Prompt structure
Prompt structure is the order and separation of a prompt’s parts — role, context, task, constraints, format, examples — which changes model behaviour independently of the words used.
Two ordering rules carry most of the benefit: put long reference material before the instructions, so the model reads the material before learning what to do with it; and restate the task at the end, because models weight the end of the context heavily.
Related terms
- XML tags in prompts
- XML tags in prompts are paired markers such as `<context>` and `</context>` used to label each section of a prompt so the model can tell them apart.
- Recency bias
- Recency bias is a model’s tendency to give more weight to instructions near the end of a prompt than to those buried earlier.
- Lost in the middle
- Lost in the middle is the observed effect that models recall information placed at the start or end of a long context more reliably than information in the middle.
More on prompt structure
See the full glossary, read the guides, or put it into practice in the prompt builder.