Glossary · Phase 6 · Modern LLM engineering
Demonstration
In fine-tuning, an example conversation showing exactly the reply you want the model to learn to give.
Try it now
The only new idea is the mask: don't train it to write the question.
Learn it properly
Supervised fine-tuning · Teach behavior from example conversations. About 40 minutes.
Other terms from the same lesson
- catastrophic forgetting: When fine-tuning on a narrow task makes a model lose skills it had before, because the same weights get overwritten.
- SFT: Supervised fine-tuning: training on example prompts with good answers, so a model learns to respond helpfully.
Never trained an AI?
Teach a small one to write, right in your browser. It takes about 20 seconds and needs no account.