Hugging Face x Microsoft

Hugging Face × Microsoft: Deepening Collaboration

Curiosity: How can partnerships between major AI platforms accelerate open-source AI adoption? What does the expanded Microsoft-Hugging Face collaboration mean for developers?

At MS BUILD, Satya Nadella announced the expansion of the partnership between Microsoft and Hugging Face to make open models and open-source AI more accessible across hardware.

Partnership Expansion

Retrieve: Key collaboration areas.

Focus Areas:

• 🆕 New AMD GPU support
• ☁️ Cloud through Azure AI
• 💻 Local inference with ONNX Runtime

Key Announcements

Innovate: Five major updates.

Announcement	Details	Impact
💻 Azure Model Catalog	Expanded Hugging Face collection with Llama 3, Mistral 7B, fine-tuned models	⬆️ Easy access
⚡ AMD MI300X	Optimized containers using Hugging Face TGI	⬆️ Performance
🤗 Phi-3 Family	Open release on Hugging Face	⬆️ Availability
🌐 WebGPU	Local inference with Transformers.js and ONNX Runtime Web	⬆️ Browser AI
💻 Dev Mode	Seamless integration between Spaces and VS Code	⬆️ Developer experience

Collaboration Benefits

Retrieve: Why this matters.

For Developers:

• ✅ Access to popular models in Azure
• ✅ Optimized hardware support
• ✅ Local inference capabilities
• ✅ Better development tools

For Ecosystem:

• ✅ More accessible open-source AI
• ✅ Cross-platform support
• ✅ Hardware optimization
• ✅ Unified developer experience

Architecture Overview

Innovate: Partnership ecosystem.

graph TB
    A[Microsoft Azure] --> B[Azure AI]
    A --> C[AMD MI300X VMs]
    A --> D[ONNX Runtime]
    
    E[Hugging Face] --> F[Model Catalog]
    E --> G[Text Generation Inference]
    E --> H[Transformers.js]
    E --> I[Spaces]
    
    B --> F
    C --> G
    D --> H
    I --> J[VS Code Dev Mode]
    
    style A fill:#e1f5ff
    style E fill:#fff3cd
    style J fill:#d4edda

Key Takeaways

Retrieve: Microsoft and Hugging Face expanded their partnership to make open-source AI more accessible across hardware, with support for AMD GPUs, Azure AI, ONNX Runtime, and improved developer tools.

Innovate: By leveraging this collaboration, you can access popular models in Azure, use optimized AMD hardware, run local inference with WebGPU, and develop seamlessly with VS Code integration.

Curiosity → Retrieve → Innovation: Start with curiosity about AI platform partnerships, retrieve insights from Microsoft-Hugging Face collaboration, and innovate by building applications that leverage both platforms’ strengths.

Next Steps:

• Explore Azure Model Catalog
• Try AMD MI300X optimization
• Test WebGPU inference
• Use VS Code Dev Mode

Learn More: https://huggingface.co/blog/microsoft-collaboration

Translate to Korean

---

허깅 페이스와 마이크로 소프트의 협력 강화

어제 MS BUILD에서 Satya Nadella는 새로운 AMD GPU, Azure AI를 통한 클라우드, ONNX Runtime 를 통한 로컬 추론을 시작으로 하드웨어 전반에 걸쳐 개방형 모델과 오픈 소스 AI에 더 쉽게 액세스할 수 있도록 Microsoft 와 Hugging Face 간의 파트너십 확장을 발표했습니다. 🚀

• 💻 Azure Model Catalog에서 Llama 3, Mistral 7B와 같은 인기 있는 개방형 모델 및 최고 성능의 미세 조정 모델을 사용하여 포옹 얼굴 컬렉션을 확장했습니다.
• ⚡ Hugging Face Text Generation Inference를 사용하여 Azure의 새로운 AMD MI300X VM에 최적화된 컨테이너
• 🤗 Hugging Face의 오픈 릴리스 Microsoft Phi-3 제품군
• 🌐 Transformers.js 및 ONNX 런타임 웹을 사용한 WebGPU를 통한 로컬 추론
• 💻 Hugging Face Spaces와 VS Code 간의 원활한 통합을 위한 새로운 개발 모드

자세히 알아보기: https://huggingface.co/blog/microsoft-collaboration