Definition
Fine-Tuning
Fine-tuning continues training a pre-trained model on your own examples so its weights adapt to a specific task, style, or vocabulary. The book's position is that most teams no longer need it: 200K-token context windows let you put documentation and examples straight into the prompt. It still earns its place for high-volume cost savings, regulated domains with specialized terminology, or output formats prompting cannot make consistent.

Explained in
Chapter 7: The New Landscape
The current state of AI. What's changed, what it means for builders.
Related terms