Skip to main content
Deliberate AcademyProfessional AI Education
AI Glossary

Instruction tuning

What is Instruction tuning?

A fine-tuning technique that trains a base language model on instruction-response pairs, making it better at following user directions. Used to create chat-optimised models from raw pre-trained models. ChatGPT and Claude are instruction-tuned models.

Example in practice

The difference between a raw GPT base model (which completes text) and ChatGPT (which follows user instructions) comes from instruction tuning — training the model on thousands of human-written request-response examples.

See also

Fine-tuningRLHF (Reinforcement Learning from Human Feedback)Base Model
← Back to Glossary