This is a Backend App for https://github.com/kvchandu/MedQuery

Create and activate a virtual environment with python 3.11, preferably using conda. One can install conda by following the steps here

conda create -n med-qa python=3.11
conda activate med-qa

Install python requirements

pip install -r requirements.txt

Paste ur openai apikey in the src/resources/config file

open_ai:
  api_key: your-key

Run the indexing script to create the vector-store for this dataset https://huggingface.co/datasets/epfl-llm/guidelines. Give the crop_size argument to specify the size of the dataset you want to crop. The normalization will be performed accordingly

python src/index_epfl.py <crop_size>

Start the fast-api server

uvicorn src.app:app --host 0.0.0.0