Deep Critic is an integrated web application designed to provide comprehensive reviews of academic papers using multiple Large Language Models (LLMs). It combines models from OpenAI, Anthropic, and Mistral to generate structured, consensus-driven feedback resembling peer-review processes at conferences such as NeurIPS.

• Integration with multiple LLMs (OpenAI, Claude, Mistral)
• PDF upload, extraction, and processing
• Automatic generation of structured paper reviews
• Consensus reviews combining insights from multiple models
• User-friendly React frontend interface

• Docker and Docker Compose
• Node.js and npm (for frontend development)
• API keys for OpenAI, Anthropic, and Mistral

git clone https://github.com/CompMotifs/deep-critic.git
cd deep-critic

Create a .env file from the provided template and add your API keys:

cp .env_template .env

Edit .env and populate your API keys:

OPENAI_API_KEY=your-openai-key
ANTHROPIC_API_KEY=your-claude-key
MISTRAL_API_KEY=your-mistral-key

Build both frontend and backend services:

docker compose build

Start all services:

docker compose up

Once running, the frontend is accessible at http://localhost:3000, and the backend API is accessible at http://localhost:8000.

deep-critic/
├── backend/
│   ├── api/               # API routes and endpoints
│   ├── app/
│   │   ├── review_engine/ # Paper review logic
│   │   └── services/      # Integration with external services (LLMs, PDF processing)
│   └── requirements.txt   # Python dependencies
│
├── frontend/
│   ├── public/            # Static files
│   │   └── index.html
│   ├── src/               # React components
│   │   ├── App.js
│   │   ├── DeepCriticApp.js
│   │   ├── DeepCriticApp.css
│   │   └── index.js
│   ├── package.json
│   ├── Dockerfile
│   └── Dockerfile.dev
│
├── docker-compose.yml
├── .env_template
└── README.md

• Improve and extend API functionality
• Enhance frontend user experience
• Add automated validation and batch review analysis
• Improve PDF processing performance
• Incorporate additional language models and inter-model communication
• Set up robust user management, authentication, database integration, and CI/CD

If you encounter issues:

1. Confirm that all API keys are correctly set in your .env file.
2. Check logs from Docker Compose:

docker compose logs backend
docker compose logs frontend

3. Verify port availability (default frontend: 3000, backend: 8000).
4. Ensure uploaded PDFs are readable and not corrupted.