# Foojay Podcast #56: Vectors in Java Code, Databases, and LLMs

> In this Foojay podcast, we enter the world of mathematics by discussing Vectors and how they are crucial for AI and machine learning. 

- URL: https://foojay.io/today/foojay-podcast-56/
- Published: 2024-07-08
- Authors: Frank Delporte
- Categories: Apache Cassandra, Databases, DataEngineering, DataStax, Machine Learning, Podcast, Project Panama


In this Foojay podcast, we enter the world of mathematics by discussing Vectors and how they are crucial for AI and machine learning.

As ChatGPT explains: **"A Vector is a mathematical structure that holds numerical values. Vectors are fundamental to the field of Artificial Intelligence, as they allow mathematical operations to be performed efficiently and form the basis of many machine learning algorithms."**

OK, but... how are these vectors crucial for the whole Artificial Intelligence evolution?

This is the last podcast of season 3, we're taking a summer break, and will be back in September with the release of Java 23 and many more OpenJDK-related topics!

## Video

{{< youtube QZAYscsyVK0 >}}

## Podcast Apps

You can listen and subscribe to the Foojay Podcast on:

* [Spotify](https://open.spotify.com/show/6CpTfgn9LirzJGAtc4ICdQ)
* [Apple Podcasts](https://podcasts.apple.com/be/podcast/foojay-io-the-friends-of-openjdk/id1652281304)
* And most others...

## Guests

* Jonathan Ellis
  * <https://www.linkedin.com/in/jbellis/>
  * <https://x.com/spyced>
* Alexander Chatzizacharias
  * <https://www.linkedin.com/in/alexander-chatzizacharias/>
  * <https://x.com/alex90_ch>

## Content

00:00 Introduction of the topic and guests  

01:57 What is a Vector?

<https://github.com/openai/tiktoken>  
<https://arxiv.org/abs/1301.3781>  
<https://towardsdatascience.com/word2vec-research-paper-explained-205cb7eecc30>  
<https://github.com/jbellis/jvector>

07:14 Vectors explained as a game

A fun and absurd introduction to Vector Databases: <https://www.youtube.com/watch?v=mQGf9hWTqSw>

09:44 Understanding tokenizers  

10:40 Do we need dedicated Vector databases?  

13:39 Vectors, LLMs, and hallucinations

Crafting your own RAG system: Leveraging 30+ LLMs for enhanced performance by Stephan Janssen: <https://www.youtube.com/watch?v=9PX5l4ETn0g>

20:40 How LLM and chat interfaces are used in companies

<https://www.bbc.com/travel/article/20240222-air-canada-chatbot-misinformation-what-travellers-should-know>

23:45 Indexing all of Wikipedia

<https://foojay.io/today/indexing-all-of-wikipedia-on-a-laptop/>  

Demo application: <https://jvectordemo.com:8443/>  
<https://openjdk.org/projects/panama/>

27:23 Evolutions in Java for vectors, LLMs, and AI

Vector API (Eighth Incubator): <https://openjdk.org/jeps/469>  

Foreign Function \& Memory API: <https://openjdk.org/jeps/454>

32:44 Is the GPU needed for vector use cases?  

35:04 Can we already use the incubator Vector API in production?  

38:27 Some predictions...

Colbert project: <https://github.com/stanford-futuredata/ColBERT>  
<https://thenewstack.io/overcoming-the-limits-of-rag-with-colbert/>

44:19 Make your vectors smaller to make them more efficient and less expensive

<https://www.sciencedirect.com/topics/engineering/vector-quantization>  
<https://huggingface.co/blog/embedding-quantization>  
<https://foojay.io/today/visualizing-brain-computer-interface-data-using-javafx/>  

Asteroids 3D in JavaFX made from AI Deep Fake Audio data: <https://www.youtube.com/watch?v=vFThM9BoTLg>

49:19 Outro