Skip to main content
Back
40 XP

Lesson 1 · 10 min

Should you fine-tune?

Fine-tuning is rarely the answer. This lesson is a decision tree for when it actually is.

The hierarchy of solutions

When the model doesn't do what you want, try in this order:

  1. A better prompt — structure, examples, constraints. Solves 95% of cases.
  2. RAG — when the gap is knowledge the model lacks. Cheaper, fresher, traceable.
  3. Fine-tuning — when the gap is style, format, or capability the base model can't be prompted into.

Most teams skip steps 1 and 2 because fine-tuning sounds more impressive. Then they spend two weeks fine-tuning what a 200-word system prompt would have solved.