Indic LLM Arena

A production-ready platform for interacting with and comparing leading LLMs from India. Built with Django REST Framework and React.

🌟 Features

Core Functionality

• 🤖 Multi-Model Chat - Interact with various AI models from leading providers
• 🔄 Comparison Mode - Compare responses from two models side-by-side
• 🎲 Random Mode - Blind test models without knowing which is which
• 📊 Real-time Streaming - Stream AI responses as they're generated
• 🌳 Message Branching - Create conversation branches and explore different paths
• 📤 Share & Export - Share conversations via links or export as JSON/Markdown/TXT

Advanced Features

• ⭐ Comprehensive Feedback - Rate responses and provide detailed feedback
• 🏆 ELO Leaderboard - Track model performance with ELO ratings
• 🔐 Dual Authentication - Google OAuth & Anonymous sessions
• 💾 Session Persistence - Continue conversations across sessions
• 📱 Responsive Design - Works seamlessly on desktop and mobile

🏗️ Architecture

Backend (Django)

backend/
├── apps/
│   ├── user/          # User authentication and profiles
│   ├── ai_model/      # AI model registry and integration
│   ├── chat_session/  # Session management
│   ├── message/       # Message handling and streaming
│   ├── feedback/      # User feedback system
│   └── model_metrics/ # Performance tracking and leaderboards
├── core/              # Core settings and configurations
└── requirements.txt   # Python dependencies

Frontend (React)

frontend/
├── src/
│   ├── app/          # App configuration and routing
│   ├── features/     # Feature-based modules
│   │   ├── auth/     # Authentication components
│   │   ├── chat/     # Chat interface and logic
│   │   ├── models/   # Model selection and testing
│   │   ├── feedback/ # Feedback components
│   │   └── leaderboard/ # Leaderboard views
│   ├── shared/       # Shared utilities and components
│   └── styles/       # Global styles
└── package.json      # Node dependencies

🚀 Getting Started

Prerequisites

• Python 3.11+
• Node.js 18+
• PostgreSQL 14+
• Redis 6+
• Google Cloud account (for OAuth)

Backend Setup

1. Clone the repository

git clone https://github.com/AI4Bharat/Chat-Arena.git
cd Chat-Arena/backend

2. Create virtual environment

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Set up environment variables

cp .env.example .env

Edit .env with your credentials:

# Database
DATABASE_URL=postgresql://user:password@localhost:5432/ai_playground

# Redis
REDIS_URL=redis://localhost:6379

# AI Model API Keys
OPENAI_API_KEY=your-openai-key
GOOGLE_API_KEY=your-google-key
ANTHROPIC_API_KEY=your-anthropic-key
META_API_KEY=your-meta-key
MISTRAL_API_KEY=your-mistral-key

# Firebase/Auth
FIREBASE_CONFIG={"apiKey":"...","authDomain":"..."}

# Security
SECRET_KEY=your-secret-key
ALLOWED_HOSTS=localhost,127.0.0.1

5. Run migrations

python manage.py migrate
python manage.py createsuperuser

6. Start backend services

# Terminal 1: Django server
python manage.py runserver

# Terminal 2: Celery worker
celery -A core worker -l info

# Terminal 3: Celery beat (for scheduled tasks)
celery -A core beat -l info

Frontend Setup

1. Navigate to frontend directory

cd ../frontend

2. Install dependencies

npm install

3. Set up environment variables

cp .env.example .env

Edit .env:

REACT_APP_API_URL=http://localhost:8000/api
REACT_APP_WS_URL=ws://localhost:8000/ws
REACT_APP_GOOGLE_CLIENT_ID=your-google-client-id

4. Start development server

npm run dev

The application will be available at http://localhost:5173

Google OAuth Setup

1. Go to Google Cloud Console
2. Create a new project or select existing
3. Enable Google Identity Toolkit API
4. Create OAuth 2.0 credentials
5. Add authorized origins:
   • http://localhost:5173 (development)
   • https://your-domain.com (production)

📱 Usage

Chat Modes

1. Direct Mode - Chat with a single AI model
2. Compare Mode - Get responses from two models simultaneously
3. Random Mode - Blind test with randomly selected models

Providing Feedback

• Star Ratings - Rate individual responses (1-5 stars)
• Preferences - Choose which model gave a better response
• Categories - Tag feedback with categories (accuracy, creativity, etc.)

Viewing Leaderboards

• Filter by category (Overall, Creative Writing, Coding, Reasoning)
• View different time periods (Daily, Weekly, All-time)
• See detailed model performance metrics

🔧 Configuration

Adding New AI Models

1. Create a model adapter in backend/apps/ai_model/adapters/
2. Register the model in Django admin
3. Add API credentials to environment variables

Customizing Categories

Edit backend/apps/feedback/constants.py to add/modify feedback categories.

🚢 Deployment

Backend Deployment (Example with Railway/Render)

1. Set up PostgreSQL and Redis instances
2. Configure environment variables
3. Deploy using provided Dockerfile or railway.toml

Frontend Deployment (Example with Vercel)

npm run build
vercel --prod

📊 API Documentation

The API documentation is available at:

• Swagger UI: http://localhost:8000/api/docs/
• ReDoc: http://localhost:8000/api/redoc/

🧪 Testing

Backend Tests

python manage.py test

Frontend Tests

npm run test

🤝 Contributing

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Built with assistance from Claude (Anthropic) - AI pair programming and architecture design
• Thanks to all the AI model providers for their APIs
• The open-source community for the amazing tools and libraries

👨‍💻 Credits

This project was developed with significant assistance from Claude 3 Opus, an AI assistant by Anthropic. Claude helped with:

• System architecture design
• Code implementation
• Best practices and patterns
• Documentation
• Problem-solving and debugging

Special thanks to the Anthropic team for creating such a capable AI assistant that made building this complex project possible!

📞 Support

For support, please open an issue in the GitHub repository or contact the maintainers.

---

Built with ❤️ and AI assistance