This repository provides the instructor embedding model as an API service. This can be run on CPU or GPU with any of the 3 model sizes, which you can set in the .env file.
We have provided the script debian_docker_install.sh for installing docker on a Debian VM with support for GPU. This script was tested on Google Cloud with L4 GPU.
sh debian_docker_install.sh
To build an run the API service first copy the .env file
cp env-example.txt .env
and then launch with:
docker compose up
The API service documentation can be accessed at http://127.0.0.1:8000/docs.