🧾 Invoice Data Extraction with LLM RAG on CPU

This project is a local Retrieval-Augmented Generation (RAG) pipeline built with LangChain, ChromaDB, and Ollama to extract structured information from invoice PDFs using LLaMA3.2:latest running on CPU.

✅ No GPU required. ✅ Entirely offline and private. ✅ Powered by sentence-transformers (all-mpnet-base-v2) for embeddings and ChromaDB for vector storage.

---

⚙️ Features

• Upload any text-based PDF invoice

• Ask structured or natural language questions like "What is the invoice number?" or "Who is the client?"

• Built on LLaMA3.2:latest (7B model via Ollama)

• Uses Sentence Transformers for semantic chunking and vector similarity

• Full offline support, privacy-preserving, and works on CPU

• UI built with Streamlit for PDF upload and querying

---

🧪 Tech Stack

• 🧠 LLM: LLaMA3.2:latest via Ollama

• 🧲 Embedding Model: sentence-transformers/all-mpnet-base-v2

• 📚 LangChain: For RAG pipeline and chaining

• 📦 ChromaDB: Fast and simple vector store

• 📄 PDF Loader: LangChain’s PyPDFLoader

• 🎛️ Streamlit: UI interface

---

🛠️ Setup Instructions

1. Install dependencies

pip install -r requirements.txt

2. Install and set up Ollama

Follow instructions at https://ollama.com Then pull the model:

ollama pull llama3:latest

---

3. Prepare PDF data

Place your text-based invoice PDFs in the data/ directory.

---

4. Generate vector store from PDF

python ingest.py

---

5. Run the app with question:

python main.py "What is the client name?"

---

📁 Folder Structure

.
├── data/                # Folder with PDF invoices
├── vectorstore/         # Persisted ChromaDB vectors
├── rag/
│   ├── pipeline.py      # RAG chain builder
├── app.py               # Streamlit interface
├── ingest.py            # PDF to vector processor
├── main.py              # CLI for querying
├── config.yml           # Config settings
├── requirements.txt

---

📌 Example Query

python main.py "What is the invoice date?"

Output:

{
  "invoice_date": "2024-03-18"
}

---