Learning path · no account required

How LLMs Work

From tokens and transformers to grounding strategies used in real LLM applications.

Students, developers, and technical PMs Reviewed:

Goal

Understand the moving parts of large language models well enough to design prompts, RAG, and evaluations intentionally.

Lessons

  1. 1 LLM overview Capabilities and limits at a glance.
  2. 2 Tokens and cost How text becomes model input.
  3. 3 Transformer architecture Attention-powered deep learning backbone.
  4. 4 Attention How models focus across context.
  5. 5 Embeddings Vectors for meaning and search.
  6. 6 Context windows What fits in a single call.
  7. 7 Pretraining Where general capabilities come from.
  8. 8 Fine-tuning Adapting behavior with data.
  9. 9 RAG grounding Connecting models to your documents.
  10. 10 Failure modes Why fluent answers can still be false.

← All learning paths · Learning hub