README.md

metadata

title: FathomDeepResearch
emoji: 🧮
colorFrom: blue
colorTo: red
sdk: docker
app_port: 7860
pinned: false
license: apache-2.0
short_description: Advanced research AI with web search capabilities

🔬 FathomDeepResearch

Advanced AI research agent powered by Fathom-Search-4B and Fathom-Synthesizer-4B models. This app provides deep research capabilities with real-time web search and intelligent synthesis.

🚀 Features

• 🧠 Advanced Reasoning: Powered by Fathom-R1-14B for sophisticated thinking
• 🔍 Real-time Web Search: Integrated search across multiple sources
• 📊 Intelligent Synthesis: Combines search results into coherent answers
• 🎨 Rich UI Components: Streamlined chat interface with progress tracking
• ⚡ Fast Performance: Optimized for Hugging Face Spaces

🛠️ How to Use

1. Enter your research question in the text box
2. Click "Research" to start the deep research process
3. Watch progress as the AI searches and synthesizes information
4. Get comprehensive answers with source citations

💡 Example Questions

• "What are the latest AI developments in 2024?"
• "DeepResearch on climate change solutions"
• "UPSC 2025 preparation strategy"
• "Comparative analysis of electric vehicle adoption"

🔧 Technical Details

Models Used

• Fathom-Search-4B: For web search and retrieval
• Fathom-Synthesizer-4B: For answer synthesis
• Fathom-R1-14B: For reasoning and planning

Architecture

• Backend: FastAPI with Gradio integration
• Frontend: React-based chat interface
• Search: Multi-source web search with Serper API
• Deployment: Docker containers optimized for HF Spaces

📋 Requirements

• Python 3.10+
• Transformers 4.35+
• Gradio 4.0+
• FastAPI
• Hugging Face Transformers

🌐 Deployment

This app is deployed on Hugging Face Spaces using Docker. The setup includes:

• Automatic model downloading
• Environment configuration
• Error handling and fallbacks
• Multi-modal capabilities

📖 License

Apache 2.0 License - See LICENSE file for details

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📞 Support

For issues or questions:

• Check the docs folder for detailed documentation
• Open an issue on the repository
• Contact the development team

🧩 Building the Docker image locally (private Hugging Face repo)

If the source is in a private Hugging Face Space, provide a token when building the image. The Dockerfile clones the repository during build using the build-arg HF_API_TOKEN.

Examples (PowerShell):

Provide token as a build-arg (less secure, visible in image history):

docker build -t fathom-deploy --build-arg HF_API_TOKEN=hf_xxx .

Using BuildKit and a secret (recommended):

 $env:DOCKER_BUILDKIT=1; docker build --secret id=hf_token,src=$env:USERPROFILE\.hf_token -t fathom-deploy .

Place your token in a file (e.g. %USERPROFILE%.hf_token) containing only the token string, then reference it with --secret. You would need to adapt the Dockerfile to read from /run/secrets/hf_token if you choose this approach.

Note: If the repository is public you can omit the build-arg and the Dockerfile will clone anonymously.