Instruction tuning is a training stage where a base language model is fine-tuned on thousands of examples of instructions paired with good responses, teaching it to follow human requests instead of merely continuing text. A raw base model asked «write a welcome email» might just ramble; an instruction-tuned model actually writes the email. It is comparable to onboarding a brilliant but unstructured new employee: the raw talent exists, and instruction tuning teaches them how to respond helpfully to actual work requests.