Fine-tuning is the process of further training a pre-trained model on selected data to adapt it to a domain, task, style, or behavior.
Fine-tuning differs from prompting and retrieval because it changes model weights rather than only changing the input context or retrieved evidence.