Fine-tuning
Fine-tuning means continuing the training of a foundation model on a specific dataset, to specialize it for a particular task or tone.
Rather than starting from scratch, fine-tuning adjusts an already trained model's parameters using a much smaller set of examples. The model keeps its general knowledge and gains a targeted skill.
For most companies, RAG is enough and costs less than fine-tuning. Fine-tuning becomes useful for locking in a precise response style or a repetitive task at scale.