A simple introduction to Vector Search

The demo is created to illustrate the basics of vector search using third-party APIs and open source models that could be run locally on CPU. You're free to play around with different models for embeddings and modify the code to run on GPU as well.

Setup

Create a virtual environment and activate it.

python -m venv <env_name>
source <env_name>/bin/activate

Clone the repository

Clone the repository.

git clone https://github.com/AbinayaM02/DevFest2023_VectorSearch.git
cd DevFest2023_VectorSearch

Install requirements

Install the necessary dependencies by executing the following command.

pip install -r requirements.txt

Run the demo

Once the setup is completed, download the embeddings file created for the demo variation 2 as per the instructions mentioned here and run the following command.

cd scripts
streamlit run search_demo.py

Cloud demo

The demo is also deployd to Streamlit cloud. You can run it directly from here. It is higly recommended to run the demo locally for demo variation 2.