Inspired from MagicalBat, I built a Machine Learning library in C that I eventually want to turn into a GPT
Project link - https://github.com/AalbatrossGuy/Teddy
Project Screenshots - https://pastes.vargoseus.com/TeddyScreenshots
Teddy (cute name, isn't it?) is currently a simple machine learning model that uses back propagation to train, learn and classify MNIST datasets. It currently has a depth of 2 since it's a pretty basic model. It has around 13.000 parameters that is enough for training it to recognize handwritten digits. I have around 3.5-4 years of experience working in C and stumbling upon MagicalBat's this video inspired me to make Teddy. The future plan is to turn it into a Language Model and eventually into a GPT which will require quite a bit of time since I need to read up on how it actually works. This project took around 4-5 months give or take since I had to balance this project and my university stuff too.
Full disclaimer: I did not use AI to build Teddy (except for that one time when my compiler suddenly stopped working for some reason and I had to converse back and forth with Claude to find a fix for it). I did, however, use AI to generate the GIFs in the github readme and the documentation for it.