Ask NERON API

A RAG-based API to answer questions about eMush using curated data.

Stack:

• FastAPI for the API, pytest for automated testing ;
• Chroma for the vector database allowing document retrieval ;
• Github Actions and Docker Swarm for automated deployment via a CI/CD pipeline on a self-managed Digital Ocean server : https://askneron.com/docs

Contributing

Prerequisites

• curl
• Docker Compose
• make
• uv

Installation

• Run the following command: curl -sSL https://raw.githubusercontent.com/cmnemoi/emush_rag/main/clone-and-install | bash
• Add an OpenAI API key to the .env file.

Usage

You can now ask questions to the API by running the following command:

curl -X POST "http://askneron.localhost/api/questions" \
     -H "Content-Type: application/json" \
     -d '{
           "question": "Do mycoalarms detect spore extraction?",
           "chat_history": []
         }'

Or by accessing the API Swagger at http://askneron.localhost/docs.

Indexing new documents

To improve the answer of the RAG model, you can index new documents in the vector database by putting them in the data directory and running the following command:

make index-documents

Development

• Lint code with make lint.
• Run tests with make test.

License

The source code of this repository is licensed under the AGPL-3.0-or-later License.