Fine-tuning

The process of further training an existing AI model on a smaller, more specific dataset so that it performs better on a particular task or domain, such as legal writing or customer support. Fine-tuning is cheaper and faster than training a model from scratch and is a common way companies customise foundation models for their own products. (See also: Foundation model, Training data)