Attempt at making a simple diffusion model from scratch in PyTorch.
Almost everything works (forward noising, denoising model, reverse noising) except for actually generating coherent images.
My first experience with working with PyTorch, so although the diffusion model doesn't generate anything useful I learned how to work with PyTorch.
A quickly runnable notebook is availalbe on Kaggle: https://www.kaggle.com/code/baconerie/simple-diffusion
Gave helpful information about implementing a diffusion model