Exploring Diffusion Models: A Hands-On Approach with MNIST
In this project, I dive into the world of diffusion models by training a Denoising Diffusion Probabilistic Model (DDPM) and a Latent Diffusion Model (LDM) from scratch on the MNIST dataset.
This code was derived from was initially referenced from a [notebook](was referenced from a notebook provided in the article, “Understanding Stable Diffusion from Scratch”, a resource made available by Harvard’s “Machine Learning from Scratch” seminar series.) provided in the article, “Understanding Stable Diffusion from Scratch”, a resource made available by Harvard’s “Machine Learning from Scratch” seminar series.
For full analysis, see this blog