Skip to index

GLOSSARY

Fine-tuning

Additional training that adapts a pre-trained model to a specific style, task or domain, using your own examples.

Fine-tuning takes a pre-trained model and continues training it on curated examples — your company's support tickets, an artist's portfolio, code in your house style — so its outputs specialize. It is how generic models become brand voices, domain experts or task specialists.

Modern practice offers cheaper alternatives: prompting with examples ('few-shot'), and retrieval systems that inject facts at answer time. Fine-tuning still wins when you need consistent style at volume or a behavior prompts cannot reliably induce — and platforms from OpenAI to Leonardo.AI and Replicate make it self-serve.

Related terms

Tools that use this

Related categories