weiwill88/Local_Pdf_Chat_RAG

Implements a complete RAG pipeline with modular components decomposing document loading, text chunking, embedding, vector storage (FAISS), and LLM generation into learnable stages. Combines dense vector retrieval with BM25 sparse retrieval, adds cross-encoder reranking and recursive retrieval for improved accuracy, and provides a Gradio interface for interactive learning. Supports pluggable LLM backends via auto-detection of local Ollama or SiliconFlow API endpoints.

842 stars. Actively maintained with 1 commit in the last 30 days.