IPCC RAG Chatbot

This is a Streamlit app that uses a Retrieval-Augmented Generation (RAG) pipeline to answer questions from the IPCC AR6 WGIII report using a FAISS vector index and Databricks-hosted LLMs and embeddings.

Setup

1. Clone this repository
2. Add your Databricks credentials in Streamlit Cloud under Secrets:

DATABRICKS_HOST = "https://<your-workspace>.databricks.com"
DATABRICKS_TOKEN = "dapi-xxxxxxxxxxxxxxxx"

3. Deploy to Streamlit Cloud

Files

• app.py: Main Streamlit interface
• rag_core.py: Core logic for vector search and answer generation
• requirements.txt: Python dependencies
• Notebooks folder: Databricks notebooks used to create embeddings vectors and development (must be run on a Databricks cluster having required libraries installed)