Knowledge MaxxingRetrieval-augmented generation
Retrieval-augmented generation retrieves relevant documents at request time and includes them in a model's prompt, so its answer is grounded in that supplied content.
Prerequisite: Information retrieval
What matters
This lets a model answer accurately about information outside its training data or too recent to have been learned, by supplying the needed facts directly in context instead.
Example
A support chatbot using retrieval-augmented generation looks up the current version of a policy document before answering, rather than relying on what it memorized during training.
Common misconception
Retrieval-augmented generation does not eliminate hallucination entirely; the model can still misread or ignore the retrieved context when generating its answer.
Recall question
What problem does supplying retrieved documents in the prompt help solve?