The NLP Course is becoming the LLM Course
Hugging Face
Hugging Face is renaming its NLP course to the LLM Course and expanding it big time. Classic NLP stuff stays, but new chapters on fine-tuning, reasoning models, and outside tools are coming.
Based on reporting by Hugging Face — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Hugging Face is giving its three-year-old NLP course a rebrand and a serious glow-up. The course, long a staple for anyone getting into open-source AI, is becoming the LLM Course, a name that better fits what it's actually teaching these days.
The shift didn't happen overnight. Over the past several months, the team quietly added chapters on fine-tuning large language models and building reasoning models like Deepseek R1, content that never really fit under the old NLP label. Add to that the momentum from the smol-course experiment and the agents course, which pulled in 100,000 students eager to learn about AI agents, and the case for a rename became obvious.
Classic NLP isn't getting tossed out, though. Chapters on classification, named entity recognition, and retrieval are sticking around and getting modernized with things like Sentence Transformers, updated zero-shot classification methods, and ModernBert. Hugging Face is making the point, correctly, that not every problem needs a giant language model thrown at it. Sometimes a smaller, interpretable model that runs locally is the smarter call.
What's new is the direction of the added material. Chapters 11 and 12, covering LLM fine-tuning and reasoning models respectively, were built in collaboration with LlamaIndex, LangChain, Unsloth, Marimo, and ML educator Maxime Labonne. That's a deliberate move away from a course that only teaches Hugging Face's own tooling. The transformers library remains the reference point for how LLMs get built, but people fine-tune and deploy models across a much wider ecosystem, and the course wants to reflect that reality instead of pretending it doesn't exist.
Interactive exercises and live sessions aren't disappearing, but they're being reserved for topics where real-time engagement actually helps. The bulk of the content will stay written and self-paced, the kind of material that doesn't go stale the moment a new model drops. Hugging Face is inviting the community to weigh in through its Hub organization if people want to suggest units or propose live sessions.
My take — AI-written commentary, not fact-checked reporting
Renaming NLP to LLM is honestly just accurate labeling, not hype chasing, and I respect that Hugging Face kept the classic NLP material instead of quietly burying it. The real signal here is the multi-vendor collaboration with LangChain, Unsloth, and others, which is what open education should look like instead of a single company's walled tutorial garden.
Read more about this at: Hugging Face