Over the past month, what started as a school project has turned into a custom-built deep learning library called pieplatter that I’ve been building from scratch. It’s been quite the ordeal, and I thought I’d document some of of the build process here.
I’ve been working hard to put the library together and get some degree of functionality. As it currently stands, it’s equipped to create small neural networks for simple problems, but I’m still adding more support and optimizations so that hopefully much bigger and more powerful networks can be created in the future.
Right now, I still haven’t put any docs together for the library, but I’ll link those when I do. Until then, I wanted to put together a fairly large series of posts to document the entire process of coding a neural network from scratch. They’ll run through all the theory, programming, and math so you can really get a good grasp on how neural networks work under the hood and how to create some simple networks from the ground up.
If you want, you can check out the first of many posts in the series here.
You can also check out my deep learning library here on github. It’s called pieplatter.