This repo implements micrograd as a WL paclet. It's a work in progres.
I recently started watching the nn-zero-to-hero series by @karpathy. I just finished the first video building micrograd from scratch and thought the natural exercise for me was to try implementing it in the WL.
One can call python from the WL, but the purpose of this paclet is to implement micrograd in a WL style, and compare that implementation with the WL Net* functions, similar to how Andrej compares micrograd functionality to pytorch.