Definition
Fine-tuning is the process of further training a pre-trained LLM on a specific, smaller dataset to adapt it for particular tasks or domains. Unlike prompt engineering (which changes the input), fine-tuning changes the model's weights to improve performance on targeted use cases.
Example
A company fine-tunes a base LLM on their internal documentation and support tickets to create a customer service agent that understands their specific products, terminology, and response patterns.
Why it matters
Fine-tuning produces the most consistent, domain-specific behavior. When off-the-shelf models don't quite match your needs, fine-tuning bridges the gap.