Glossary
Fine-tuning
Additional training of an existing model on a smaller, focused dataset to adapt it to a task.
Fine-tuning takes a model that has already been trained and trains it further on a narrower set of examples, such as medical notes, legal documents, or examples of a desired writing style. It is much cheaper than training a model from scratch.
Companies use fine-tuning to make a general model better at a specific job or to shape its behavior. Fine-tuning can also weaken safety behavior that a developer built in, which is one reason open-weight releases raise policy questions.
A fine-tuned model is still built on its base model and inherits many of its strengths and weaknesses.