AI Glossary
What is Fine-Tuning?
Further training an existing AI model on a smaller, specific set of examples so it performs better at a particular task, style or domain.
A general model knows a little about everything. Fine-tuning teaches it to be better at something specific, for example writing support replies in a company's tone, or classifying a particular kind of document, by training it further on many examples of that task.
When it is used: mostly by companies and developers who need consistent output at scale and have hundreds or thousands of good examples.
What most people need instead: for everyday work, good prompts usually achieve the same goal much more cheaply:
- Paste examples of the style you want (few-shot prompting)
- Save standing instructions in your AI tool's settings
- Give the AI the relevant documents to work from (RAG)
Try those first. Fine-tuning is a bigger project with its own costs and data-privacy considerations.