Definition
Fine-tuning is further training of a pre-trained model on an organisation's own examples so that it learns specific tasks, formats or vocabulary.
Fine-tuning vs retrieval
Fine-tuning teaches behaviour; retrieval supplies facts. Facts that change belong in retrieval, while repeated tasks and specialist language benefit from fine-tuning.
Ownership
When fine-tuning runs on your infrastructure, the resulting weights can belong to you.
See RAG vs fine-tuning.