huggingface/deep-rl-class

This repo contains the Hugging Face Deep Reinforcement Learning Course.

What it is
A free online course that teaches Deep Reinforcement Learning, hosted by Hugging Face. The course combines theory lessons with hands-on coding exercises (notebooks).

What it contains

  • Course materials: Markdown files and Jupyter notebooks for each unit.
  • Website: The full course is available at huggingface.co/deep-rl-course/unit0/introduction.
  • Syllabus: A separate page lists the topics covered.
  • Community support: The repository’s issue section has solutions to common bugs in the exercises.

Current status
The course is in a low-maintenance state — it’s still a good learning resource, but two features are no longer working:

  • Unit 7 (AI vs AI): The multiplayer feature is non-functional, but you can still train an agent to play soccer and watch it perform.
  • Leaderboard: No longer operational.
    All other theory and practical exercises remain fully accessible.

How to use it

  • Sign up for updates via the link in the README.
  • Star the repository to support the project (optional).
  • Access the course directly through the provided website link.

Citation
If you use the course in academic work, the README provides a BibTeX entry to cite it.

Related

  • Project
    sktime/pytorch-forecasting

    A PyTorch-based library for time series forecasting that provides a high-level API and state-of-the-art deep learning architectures like Temporal Fusion Transformers and N-BEATS.

  • Project
    Nixtla/statsforecast

    A high-performance library for univariate time series forecasting that provides fast, scalable implementations of statistical and econometric models.

  • Project
    tencentmusic/cube-studio

    Cube Studio is an integrated development environment for data science and AI, focusing on providing consistent environments across the development lifecycle.

  • Project
    humphd/have-fun-with-machine-learning

    A hands-on, beginner-friendly guide for programmers to build and train image classification models using the Caffe framework and DIGITS without needing a background in AI or expensive hardware.