Hugging Face's logo

Spaces:
Wills17
/
Document-RAG-System
Running

App Files Community
Document-RAG-System / README.md
Wills17's picture
Wills17
Update README.md
fffc3d6 verified
preview code
|
raw
history blame
3.42 kB
metadata
title: Document RAG Chatbot
emoji: 🤖
colorFrom: indigo
colorTo: purple
sdk: docker
app_file: flask_app.py
pinned: false
short_description: An intelligent chatbot that understands your documents

Document RAG Chatbot

An intelligent, context-aware chatbot that understands your documents.
Upload a PDF or text file, and it will answer questions using only the information inside — no hallucinations, no fluff.

Built with Flask, LangChain, and Google Gemini, this project demonstrates a clean, modular approach to Retrieval-Augmented Generation (RAG) in action.

Highlights

  • Upload and analyze PDF or TXT files
  • Uses FAISS for fast semantic search
  • Embeddings powered by HuggingFace Sentence Transformer
  • Answers generated by Gemini 2.5 Flash
  • Works with frontend-provided API key (no server-side storage)
  • Clean, responsive interface for smooth chat interaction

Tech Stack

Component Technology
Backend Flask
Language Model Google Gemini
Vector Store FAISS
Embeddings HuggingFace all-MiniLM-L6-v2
Frontend HTML, CSS, JavaScript
Framework LangChain

Getting Started

1️⃣ Clone the Repository 

git clone https://huggingface.co/spaces/<your-username>/Document-RAG-System
cd Document-RAG-System

2️⃣ Set Up Your Environment 

python -m venv venv
# Activate
venv\Scripts\activate     # Windows
source venv/bin/activate  # macOS/Linux

3️⃣ Install Dependencies 

pip install -r requirements.txt

Run the App 

python flask_app.py

Once running, open your browser and go to: 👉 http://127.0.0.1:5000/

Get Your Gemini API Key

  1. Visit Google AI Studio
  2. Generate a Gemini API Key
  3. Paste it in the “API Key” field on the webpage
  4. Save and start chatting!

Your key is never stored — it’s used only in your current session.

How It Works

Here’s what happens behind the scenes:

  1. You upload your document.
  2. The file is split into small chunks (for efficient retrieval).
  3. Each chunk is embedded into a vector using HuggingFace embeddings.
  4. FAISS indexes these vectors for quick similarity search.
  5. When you ask a question, relevant chunks are retrieved and sent to Gemini.
  6. Gemini generates a focused, contextual answer — grounded in your document.

That’s Retrieval-Augmented Generation (RAG) in a nutshell.

Example Use Cases

  • Summarize long reports
  • Extract key information from research papers
  • Study assistant for textbooks
  • Legal, medical, or technical document Q&A
  • Analyze and interpret crypto project whitepapers — understand tokenomics, roadmap, and team details before investing

Customization

You can tweak:

  • chunk_size and chunk_overlap in chatbot.py
  • The system message for tone or depth
  • The Gemini model (to gemini-2.5-flash, gemini-1.5-pro, etc.)

Author

Williams Odunayo Machine Learning Engineer | Builder of useful AI systems😉 🔗 GitHub • LinkedIn

License

Released under the MIT License. Free to use, modify, and build upon - attribution is appreciated.