Instruction-Tuned Model
A model fine-tuned on instruction-response pairs to follow natural-language directives reliably.
Full Definition
Instruction tuning adapts a base language model by training it on a curated dataset of (instruction, response) pairs that cover diverse tasks — summarisation, translation, question answering, coding, and more. After instruction tuning, the model follows natural-language commands rather than just continuing text. This was the key insight behind InstructGPT (2022) and is now standard practice. Instruction-tuned models are dramatically more useful for end-users because they generalise instruction-following to tasks not in the training set. Most commercially deployed models — ChatGPT, Claude, Gemini — are instruction-tuned versions of their underlying base models.
Examples
InstructGPT responding helpfully to 'List five benefits of meditation' instead of just continuing the sentence fragment.
Llama 3 Instruct following the instruction 'Rewrite this paragraph more formally' without needing few-shot examples.
Apply this in your prompts
PromptITIN automatically uses techniques like Instruction-Tuned Model to build better prompts for you.
Related Terms
Instruction Tuning
Supervised fine-tuning on diverse instruction-response pairs to improve a model'…
View →Base Model
A model trained only on next-token prediction over a large corpus, before any in…
View →Fine-Tuned Model
A pretrained model whose weights have been updated on a specific dataset for a t…
View →