The Complete Hugging Face Learn Guide: 12 Courses to Run, Fine-Tune, and Deploy Models Yourself

Hugging Face's Learn section has quietly grown to 12 courses this year — and half of them are brand new.

This sets it apart from every other platform covered so far. While others teach you "how to use AI," HF teaches you how to run, fine-tune, and publish models yourself. You walk away with something real to show for it.

Main portal: huggingface.co/learn

Here are the 10 courses most worth your time.

1. LLM Course (formerly the NLP Course — fully rewritten this year)

🔗 huggingface.co/learn/llm-course

Covers everything from "what is a transformer" all the way to fine-tuning and deploying to the Hub. Completely rewritten this year — this is the backbone of HF Learn.

2. Agents Course (new this year — the hottest one)

🔗 huggingface.co/learn/agents-course

An end-to-end course on building agents, featuring three full implementations: smolagents / LlamaIndex / LangGraph. The most popular course this year — essential if you're building agents.

3. MCP Course (new this year)

🔗 huggingface.co/learn/mcp-course

HF's official MCP tutorial, complementing Anthropic's MCP course — this one focuses on integration within the HF ecosystem.

4. smol course (new this year — lean and focused)

🔗 huggingface.co/learn/smol-course

Dedicated entirely to post-training — the most distilled course on fine-tuning large models available. If you're short on time, start here.

5. Robotics Course (new this year — uses the LeRobot library)

🔗 huggingface.co/learn/robotics-course

Your entry point into AI + robotics — a domain none of the other nine platforms even touch.

6. Diffusion Course

🔗 huggingface.co/learn/diffusion-course

A must if you're looking to build on top of Stable Diffusion or Flux.

7. Computer Vision Course (community collaboration)

🔗 huggingface.co/learn/computer-vision-course

Covers ViT, SAM, and DINO — much more current than a traditional CV course.

8. Audio Course

🔗 huggingface.co/learn/audio-course

The go-to resource if you're working on speech, music generation, or TTS.

9. Deep RL Course

🔗 huggingface.co/learn/deep-rl-course

A classic introduction to reinforcement learning, paired with HF's own training environments.

10. Open-Source AI Cookbook

🔗 huggingface.co/learn/cookbook

A collection of hands-on notebooks. After working through the first nine courses, come here to grab production-ready code.

HF's One Killer Feature

Every course can be run directly on Spaces — no local setup required, and HF provides the GPU.

Not one of the other nine platforms can say the same. If you don't have a local GPU, this single fact is enough to put HF at the top of your list.

My Study Recommendations

Start with the LLM Course: Fully rewritten this year — it's the foundation for understanding the entire HF ecosystem
Make the Agents Course your top priority: The hottest area this year, covering three agent implementations in one place
The Robotics Course is one-of-a-kind: If you want to break into AI + robotics, this is a rare entry point
Save the Cookbook for last: Build up your skills with the first nine courses, then use the Cookbook to grab real-world code
Take advantage of Spaces: You can complete every course without a GPU — that's HF's biggest edge

FAQ

Can I take Hugging Face courses without a GPU?

Yes — and this is HF's signature advantage: every course runs directly on Spaces, no local environment needed, with GPU provided by HF. None of the other nine platforms offer anything like it.

Which Hugging Face course should I start with out of the 12?

Start with the LLM Course (completely rewritten this year, it's the foundation for understanding the entire HF ecosystem), then move on to the Agents Course — the most popular course this year, covering smolagents / LlamaIndex / LangGraph in a single place.

How is HF fundamentally different from other learning platforms?

Other platforms teach you "how to use AI." HF teaches you "how to build AI" — running models yourself, fine-tuning them, and publishing them. You finish with something real in your hands. Almost everyone who's serious about open-source LLMs started here.

The One-Sentence Summary

Hugging Face Learn doesn't teach you to use AI — it teaches you to build it. 12 courses spanning LLMs, agents, fine-tuning, and robotics, all runnable on Spaces for free (GPU included). If you're serious about open-source LLMs, this is where you start.

🎬 All 10 Platforms Covered — Two Years of AI Learning, Completely Free

That wraps up all 10 AI learning platforms. If you want a structured path from zero to engineer-level skills you can put on a résumé, the combination below covers the full journey — entirely for free:

→ The Complete Claude Certified Architect Roadmap (Anthropic) → OpenAI Academy vs. Anthropic: Full Comparison (OpenAI) → Google Skills: Gamified AI Learning Paths (Google) → The Complete Guide to Free NVIDIA DLI Courses (NVIDIA) → The Complete Guide to Microsoft AI Certifications (Microsoft) → Free AI Courses on AWS Skill Builder (AWS) → The Complete Guide to Meta's PyTorch Tutorials (Meta) → Free AI Courses on IBM SkillsBuild (IBM) → The Complete DeepLearning.AI Course Selection Guide (DeepLearning.AI)

Bookmark this post and work back through the previous nine — the full path is all there. Every bit of AI learning you'd otherwise pay for over the next year or two? You won't need to spend a cent.

📚 The Complete Free AI Learning Map Across 10 Platforms → One table, role-based learning paths, and the full series index.