bRAG-langchain
A hands-on notebook series that teaches Retrieval-Augmented Generation from an introductory setup through advanced techniques such as multi-querying, logical and semantic routing, and custom RAG pipeline construction.
bRAG-langchain is a collection of Jupyter notebooks that teach Retrieval-Augmented Generation step by step, starting from a boilerplate, fully customizable RAG chatbot and progressing to more advanced constructions. An introductory notebook covers environment setup, document loading and preprocessing, embedding generation, setting up a vector store such as ChromaDB or Pinecone, and a basic retrieval-and-generation pipeline as a baseline.
Later notebooks build on that baseline with multi-query techniques that diversify retrieval and compare results against a single-query pipeline, and with routing and query construction, including function-based logical routing that classifies a query by programming language and embedding-based semantic routing that directs a question toward a math or physics prompt, plus structured search schemas for filtering by metadata.