fairDataSociety / fave-pdf-reader

fave-pdf-reader

This is a demo application for Fave intigrated with langchain. You can upload any pdf file and embed the content in FaVe. Also you can ask questions once the upload is finished.

Prerequirements

• ENS based FDS account
• Docker

Run Fave

We need docker to run fave and a vectorizer service. You can use this docker compose file to quickly setup FaVe and vectorizer.

NOTE: Please update the required details to run FaVe in the docker compose file.

How to run this docker compose file

• Copy the code
• Create a file in your local system docker-compose.yaml.
• Paste the code in that file and change the values for BEE_API_ENDPOINT, BLOCKCHAIN_RPC_ENDPOINT, BATCH_ID, FDS_USERNAME, FDS_PASWORD, POD_NAME.
• Make sure Docker is running. Then is the same directory pf the docker-compose.yaml file, run docker compose up.

docker compose up will download both the images from dockerhub and run them. After you run the command you FaVe instance should be running on port 1234.

How to run this Demo app

• Run pip install -r requirements.txt.
• Change the collection name in the app.py file.
• You have to get HUGGINGFACEHUB_API_TOKEN from here https://huggingface.co/docs/hub/security-tokens
• Rename .env.example file to .env and paste the token. HUGGINGFACEHUB_API_TOKEN=<TOKEN>
• Run streamlit run app.py.