Building a Modern Local RAG System with Ollama and LangChain

Interactive knowledge graph visualization of Graph Neural Network (GNN) research papers using Retrieval-Augmented Generation (RAG).

Nikhil Singh

Building ML solutions is art.

Local RAG implementation on the topic of GNN

Introduction

Motivation: Navigating the GNN Research Landscape

Project Objectives

Architecting the solution

Ollama & Llama3
LangChain
DuckDB
Streamlit

Building the GNN Knowledge Graph Pipeline

PDF Text Extraction
Information Extraction: Unveiling Entities and Relationships
Knowledge Graph Construction

Retrieval-Augmented Generation (RAG) in Action

Interactive Visualization: Exploring the GNN Universe

Key Insights and Lessons Learned

Codebase: Running locally

Conclusion and Future Directions