📘 Retrieval-Augmented Generation (RAG) PDF QA System

A lightweight Retrieval-Augmented Generation (RAG) system designed to answer queries based on NCERT textbook PDFs using transformer embeddings and large language models (LLMs). Optimized to run on home computers with minimal resources.

---

✨ Features

• PDF-based question answering with semantic chunking
• Fast document retrieval using FAISS
• Embeddings from MPNet, MiniLM, and fine-tuned MiniLM
• Streamlit interface for easy interaction and model comparison
• Fine-tuned using Multiple Negatives Ranking (MNR) loss
• Query handling through OpenRouter LLM API (e.g., Gemma, DeepSeek, LLaMA)