LlamaIndex RAG Examples

A compact, hands-on collection of Retrieval-Augmented Generation (RAG) reference implementations built with LlamaIndex. Each example is a self-contained Jupyter notebook that you can read, run, and adapt to your own documents and models.

The repository contrasts two complementary approaches:

Example	Technique	Best for	LLM / Backend
GraphRAG	Property-graph knowledge index	Complex, multi-hop questions that connect facts across documents	OpenAI
Simple Vector Index	Dense vector retrieval	Straightforward semantic search / Q&A over documents	Groq + Hugging Face

---

Why two examples?

• GraphRAG builds a knowledge graph over your corpus, so it can reason across multiple sources and answer questions that require chaining several facts together (multi-hop reasoning). The notebook also builds a baseline vector index side-by-side so you can directly compare the two approaches on the same questions.
• Simple Vector Index is the classic RAG pipeline: parse documents, embed them, store the vectors, and answer questions with an LLM. It is the fastest way to get a working RAG system running over your own PDFs.

If you are new to RAG, start with the Simple Vector Index Example. If you need to answer complex questions that span multiple documents, explore the GraphRAG Example.

---

Repository structure

llamaindex-rag-examples/
├── GraphRAG Example/              # Graph-based RAG with multi-hop reasoning (OpenAI)
│   ├── data/                      # Sample news corpus (JSON)
│   ├── main.ipynb                 # Walkthrough notebook
│   ├── requirements.txt           # Example-specific dependencies
│   ├── .env.example               # Template for API keys
│   └── README.md
├── Simple Vector Index Example/   # Classic vector RAG (Groq + Hugging Face)
│   ├── data/                      # Sample PDF document
│   ├── main.ipynb                 # Walkthrough notebook
│   ├── requirements.txt           # Example-specific dependencies
│   ├── .env.example               # Template for API keys
│   └── README.md
├── requirements.txt               # Combined dependencies for both examples
├── LICENSE
└── README.md

---

Quick start

Each example is independent and ships with its own README.md, requirements.txt, and .env.example. In general:

# 1. Clone the repository
git clone https://github.com/<your-username>/llamaindex-rag-examples.git
cd llamaindex-rag-examples

# 2. Move into the example you want to run
cd "Simple Vector Index Example"      # or: cd "GraphRAG Example"

# 3. Create and activate a virtual environment
python -m venv .venv
source .venv/bin/activate              # On Windows: .venv\Scripts\activate

# 4. Install dependencies
pip install -r requirements.txt

# 5. Configure your API keys
cp .env.example .env                   # then edit .env

# 6. Launch Jupyter and open main.ipynb
jupyter notebook

See each example's README.md for the exact API keys it requires.

---

Requirements

• Python 3.11+
• A Jupyter environment (jupyter notebook or jupyter lab)
• API keys depending on the example:
  • GraphRAG Example — an OpenAI API key
  • Simple Vector Index Example — a Groq API key, a LlamaIndex Cloud (LlamaParse) key, and a Hugging Face token

---

Author

Written and developed by Amin Amiri.

License

Released under the MIT License. © 2026 Amin Amiri.