cloneofsimo/minDiffusion resource

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1196 stars · 152 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A self-contained, minimalistic PyTorch implementation of denoising diffusion probabilistic models (DDPM) intended for education. The core implementation fits in under 200 lines of code and includes training scripts for MNIST, CIFAR-10, and CelebA datasets.

Use cases

• learn how diffusion models work from a minimal codebase
• understand DDPM training and sampling in PyTorch
• generate images with a simple diffusion model
• study U-Net and time embedding implementations for diffusion
• get a starting point before reading larger diffusion codebases
• train a small diffusion model on CIFAR-10 or MNIST

When to choose

• you want to learn the internals of diffusion models with minimal code
• you need a compact, readable DDPM reference implementation
• you are teaching or studying generative diffusion models
• you want a quick baseline before moving to larger frameworks like Hugging Face diffusers

When to avoid

• you need production-ready or feature-complete diffusion tooling
• you require DDIM, classifier guidance, or latent diffusion
• you want a maintained library with a license and active releases
• you need efficient large-scale training or high-resolution image generation

Facets

learning-resource · maturity maintenance

machine-learning deep-learning machine-learning deep-learning artificial-intelligence tutorials python diffusion-models ddpm educational pytorch image-generation gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="cloneofsimo/minDiffusion")