Vector embedding demos

A vector embedding encodes an input as a list of floating point numbers.

"dog" → [0.017198, -0.007493, -0.057982, 0.054051, -0.028336, 0.019245,…]

Different models output different embeddings, with varying lengths.

Model	Encodes	Vector length
word2vec	words	300
Sbert (Sentence-Transformers)	text (up to ~400 words)	768
OpenAI text-embedding-ada-002	text (up to 8191 tokens)	1536
OpenAI text-embedding-3-small	text (up to 8191 tokens)	256-1536
OpenAI text-embedding-3-large	text (up to 8191 tokens)	256-3072
Azure AI Vision	image or text	1024

Vector embeddings are commonly used for similarity search, fraud detection, recommendation systems, and RAG (Retrieval-Augmented Generation).

This repository contains a visual exploration of vectors, using several embedding models.

Getting started

Before running the notebooks, install the requirements:

pip install -r requirements.txt

Then explore these notebooks:

• Generate new OpenAI text embeddings
• Compare OpenAI and Word2Vec embeddings
• Vector similarity
• Vector search
• Generate multimodal vectors for dataset
• Explore multimodal vectors
• Vector distance metrics
• Vector quantization
• Vector dimension reduction (MRL)

These notebooks are also provided, but aren't necessary unless you're generating new embeddings data.

• Generate OpenAI embeddings for datasets
• Generate Word2Vec embeddings for datasets

Deploying Azure OpenAI embedding models

If you need to generate new OpenAI embeddings, you'll need access to the embedding models via the API. This project includes infrastructure as code (IaC) to provision an Azure OpenAI deployment of "text-embedding-3-large". The IaC is defined in the infra directory and uses the Azure Developer CLI to provision the resources.

1. Make sure the Azure Developer CLI (azd) is installed.

2. Login to Azure:

   azd auth login

   For GitHub Codespaces users, if the previous command fails, try:

    azd auth login --use-device-code

3. Provision the OpenAI account:

   azd provision

   It will prompt you to provide an azd environment name (like "vector-demos"), select a subscription from your Azure account, and select a location. Then it will provision the resources in your account.

4. Once the resources are provisioned, you should now see a local .env file with all the environment variables needed to run the scripts.

5. To delete the resources, run:

   azd down

Resources

Each notebook has resources at the bottom to dig further into that topic. Here are some additional general resources:

• Poster: Visual exploration of vectors (PyCon 2025)
• Livestream Recording: Vector Embeddings from Python + AI Series
• Vector compression techniques