RAG Researcher

RAG Researcher is a retrieval-augmented generation system for analyzing and understanding research papers, running entirely on local models so no data leaves your machine.

Features

• Local AI: Llama 3 via Ollama for generation and nomic-embed-text for embeddings, fully offline.
• Multi-query retrieval: generates several variations of each question to retrieve better-matching document chunks.
• PDF processing: robust loading and configurable text chunking, stored in a ChromaDB vector database.
• Interactive chat: terminal interface with streaming answers and the option to inspect the retrieved source chunks.

Tech stack: Python · LangChain · Ollama · Llama 3 · ChromaDB · RAG

Source code