A tiny scalar-valued autograd engine and a neural net library in Zig. It is inspired by the micrograd project from Andrej Karpathy.
- Youtube video: The spelled-out intro to neural networks and backpropagation: building micrograd.
- micrgrad on GitHub.
I wrote it in Zig for education purpose, and for fun of course. :)
Examples can be found in the examples folder.
- expression.zig: a backpropagation on an expression.
- tiny_dataset: a neural network on a tiny dataset.
To run them:
% zig build run -Dexample=expression
% zig build run -Dexample=tiny_dataset% zig build test --summary allMIT