This project is a machine learning framework that includes various modules for data loading, model training, evaluation, and visualization. The framework is designed to be similar to Pytorch, but it is based on Numpy. The goal of this project is to help people understand the underlying mechanism of deep learning frameworks and to provide a simple and easy-to-understand framework for educational purposes.
To install the required dependencies, run:
pip install -r requirements.txt
Run hand-written digits recognition demo at demo/train_mlp.py
.
cd demo
python train_mlp.py
The logging system is inspired by:
https://github.com/gmberton/deep-visual-geo-localization-benchmark