Skip to content

Latest commit

 

History

History
40 lines (30 loc) · 1.32 KB

README.md

File metadata and controls

40 lines (30 loc) · 1.32 KB
Raw

Infinispan vectors - Python demo

A set of demo on how to use Infinispan as a vector DB in python projects

Requirements

  • Python 3.9 environment
  • pip
  • Jupiter Lab
  • Docker (or you'll need to start Infinispan before demo)

Setup

pip install jupyterlab
jupyter-lab

jupiter-lab will open a web gui in your browser, from there you can run the demo notebooks

similarity-search-demo-1.ipynb

Based on all-MiniLM-L12-v2 model. Relay on a 28k+ news db that you can use to do similarity search.

similarity-search-demo-2.ipynb

Based on all-MiniLM-L12-v2 model. Performs a similarity search on a set of random sentences. Content is store is a separate cache and referred in the vector cache via key id.

question-answer-demo.ipynb

This demo needs an openAI api key: it must be stored in a .env file in the form OPENAI_API_KEY=<your-key-here>

Implement a question answer bot that can answer to question about pdf AI book provided as an input data. For this an OpenAI api key is needed.

Run Infinispan

By default, Infinispan is run in a docker container at the start of the demo. If Docker is not a good option, you can download an Infinispan 15+ and run it with the provided infinispan-noauth.yaml configuration. Remember to skip the docker cell when running the demo.