---
title: 'ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT'
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
license_note: 'paper is arXiv non-exclusive distribution license (verified on arXiv abs page 2026-11-12); this pointer note is our own words and is CC BY 4.0; we link to the paper rather than reproduce it'
source: https://iinventedlight.com/library/notes/2004-12832-colbert.md
source_license: 'arXiv non-exclusive distribution license'
attribution: 'Pointer by iinventedlight (AI-assisted). Source: Omar Khattab, Matei Zaharia, ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT (arXiv non-exclusive distribution license). This note links to the paper rather than reproducing it.'
author: 'iinventedlight (AI-assisted)'
date: 2026-11-12
status: published
approved_by: "Cole Withington, 2026-11-12 14:44 PT, Librarian chat widget t51s316: \"Yes, publish all 100 second-list arXiv notes\""
checked_on: 2026-11-12
as_of: 2026-11-12
confidence: medium
untested: 'Pointer note only. Paper not reproduced. Abs page and license checked 2026-11-12.'
category: arxiv
topic_area: 'ai-research: RAG'
tags: arxiv, RAG, retrieval, late-interaction
arxiv_id: 2004.12832
alias: 2004.12832-colbert
type: pointer
format: pointer
sources:
  - id: S1
    title: 'ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT'
    publisher: arXiv
    authors: 'Omar Khattab, Matei Zaharia'
    url: https://arxiv.org/abs/2004.12832
    version_read: 'v2 (4 Jun 2020)'
    accessed: 2026-11-12
    license: 'arXiv non-exclusive distribution license'
---
# ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT

## Summary
[ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT](https://arxiv.org/abs/2004.12832) (2020) introduced ColBERT, a late-interaction retrieval architecture. Instead of compressing a document into one vector, it keeps token-level BERT embeddings and delays the interaction between query and document until scoring time, aiming for both effectiveness and much faster retrieval than full cross-encoders.

What it introduced: contextualized late interaction over BERT for passage search, so that fine-grained token matches can be precomputed for documents and then compared with a query without running a full transformer on every pair.

What it led to: later ColBERT-style retrievers and a standard alternative to single-vector dense retrieval in RAG stacks. Live note: [2005-11401-retrieval-augmented-generation](https://iinventedlight.com/library/notes/2005-11401-retrieval-augmented-generation/).

Use it when you are choosing between a bi-encoder and a late-interaction retriever. Where to read it: https://arxiv.org/abs/2004.12832 (v2 (4 Jun 2020)).

The paper is posted on arXiv under the arXiv non-exclusive distribution license. This note links to it rather than reproducing it. The note itself is CC BY 4.0.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Pointer by iinventedlight (AI-assisted). Paper license: arXiv non-exclusive distribution; this note links to it and does not reproduce it.

---

Agents: [/agents.md](/agents.md) | [/.well-known/agent-library.json](/.well-known/agent-library.json).
