Hugging Face's Learn platform now lists 12 free courses spanning the major domains of applied AI. The lineup ranges from introductory LLM work to specialized tracks on robotics, diffusion models, and Anthropic's Model Context Protocol.
The MCP course, built in partnership with Anthropic, covers the protocol that lets AI models connect to external tools and data sources. It includes hands-on assignments using Hugging Face's hosted MCP server and offers a completion certificate. The robotics course uses LeRobot, Hugging Face's open-source library for robot learning, walking students through imitation learning and reinforcement learning approaches. Both courses are self-paced with community support through Discord.
A smaller track called "smol-course" targets developers who want to skip the fundamentals. It focuses on fine-tuning using TRL and Transformers, covering supervised fine-tuning and preference alignment in roughly four hours per week. The course uses SmolLM3, a 3-billion parameter model designed for accessibility.
Other offerings include deep reinforcement learning (using Stable Baselines3 and CleanRL), computer vision, audio processing, and ML for games. Dylan Ebert, a Hugging Face developer advocate, created the ML for 3D course, which covers gaussian splatting and generative 3D techniques. Most courses offer certificates upon completing assignments and quizzes.
The Bottom Line: Hugging Face](https://aihola.com/article/how-to-use-hugging-face-guide) has quietly assembled one of the more comprehensive free AI curricula available, with the MCP and robotics additions filling gaps that most competitors haven't touched.
QUICK FACTS
- 12 courses available at huggingface.co/learn
- MCP course developed with Anthropic
- Robotics course uses LeRobot library
- Certificates available on most tracks
- All courses free and open-source
