# ColBERT

Stores an embedding per token and scores a document by matching query tokens against document tokens.

## What it is used for

High-accuracy retrieval with fine-grained matching.