learngenai.space
Module 1

How LLMs Work

Build a correct mental model of tokens, embeddings, attention, and next-token generation.

~90 minutes

Learning path

  1. 1Welcome: You Are Going to Build AI Systems~10 min · Guest free
  2. 2What Actually Is AI?~12 min · Guest free
  3. 3What Is a Model?~10 min · Guest free
  4. 4How Does a Model Learn?~14 min
  5. 5Training, Validation, Testing, Inference~10 min
  6. 6Neural Networks Without the Scary Math~12 min
  7. 7Tokens: LLMs Don't Read Text Like You Do~14 min
  8. 8Context Windows~12 min
  9. 9Embeddings: Meaning as Numbers~14 min
  10. 10Why Transformers?~10 min
  11. 11Attention~12 min
  12. 12Query, Key, Value~12 min
  13. 13Self-Attention~14 min
  14. 14Inside a Transformer Block~12 min
  15. 15Encoder, Decoder, and Decoder-Only Models~10 min
  16. 16What Is an LLM?~8 min
  17. 17The Flagship Next-Token Simulation~14 min
  18. 18Logits, Probabilities, Softmax~10 min
  19. 19Sampling~14 min
  20. 20Why Can an LLM Follow Instructions?~10 min
Milestone

Build a Mental Model of an LLM

You have the pieces. Now assemble the full next-token story as an architecture: from raw text to the next generated token. When it works, we change one constraint and ask you to diagnose the failure like an engineer — not like a spectator.

Open challenge