// the find
mlabonne/llm-course
Course to get into Large Language Models (LLMs) with roadmaps and Colab notebooks.
A free, three-part roadmap (fundamentals, 'LLM Scientist', 'LLM Engineer') that links out to articles, videos, and Colab notebooks covering everything from linear algebra to RLHF, quantization, and deployment. It's for people who want a structured self-study path through LLM theory and tooling rather than a library or framework to install.
The three-tier structure (math/Python basics → training/fine-tuning/alignment → production deployment) is genuinely well-sequenced, not just a link dump in random order. Several of the Colab notebooks (AutoQuant, LazyMergekit, fine-tuning scripts) are the author's own working tools, not just reference material, so you can actually run something instead of only reading. Coverage is broad and current enough to include GRPO, DPO, and modern quantization (AWQ/GPTQ) alongside the fundamentals.
This is a README, not code — there's nothing to clone and run, no tests, no package, so 'last push' dates and stars are really measuring content edits, not software health. Most value lives off-repo: Colab notebooks on personal Google Drive links and blog posts on an external site, both of which rot independently of this repo and aren't version-controlled alongside it. No changelog or versioning on the course content itself, so you can't tell at a glance which sections were updated recently versus left stale since 2022-2023. The 'toggle section' README format means a lot of the real content is collapsed and easy to skim past on GitHub's renderer.