General Neural Network training platform for classification or regression

This projects represents a basic playground for deep neural network training for learning purposes of the LEGO AI Team. It aims to use modern solutions (pytorch, lightning) with easy to use on a flexible platform.

Install

git clone repo
cd repo
conda create -n nnet python=3.10
pip install -r requirements.txt

Usage

The nn folder contains the model config files, and a basic implementation of a deep neural network. The scripts contains the config of the training methods (the data augmentations, epoch, early stopping, and so). It will have separate scripts for training, testing, etc.

The tutorial.ipynb contains a demo of usage.

Contribution

Feel free to contribute, report issues or request features!