A new press release reports, “Synopsys, Inc. today announced support for the Open Neural Network Exchange (ONNX) format in the upcoming release of its DesignWare® ARC® MetaWare EV Development Toolkit, a complete set of tools, runtime software and libraries to develop vision and artificial intelligence (AI) applications for ARC EV6x Embedded Vision Processor IP.  ONNX is an open standard for representing deep learning models that enables trained models to be transferred between AI frameworks. By importing models in the ONNX format, Synopsys’ MetaWare EV Development Toolkit will enable developers to train models in any of the frameworks supporting ONNX and then map the models to the convolutional neural network (CNN) engine of the EV6x Processor Family.”

The release continues, “ONNX, a community project created by Facebook and Microsoft, is an open ecosystem for interchangeable AI models that provides a common way to represent neural network models. ONNX provides an open source format that enables models to be trained in one framework and transferred to another for inference. ONNX models are currently supported in Caffe2, Microsoft Cognitive Toolkit, MXNet, PaddlePaddle, and PyTorch, and there are connectors for many other common frameworks and libraries.”

John Koeter, vice president of marketing for IP at Synopsys, commented, “We see the need for greater interoperability in the AI tools community so developers don’t get locked into a single AI framework for the entire lifecycle of their project… By supporting ONNX in our ARC MetaWare Development Toolkit, Synopsys gives AI developers the ability to train models in one framework and later transfer them to another framework for inference. As a result, adopters of the EV6x’s CNN engine are able to choose the best combination of AI tools and frameworks for getting their designs to production in the shortest possible time.”

Read more at PR Newswire.

Photo credit: Synopsys