A small autodiff and neural network library patterned on Karpathy's micrograd, but with support for matrics and vectors expressed as numpy arrays.

Written for my own clarity, but I hope it helps you too.

• Install with pip install -e . The only runtime dependency is numpy.
• Test with pytest.
• main.py runs a little demo of an MLP doing regression against random data.
• nbs/demo.ipynb is Karpathy's original demo notebook, adapted to use my library