Khronos Launches NNEF, OpenVX NNE
The Khronos Group, the organization behind OpenGL, Vulkan, and other industry standards, is expanding their footprint when it comes to new neural network initiatives.
Khronos is today announcing the Neural Network Exchange Format (NNEF) as an alternative to proprietary formats used today by neural network software. NNEF is designed for portability across inference engines and other toolkits.
The OpenVX specification is meanwhile being expanded with a Neural Network Extension. This extension is an architecture for executing CNN-based inference in OpenVX graphs with import/export support to OpenVX objects.
Developers dealing with neural networks can learn more about these new standards push can read the Khronos press release for all of the details.
Khronos is today announcing the Neural Network Exchange Format (NNEF) as an alternative to proprietary formats used today by neural network software. NNEF is designed for portability across inference engines and other toolkits.
The OpenVX specification is meanwhile being expanded with a Neural Network Extension. This extension is an architecture for executing CNN-based inference in OpenVX graphs with import/export support to OpenVX objects.
Developers dealing with neural networks can learn more about these new standards push can read the Khronos press release for all of the details.
2 Comments