Guide
Glossary
Fine-tuning
Fine-tuning is further training of an existing AI model on a set of your own examples, so it consistently follows a particular style, format or task. The examples change the model itself rather than being supplied with each request. Fine-tuning is good at teaching how to respond, such as a house style for file notes, but a poor way to teach facts that change, like current precedents or policies. For answers drawn from firm documents, most firms use retrieval-augmented generation instead, which is easier to update and can show its sources. Adding instructions or files to a chat assistant's project is not fine-tuning.
Also called model fine-tuning, custom model training
Last updated:
Example
In a law firm
A commercial law firm wants its AI assistant to answer questions from its precedent bank. Fine-tuning a model on the precedents would fix today's versions into the model and hide which document each answer came from. The firm instead indexes the precedents for retrieval, so updates apply straight away and every answer links to the source clause.
Guides that explain it in context
Secure by design. Set up correctly. Fully managed.
Talk to us before you commit to anything
Start with a free 45-minute discovery call. We look at your systems and priorities, then recommend a first step with a fixed scope, or tell you if we are not the right fit.