huggingface/diffusion-models-class

Materials for the Hugging Face Diffusion Models Course

What it solves

For people who want to understand and build diffusion models — the technology behind modern image and audio generators — this free course provides a structured path. It solves the problem of scattered resources by combining theory, practical notebooks, and community support, so learners can go from basics to training their own models.

How it works

The course is split into at least four units, each covering a specific topic. Every unit mixes theoretical background with hands-on Jupyter notebooks that use the Hugging Face Diffusers library. You start with an introduction to diffusion models and Diffusers, then move to fine-tuning and adding guidance, explore Stable Diffusion, and finally dive into advanced techniques. Along the way, you train models from scratch, fine-tune them on new datasets, and create custom pipelines. The README also mentions competitions and awards for the best pipelines, and the course is being translated into multiple languages by the community.

Who it’s for

Learners with good Python skills and basic knowledge of deep learning and PyTorch. The course is completely free and requires a Hugging Face account only if you want to push your custom models and pipelines to the Hub.

Highlights

  • Free and open to everyone
  • Covers theory, training, fine-tuning, and guidance
  • Hands-on notebooks using the popular Diffusers library
  • Community support via Discord and a signup form
  • Multiple language translations contributed by the community
  • Planned competitions and swag for the best pipelines and demos

Related

  • Project
    ourownstory/neural_prophet

    An interpretable time series forecasting framework built on PyTorch that combines neural networks with traditional time-series algorithms.

  • Project
    Nixtla/neuralforecast

    NeuralForecast is a Python library that provides a large collection of modern neural‑network models for time‑series forecasting, with a scikit‑learn‑like API, support for exogenous variables, probabilistic outputs, hyper‑parameter tuning, and safe model saving.

  • Project
    limix-ldm-ai/LimiX

    LimiX is a foundation model for structured tabular data that performs classification, regression, and missing-value imputation without task-specific training.

  • Project
    microsoft/FLAML

    A lightweight Python library for efficient automated machine learning and hyperparameter tuning that optimizes model performance under resource constraints.