https://github.com/micheldumontier/gba-rag
This is a demonstration project to provide LLM-based question answering over the website and documents of German Federal Joint Committee (G-BA).
https://github.com/micheldumontier/gba-rag
health llama3 llamaindex llm qdrant-vector-database
Last synced: about 1 month ago
JSON representation
This is a demonstration project to provide LLM-based question answering over the website and documents of German Federal Joint Committee (G-BA).
- Host: GitHub
- URL: https://github.com/micheldumontier/gba-rag
- Owner: micheldumontier
- License: mit
- Created: 2024-07-12T17:20:10.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2024-07-30T14:17:01.000Z (almost 2 years ago)
- Last Synced: 2025-03-28T15:47:02.880Z (about 1 year ago)
- Topics: health, llama3, llamaindex, llm, qdrant-vector-database
- Language: Jupyter Notebook
- Homepage:
- Size: 31.3 KB
- Stars: 2
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# gba-rag
This is a demonstration project to provide question answering over the website and documents of German [Federal Joint Committee (G-BA)](https://www.g-ba.de/). The G-BA is the highest decision-making body of the joint self-government of physicians, dentists, hospitals and health insurance funds in Germany.
This project uses the following technologies:
- [LLamaIndex](https://www.llamaindex.ai/) framework
- [Ollama](https://ollama.com/) to serve up an LLM (namely LLama3)
- [Fastembed](https://github.com/qdrant/fastembed) to embed text with lightweight embedding models
- [Qdrant](https://qdrant.tech/) vector store to store and retrieve embeddings
It also assumes you have a copy of the website.
## Deployment
### prepare the local environment
```bash
python -m venv .venv
```
### activate the local environment
```bash
source .venv/bin/activate
```
### Deploy the vector database
The `docker-compose.yml` file contains the basic setup for running the Qdrant vector store. You can modify the basic settings there and in the `qdrant_config.yml` file (including the password). Currently, you'll need to modify the notebook if you changed the default settings.
```bash
cd vectordb
docker compose up -d
```
### Open and run the ipython notebook
The `notebook/gba-llamaindex.ipynb` notebook contains the code to run through the demo project.