environment with python -m venv env. Activate the environment with

• Windows CMD: .\env\Scripts\activate.bat
• Windows Powershell: .\env\Scripts\Activate.ps1
• Linux/MacOs: source env/bin/activate

Install dependencies with pip install -r requirements.txt. Save modified dependencies with pip freeze > requirements.txt.

Run model.py with python to generate the model.

Use pnpm run build to build the project.

Build output is in the docs folder, which is hosted by github pages.