Skip to content
DKKB

Large language model

Section titled “Large language model”

A large language model, or LLM, is a language model trained at substantial scale so it can perform a broad range of language tasks from instructions and context.

The term matters because an LLM is usually used as a general model inside a larger system, not as a complete application by itself.

Prompting, retrieval, tools, structured output, evaluation, and authorization are separate system concerns. The related DKKB entries cover those engineering decisions in more depth.

Used in DKKB

Context engineering for LLM systemsDeliberately select and structure the information a model receives instead of treating the context window as an unbounded memory store.
Context engineering for LLM systemsDeliberately select and structure the information a model receives instead of treating the context window as an unbounded memory store.
Retrieval-augmented generationRetrieve external knowledge for a task and provide the selected evidence to the model at generation time.
Evaluate LLM behavior and hallucination riskMeasure model behavior against representative tasks and verify externally checkable claims instead of treating fluent output as evidence.