LLM-RAG-demo

About

This repository provides a small, end-to-end Retrieval-Augmented Generation (RAG) demo inspired by our work on DMP-Chef.
RAG is a practical approach for improving the accuracy and trustworthiness of LLM outputs by retrieving relevant context from a local knowledge base (e.g., web pages, PDFs, notes) and using that context during generation.

In this demo, we collect a small snapshot (~20 pages) from the FAIR Data Innovations Hub website, save it locally as .txt, build a FAISS vector index using Ollama embeddings, and compare No-RAG vs RAG answers side by side.

👉 Blog post: How to Quickly Set Up a RAG System: A Practical Guide Inspired by Our Work on DMP-Chef
👉 Code repo: fairdataihub/LLM-RAG-demo

---

Standards followed

The overall codebase is organized in alignment with the FAIR-BioRS guidelines. The Python code in the primary Jupyter notebook, main.ipynb, follows PEP 8 style conventions (including comments and docstrings). All required dependencies are listed in requirements.txt.

---

1) Clone the repository

git clone https://github.com/fairdataihub/LLM-RAG-demo.git
cd LLM-RAG-demo

---

2) Create and activate a Python environment

macOS / Linux

python3 -m venv .venv
source .venv/bin/activate

Windows (PowerShell)

python -m venv .venv
.\.venv\Scripts\Activate.ps1

Upgrade pip (recommended):

python -m pip install --upgrade pip

---

3) Install dependencies

pip install -r requirements.txt

If you don’t have a requirements.txt yet, create one (recommended) and include:

• langchain, langchain-community, langchain-ollama
• faiss-cpu
• requests, beautifulsoup4

---

4) Install and run Ollama

Install Ollama: https://ollama.com

Make sure Ollama is running:

ollama serve

Pull the models used in this demo:

ollama pull llama3.2
ollama pull nomic-embed-text

Optional: confirm Ollama is reachable

• Open: http://localhost:11434/api/tags in your browser
  or

curl http://localhost:11434/api/tags

---

5) Run the demo (recommended order)

This repo may include the demo as a script, notebook, or both. Use whichever exists in your repo structure.

Run as a Jupyter notebook

If you have a notebook (RAG_Application.ipynb):

pip install notebook
jupyter notebook

Then open the notebook and run cells in order:

1. Crawl website pages into data/fairdata_texts/
2. Split into chunks
3. Build FAISS index (saved locally)
4. Compare No-RAG vs RAG outputs

---

6) Expected outputs

After crawling:

• data/fairdata_texts/*.txt

After indexing:

• faiss_index_fairdata/ (or similar folder)

When you run the comparison step, you should see side-by-side answers for:

• No-RAG (LLM answers without documents)
• RAG (LLM answers using retrieved context)

---

License

This work is licensed under the MIT License. See LICENSE for more information.

---

Feedback and contribution

Use GitHub Issues to submit feedback, report problems, or suggest improvements.
You can also fork the repository and submit a Pull Request with your changes.

---

How to cite

If you use this code, please cite this repository using the versioned DOI on Zenodo for the specific release you used (see CITATION.cff file)