Concept

Fine-tuning

Additional model training on task-specific examples to change behavior or specialize performance.

In practice

A classifier can be fine-tuned on consistently labeled examples.

Why it matters

Use the term as an operating concept, not jargon. Ask what it changes in cost, quality, evidence, latency, security or user experience.

Back to glossary →