Fine-tuning
The second training stage, where a pre-trained model is further trained on a smaller, task-specific dataset. Because the model already carries general knowledge, fine-tuning converges faster and needs far less labelled data to reach strong performance.
An imaging team fine-tunes a general chest X-ray model on 800 labeled films from its own scanners, and accuracy on a held-out slice of those same films climbs. Nobody rechecks the findings the local set barely contained, such as small pneumothoraces, which the updated model may now detect less reliably than the original did. Ask for results on the original broad test set as well as the local one, broken out by finding, and treat any drop as a reason to hold the update.
Terms like this come up in real clinical scenarios across the HelloAI courses: bite-sized modules with verifiable certificates. An account takes one minute, no password needed.
Sign in →