huggingface/course

The Hugging Face course on Transformers

What it is

The huggingface/course repository holds the source content for the Hugging Face course — a free, open-source educational resource. The course teaches how to apply Transformer models to tasks in natural language processing (and other fields), and how to use the Hugging Face ecosystem: the Transformers, Datasets, Tokenizers, and Accelerate libraries, plus the Hugging Face Hub (a platform for sharing models).

What the course covers

  • Using Transformer models for NLP tasks and beyond
  • Working with the Hugging Face libraries and Hub
  • All content is free and open-source

Languages and translations

The course is available in many languages, including English, French, German, Spanish, Bengali, Persian, Hindi, Japanese, Korean, Russian, Chinese (simplified and traditional), and more. Most translations are marked as work in progress (WIP); English and French appear complete. Each language has its own folder under chapters/ and its own set of authors.

How to contribute a translation

The README gives a step-by-step process for volunteers:

  1. Open an issue on the repo to check if your language is already being worked on, and claim chapters.
  2. Join the Discord server for quick discussion with other translators.
  3. Fork the repository and clone it locally.
  4. Copy the English files into a new language folder (using an ISO language code).
  5. Translate the content — the MDX files and the _toctree.yml table of contents. Only include sections you've actually translated.
  6. Build locally with the doc-builder tool to preview how the course will look.
  7. Submit a pull request after running the style checks, tagging the maintainers for review.

Jupyter notebooks

The code from the course is hosted in a separate repository (huggingface/notebooks). You can also generate the notebooks locally by running a script (utils/generate_notebooks.py) that extracts code snippets from the chapters.

Contributing new chapters

The project is not currently accepting community contributions for new chapters — those instructions are only for the Hugging Face authors themselves.

Related

  • Project
    53AI/53AIHub

    An open-source AI portal that allows users to quickly build and operate a production-grade hub for AI agents, prompts, and tools by integrating with existing AI platforms.

  • Project
    Adkid-Zephyr/anti-defensive-writing-Skill

    A collection of AI prompts and agent skills designed to help researchers rewrite academic papers by eliminating defensive, self-doubting language and focusing on clear, persuasive arguments.

  • Project
    QwenLM/Qwen3.8

    A series of open-weight large language models (Qwen3.5, 3.6, and 3.8) that provide high-performance capabilities in coding, professional research, and autonomous agentic tasks.

  • Project
    Chevey339/kelivo

    Kelivo is an open‑source, Flutter‑based LLM client for Android, iOS, macOS, Windows and Linux. It lets you add any major LLM provider (OpenAI, Gemini, Claude, etc.) via API keys or account sign‑in, store chats and files locally, and optionally back them up to WebDAV/S3. A standout feature is *workspaces*: a sandboxed Linux environment (PRoot on Android, iSH on iOS) or native shell on desktop where the model can read/write files, run commands, and use tools like web search, image generation, speech‑to‑text, and custom “skills”. The app includes long‑term memory extraction, multilingual UI, theming, and extensive privacy controls, all under the AGPL‑3.0 license.