smol-course

The smol-course offers an augmented documentation experience to help you get started. Participants can discover the course syllabus, learn about the certification process, and meet the team behind the course. The course is particularly suited for software developers and engineers looking to fast-track their skills in fine-tuning large language models. If you're not in this category, you may want to explore the LLM Course instead.

Throughout the course, participants will study various topics, including instruction tuning, supervised fine-tuning, preference alignment, evaluation, and vision language models. They will learn to use established fine-tuning frameworks and tools like TRL and Transformers. Additionally, the course encourages sharing projects and exploring fine-tuning applications created by the community. Participants will also have the opportunity to participate in challenges, where they can evaluate their fine-tuned models against those of other students.

The course is structured into foundational units that cover theoretical concepts, hands-on sections for practical application, use case assignments for real-world problem-solving, and collaborations with Hugging Face’s partners to provide the latest fine-tuning implementations and tools. It is a living project that evolves with participant feedback and contributions, encouraging engagement through GitHub and Discord.

To enroll, participants should have a basic understanding of AI and LLM concepts, familiarity with Python programming, and experience with deep learning frameworks like PyTorch. The course requires only a computer with internet access, preferably with GPU capabilities, and an account to access resources and create projects. The certification process is free, offering certificates for both fundamentals and completion based on course participation and project submission. Each chapter is designed to be completed in approximately one week, with a recommended pace to help participants stay on track.