What is Fine-tuning?
Training a pre-trained LLM with domain-specific data to specialize it for particular tasks. Requires much less data and compute than training from scratch. Good for stable output formats or specialized knowledge, but more expensive than RAG.