Instruction tuning

Fine-tuning a pretrained base model on examples of instructions paired with the response they should produce, so the model learns to follow a request rather than merely continue text. This is the step that turns a raw language model into something that reliably obeys a system prompt.

Why exams ask this

Tested as the reason a base model and a chat-ready model behave differently on the identical prompt. A base model completes text, an instruction-tuned model follows a request, and the exam wants that distinction named directly rather than described as "better."

Relevant to

Related concepts

Resources

No resources linked to this concept yet.