---
title: "Late chunking"
sidebarTitle: "Late Chunking"
description: "Retrieve whole documents first, chunk them afterwards."
---

Inverts the usual order: documents are retrieved first and chunked just before
synthesis. Chunking late preserves context that pre-chunking discards, since a
chunk is cut with knowledge of the query it is answering.

<Note>
This example is a runnable notebook in the
[vectordb-recipes](https://github.com/lancedb/vectordb-recipes) repository rather than a page here.
</Note>

<CardGroup cols={2}>
  <Card title="Open in Colab" icon="google" href="https://colab.research.google.com/github/lancedb/vectordb-recipes/blob/main/examples/Advanced_RAG_Late_Chunking/Late_Chunking_(Chunked_Pooling).ipynb">
    Run it end to end in the browser.
  </Card>
  <Card title="View on GitHub" icon="github" href="https://github.com/lancedb/vectordb-recipes/tree/main/examples/Advanced_RAG_Late_Chunking">
    Read the source.
  </Card>
</CardGroup>