# RLHF: Reinforcement Learning from Human Feedback — From Ouyang 2022 InstructGPT to ChatGPT

> Source: https://sukruyusufkaya.com/en/learn/llm-muhendisligi/rlhf-ouyang-2022-instructgpt-chatgpt
> Updated: 2026-05-13T11:17:37.253Z
> Category: LLM Mühendisliği
> Module: Module 15: RLHF + DPO — Alignment & Preference Optimization
**TLDR:** Full anatomy of RLHF: SFT model → reward model training (Bradley-Terry) → PPO RL training. Ouyang 2022 InstructGPT paper, 3-stage pipeline, KL divergence penalty, reward hacking concerns. ChatGPT's secret sauce. Turkish RLHF challenges (human annotator pool, cultural nuances).