DocuQueryRAG

DocuQueryRAG is a Retrieval-Augmented Generation (RAG) system for question answering over documents (PDF, DOCX, PPTX, TXT). It uses sentence embeddings, FAISS vector search, and local LLMs (like Mistral via Ollama) to provide context-aware answers from uploaded files.

Features

• Supports multiple file types: PDF, DOCX, PPTX, TXT
• Text chunking and semantic embedding with all-MiniLM-L6-v2
• Fast similarity search with FAISS
• Flexible LLM integration via Ollama (default: mistral:7b)
• Command-line option to switch models (--model) , but first you have to download that model from ollama site.
• Interactive web UI powered by Gradio

---

Installation

1. Clone the repository:

git clone https://github.com/yourusername/DocuQueryRAG.git
cd DocuQueryRAG

Install dependencies:

pip install -r requirements.txtt

Run the app with the default Mistral model:

pip install -r requirements.txtt

python app.py --model mistral:7b

For changing the Model you just have to give the model name here.