- A foundational set of instructions that defines an LLM’s persona, operational constraints, and behavioral tone. It functions as a persistent layer, ensuring the model adheres to specific formatting, safety, and stylistic guidelines throughout a conversation.
- Related terms: Prompt engineering · Large language model · Prompt injection attack · Zero-shot prompting
- Previous term: System Object Model
- Next term: Systems Biology Markup Language
- Random term: Routing information base (webglossary.info/random 🎲)