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?
3 / 31
BackNext

A model is a file

We've spent this whole module building a network: neurons in layers, connected by weights, slowly trained until the weights settle into useful values. So what do you actually have once the training stops?

You have a model.

The network is the machine: the layout of neurons and the rules for passing signals through it. The model is what that machine becomes once it's trained, the exact weights it landed on. Same structure, but now filled in with everything it learned. When people release "a model," this is what they mean: a network frozen at the moment it got good.

And when people talk about it, the word can sound like something grand. A system. A mind. Something that lives in the cloud.

But a model is a file. A plain file, sitting on a hard drive, the same way report.pdf or vacation-photos.zip does. When researchers share their work publicly, they publish that file (a huge one, often tens or hundreds of gigabytes). And if you opened it, here is what you would find inside:

Numbers. Billions of them. They are organized into groups and layers, but none of it looks like knowledge. No words. No rules. No instructions written by hand. Just numbers, each one a weight.

Those numbers are the model. Everything it knows, every pattern it learned, every capability it has, lives in that list. Nothing else.

But a list of numbers on its own doesn't do anything. There's a second piece, and without it, the numbers are meaningless.

Citations
Citations