PyTorch 1.0 takes the modular, production-oriented capabilities from Caffe2 and ONNX and combines them with PyTorch's existing flexible, research-focused design to provide a fast, seamless path from research prototyping to production deployment for a broad range of AI projects. With PyTorch 1.0, AI developers can both experiment rapidly and optimize performance through a hybrid front end that seamlessly transitions between imperative and declarative execution modes. Data Scientists can develop models in PyTorch 1.0, which are saved in ONNX as the native format and directly use them in applications built on Windows ML and other platforms that support ONNX.
At Microsoft we believe bringing AI advances to all developers, on any platform, using any language, in an open and interoperable AI ecosystem, will help ensure AI is more accessible and valuable to all. Microsoft’s support for ONNX is an example of this – ONNX allows developers to choose the right framework for their task, framework authors can focus on innovative enhancements, and hardware vendors can streamline optimizations.
Azure Machine Learning Services provides support for a variety of frameworks including TensorFlow, Microsoft Cognitive Toolkit, and soon PyTorch 1.0 is another example. Azure infrastructure services, of course, lets you use any framework, even beyond this list because it is an open compute fabric with cutting edge hardware like the latest GPUs. Microsoft is also deeply engaged in the AI Open Source community with Visual Studio Code Tools for AI, Microsoft Cognitive Toolkit and ONNX to provide transparency, faster innovation and interoperability.
To learn more about Microsoft’s Azure AI Platform visit http://azure.com/ai.