Invoice data processing with Llama2 13B LLM RAG on Local CPU

Youtube: Invoice Data Processing with Llama2 13B LLM RAG on Local CPU

---

Quickstart

RAG runs on: LlamaCPP, Haystack, Weaviate

1. Download the Llama2 13B model, check models/model_download.txt for the download link.
2. Install Weaviate local DB with Docker

docker compose up -d

3. Install the requirements:

pip install -r requirements.txt

4. Copy text PDF files to the data folder.
5. Run the script, to convert text to vector embeddings and save in Weaviate vector storage:

python ingest.py

6. Run the script, to process data with Llama2 13B LLM RAG and return the answer:

python main.py "What is the invoice number value?"