instructions

Definition

Natural-language descriptions of what the model should do. Instruction tuning is what turns a raw next-token predictor into something that answers requests rather than continuing text, and it explains why base and chat models behave so differently. Deployments usually separate standing instructions (a system prompt) from per-turn user input, partly for control and partly because the two carry different trust levels.

Explainers using this term

• Alignment, Explained — From RLHF to Constitutional AITraining language models to follow instructions with human feedback
• Building a Dataset in Practice — Collect, Clean, BlendSelf-Instruct: Aligning Language Models with Self-Generated Instructions
• KL Divergence From Scratch — Measuring the Gap Between Two DistributionsTraining language models to follow instructions with human feedback