This repository contains my own code working through Andrej Karpathy's "Neural Networks: Zero to Hero".
- Each folder represents one of the lessons.
- Instead of building the language model with baby names like in the series, I used a dataset with Dungeons and Dragons character names. This dataset was sourced from janelleshane/DnD-characters. Thank you!
I can highly recommend the course. Instead of starting by using huge libraries to do the heavy lifting like in so many other courses, the course teaches you the fundamental math and processes by building them yourself from scratch.
This takes a bit longer to eventually get a working model, but: the fundamentals never change, no matter the size or even type of model. Additionally, the course itself is very well done and clearly explained... and it's free! I'd have paid for it.