# AI Instruction Tuning The process of fine-tuning a base language model on datasets of instruction-response pairs to make it follow human instructions. This is the key step that transforms a next-token predictor into a useful assistant. Without instruction tuning, models generate plausible text continuations rather than helpful answers. Pioneered by FLAN (Google, 2022) and InstructGPT (OpenAI, 2022). Often combined with [[Reinforcement Learning From Human Feedback (RLHF)]] for further alignment. ## References ## Related - [[Large Language Models (LLMs)]] - [[Reinforcement Learning From Human Feedback (RLHF)]]