Skip to main content
Azure

Last month we introduced the Open Neural Network Exchange (ONNX) format with Facebook to increase interoperability and reduce friction for developing and deploying AI. Since then we’ve talked with many companies that share our goals and recognize the benefits of the ONNX open ecosystem. Today, Facebook and Microsoft are excited to share the following companies are announcing their support for the ONNX format:

We are also adopting the ONNX format widely at Microsoft. Today we are releasing preview support for ONNX in Cognitive Toolkit, our open source, deep learning toolkit. Cognitive Toolkit users can get started by following the instructions on GitHub to install the preview version. We encourage users to try it out and send us feedback.

We are also announcing that Project Brainwave, our FPGA-based deep learning platform for accelerating real-time AI, will support the ONNX format. By supporting it, Project Brainwave will help customers accelerate models from a variety of frameworks. The ONNX format also enables Project Brainwave innovation by letting us focus on acceleration optimizations instead of duplicative code for converting models.

It’s exciting times for AI. Just two weeks ago we announced major innovations to the Microsoft AI platform, including new and updated services and tools such as Azure Machine Learning and Visual Studio Code Tools for AI. We believe saving time with a productive AI platform while increasing interoperability will make it easier for developers to innovate and deploy AI on their terms. We are excited about the companies announcing support for the ONNX ecosystem today and invite others in the community to join the effort. You can visit https://onnx.ai to learn more.

Eric Boyd

Eric Boyd posts

Eric Boyd leads the global AI Platform team within Microsoft’s Cloud + AI division – a team whose mission is to deliver the Copilot Stack for Microsoft, Enterprises and Developer, powering every application with intelligent capabilities. To deliver on the mission, the team manages the deployment and optimization of a broad selection of best-in-class foundation and open source Generative AI models that support customer choice and cost effectiveness. The AI Platform team also spearheads the development of products, tools and services that help enterprises scale transformative AI initiatives, including Azure OpenAI Service, Azure Machine Learning, AI Services (like Speech, Language and Vision), AI Search fueled by RAG, Model Customization, Responsible AI/Content Safety, and live-site internal platforms that provide data, experimentation, and Graphics Processing Units (GPU) cluster management to groups across Microsoft.  

See Eric Boyd posts

Explore Microsoft Foundry

The future of AI starts here. Envision your next great AI app with the latest technologies. Get started with Azure.