---
id: wiki-2026-0508-instruction-tuning
title: Instruction Tuning
category: 10_Wiki/Topics
status: duplicate
canonical_id: wiki-2026-0508-fine-tuning
duplicate_of: "[[Fine-tuning]]"
aliases: [instruction tuning, IFT, FLAN, Alpaca, ShareGPT]
source_trust_level: A
confidence_score: 0.96
verification_status: redirected
tags: [duplicate, instruction-tuning, sft]
last_reinforced: 2026-05-10
github_commit: pending
---

# Instruction Tuning

> **이 문서는 [[Fine-tuning]] 의 specialization 입니다.** Canonical 문서로 redirect.

## 핵심 요약 (instruction-specific)
- 매 (instruction, response) pair 의 SFT.
- 매 FLAN (Wei 2021), Alpaca, ShareGPT, Dolly.
- 매 RLHF / DPO 의 의 의 prerequisite.
- 매 LIMA (1000 high-quality > 100k noisy) 매 data quality emphasis.
- 매 modern: 매 multi-turn + tool use + reasoning data.

## 🔗 Graph
- 부모: [[Fine-tuning]] (canonical)
- Adjacent: [[Foundation-Models]] · [[GRPO]]

## 🕓 변경 이력
| 날짜 | 변경 |
|---|---|
| 2026-05-08 | Phase 1 |
| 2026-05-10 | 중복 처리 — canonical 문서로 redirect |