Testing the lightweight RAG System on local setup using Ollama for local LLM operations.

This project demonstrates a lightweight RAG system using two Jupyter notebooks:

• lightrag-ingestion-notebook.ipynb - Handles document processing and embedding generation
• lightrag-query-notebook.ipynb - Implements querying and response generation

• Fully local RAG implementation using Ollama
• Uses nomic-embed-text for embeddings
• Supports PDF and text document ingestion
• Chunking with configurable overlap
• Vector similarity search for relevant context retrieval
• Flexible LLM model selection for response generation

• Python 3.x
• Ollama installed and running locally
• Required Ollama models:
  • nomic-embed-text (for embeddings)
  • Your choice of LLM (e.g., llama2, deepseek-r1:8b)

1. Pull and start required Ollama models:

   ollama pull nomic-embed-text
   ollama pull deepseek-r1:8b
   

2. Verify that the models are running:

   ollama list

3. Run the ingestion notebook to process documents and generate embeddings:

   jupyter notebook lightrag-ingestion-notebook.ipynb

4. Run the query notebook to query the processed documents and generate responses:

    jupyter notebook lightrag-query-notebook.ipynb