Fine-tuning: definition

Fine-tuning is further training a base model on your own examples so it learns a specific style, format, or task more reliably than prompting alone can achieve.

Fine-tuning changes the model’s weights, unlike RAG, which only changes what you show it at question time. It shines when you need a consistent voice or a narrow classification done thousands of times, and you have good labelled examples.

It is also the more expensive, slower-to-update option, so we reach for prompting and RAG first and only fine-tune when the task clearly earns it.

Related terms

Large language model (LLM): A large language model is a neural network trained on huge amounts of text to predict the next token, which lets it write, summarise, classify, and answer questions across almost any topic.
Prompt engineering: Prompt engineering is the craft of writing the instructions, examples, and context you give a model so it produces reliable, correctly formatted output for a specific task.
Retrieval-augmented generation (RAG): Retrieval-augmented generation is a technique that fetches relevant passages from your own documents at question time and hands them to the model, so its answer is grounded in your data instead of its training.
Inference: Inference is the act of running a trained model to get an output, and inference cost is what you pay each time the model answers, measured in tokens.