Process of continuing to train a pretrained foundation model on your own labeled dataset to specialize it for a specific task or domain, adjusting the model's weights (unlike RAG or in-context learning, which don't modify the model).
In AIF-C01 and GENAI-L it's compared with RAG and in-context learning on a cost and complexity scale: in-context learning (cheapest, no training) < RAG (moderate) < fine-tuning (requires labeled data and compute) < pre-training from scratch (most expensive). Question: adapt a model's tone to the company's brand voice using your own examples → fine-tuning.
Fine-tuning is a concept that can show up on the exam. Check the full guide to see it in context.