A see-through automatic differentiation library.
This library is my effort to understand reverse automatic differentiation, most famously used in TensorFlow. I have used nothing but numpy and written everything from scratch (see credits). The end goal of this exercise is to understand autodiff and write usable code with reasonable performance.
The initial version of this library is almost completely based on Jonathan Kernes's amazing blog Build Your Own Automatic Differentiation Program. Do check out his blog!
Feel free to open a PR for small changes. For medium to large sized changes, consider opening an issue first to discuss regarding the same.