omnigrad 0.2.1

A scalar-valued autograd engine, consider this like an extension of micrograd with support of more complex functions

# omnidiff This project is a lightweight Python library that implements reverse-mode automatic differentiation. It is designed as a learning-oriented extension of Andrej Karpathy’s Micrograd, providing support for a broader range of mathematical operations and functions.

ref: https://github.com/karpathy/micrograd

Similar to Micrograd, this library operates exclusively on scalar values, enabling a clear and intuitive understanding of the underlying mechanics of automatic differentiation.

In addition to computing gradients, you can use this library alone to construct and train neural networks, with a particular focus on binary classification tasks.