Create FAISS vector db and chunk PDF docs for query by Lllama 2 (UI and CMD line versions)
If don't already have (you likely have the rest used): pip install langchain pip install pypdf pip install faiss-cpu pip install ctransformers pip install sentence-transformers pip install streamlit
Prepare your PDF docs in single location folder. Update create_db.py with the path to your docs. Run create_db.py and ensure vectorstore was created. Download your model locally (for this, it should be a GGML model).
To run with Streamlit UI: use myui.py after updating paths. 'streamlit run myui.py'
To run CMD line: Update main.py with all your paths, etc. Run main.py with a query. Example: python main.py "What are 3 types of verifications that can be used for Medicaid SSI unearned income?"