---
title: "Discussion on \"When Does Similarity Stop Working?\" | Hashnode"
description: "Discussion on \"When Does Similarity Stop Working?\". The Concentration Problem at the Heart of Vector Search\n\n\nEvery major embedding model produces hundreds or thousands of dimensions. You use cosine similarity on them. There is a theorem that says this"
url: "https://hashnode.com/posts/when-does-similarity-stop-working/69db9461aadf1107e2db0260"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [When Does Similarity Stop Working?](https://okonu.hashnode.dev/when-does-similarity-stop-working?utm_source=hashnode&utm_medium=feed)

The Concentration Problem at the Heart of Vector Search Every major embedding model produces hundreds or thousands of dimensions. You use cosine similarity on them. There is a theorem that says this

Apr 5 · 20 min read · okonu.hashnode.dev  
Read the article

## Written by

[Ian Okonu](https://hashnode.com/@Okonu)

@Okonu · Nairobi, Kenya

  
[View profile](https://hashnode.com/@Okonu)

ML &AI Architect

LLM Orchestration · RAG Pipelines · Multi-Agent Systems | Production AI for Fintech, Legal &amp; Gov on Azure · GCP · AWS

## Responses

Markdown supported · ⌘↵ to post

No responses yet.
