Loading slide

What a Trained Model Actually Is

  1. 01What a Trained Model Actually Is
  2. 02Memory card
  3. 03A model is a file
  4. 04Two files
  5. 05How the two fit together
  6. 06A Piano
  7. 07The valuable file
  8. 08You connect to it
  9. 09Where does training data come from?
  10. 10Cleaning the data
  11. 11The network can't read
  12. 12Guess the next word
  13. 13Right or wrong
  14. 14One nudge, a billion times
  15. 15Take a breath
  16. 16Where your memories live
  17. 17A base model is not an assistant
  18. 18Why it's called pre-training
  19. 19The weights are the model
  20. 20The scale of it
  21. 21Why so many?
  22. 22What a weight actually is
  23. 23Everywhere and nowhere
  24. 24Using the model: inference
  25. 25The whole arrangement
  26. 26Reinforce your understanding
  27. 27Question: What is a model, really?
  28. 28Question: Learning from raw text
  29. 29Question: The model doesn't know what it doesn't know
  30. 30Quiz: answer
  31. 31The machine learned something real.
  32. 32Want to go deeper?
6 / 31
BackNext

A Piano

That was the machinery. Now a way to feel it.

Picture a piano, and picture Beethoven sitting down to play.

The piano is the program. Built once, by skilled people, it doesn't change. Every key is in the same place. The instrument is identical whoever sits down at it.

Beethoven is the weights. Everything he knew lived inside him, built up through years of practice. Thousands of hours of repetition, until the patterns became automatic. When he played, he was not actively thinking about which finger goes where. It just flowed. Ask him to explain exactly what happened in his hands and his mind, and he couldn't fully tell you. The knowledge was real, but it didn't live anywhere you could point to.

A trained model is the same pairing: a fixed instrument anyone can own, and inside it, everything learned, carried in a form impossible to fully explain.

Go ahead, play a few notes.

Citations
Citations