What is RAG (Retrieval-Augmented Generation)?

A retrieval method that generates multiple query variants, retrieves results for each, and fuses the rankings.

Additional inference computation used during response generation to improve quality or reasoning.

Finding relevant documents or records from a knowledge source for a query.

Compressing knowledge from a large teacher model into a smaller student model.

The amount of processing power consumed while producing each response.

The latest point in time reflected in a model's training data.