# Instruction Tuning

> More training on asks paired with good answers. Then the model does what you ask.

**Field:** Artificial Intelligence  
**Category:** LLMs  
**Also known as:** Supervised Fine-Tuning, SFT

## Think of it like

Think of it like you show a helper lots of notes. Each one has the right reply under it.

## Definition

Instruction tuning is fine-tuning on many pairs of a request and a good reply. It is the step that turns a base model into one that follows requests. The pairs cover many kinds of task, like 'sum this up', 'write a poem' or 'fix this code'. They are written by people or by other models, then checked. It is also called supervised fine-tuning.

## Example

A team writes 50,000 pairs such as 'List three uses for salt' with a good list under it. After training on them, their base model starts to answer like a helper.

## Why it matters

It is the step that makes a model useful to talk to. Without it, you get a text guesser, not a helper.

## Common mistake

Thinking more pairs always help. A small set of very good pairs often beats a huge set of sloppy ones.

## Three principles

- **Request and reply pairs.** Many kinds of task.
- **Turns base into helper.** It learns to follow.
- **Quality over count.** Good pairs matter most.

## Related terms

- [Fine-Tuning](https://learn.benibix.com/term/ai/fine-tuning)
- [Instruct Model](https://learn.benibix.com/term/ai/instruct-model)
- [Base Model](https://learn.benibix.com/term/ai/base-model)
- [Reinforcement Learning from Human Feedback (RLHF)](https://learn.benibix.com/term/ai/reinforcement-learning-from-human-feedback)

## Taught in

- [Post-Training and RLHF](https://learn.benibix.com/learn/ai-llms/post-training-and-rlhf): LLMs, 8/15

![Three stages from left to right. Pretraining on trillions of tokens makes a base model. Instruction tuning on many examples makes it follow requests. Feedback from people makes it a helpful chat model.](https://learn.benibix.com/diagrams/llm-training-stages.svg)

---

Source: https://learn.benibix.com/term/ai/instruction-tuning
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-18
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
