Share This Post

News/PR

AWS backs deep-learning framework by extending ONNX support for Apache MXNet

AWS backs deep-learning framework by extending ONNX support for Apache MXNet

Amazon Web Services recently announced an open-source project – ONNX-MXNet, which is a Python package designed to import ONNX (Open Neural Network Exchange) based deep learning models into the Apache MXNet.

Apache MXNet is a deep learning framework that offers APIs to support multiple languages like Python, R and Scala. The ONNX support for MXNet will enable developers to build and train deep learning models with other frameworks like Microsoft Cognitive Toolkit, PyTorch or Caffe 2, and import them into MXNet.

The ONNX project, introduced back in September this year, by its creators Microsoft Corp. and Facebook Inc. is a format built for deep learning neural networks, which makes machines learn tasks instead of being explicitly programmed. The two companies also said that the framework will be standardized such that the models trained on one framework can be easily transferred to another.

With growing usage of deep learning and AI, more and more hardware vendors and frameworks are supporting ONNX, and AWS is another to add to this list.

The news is a positive development for artificial intelligence and deep learning practitioners who can flexibly build, deploy and move models between various frameworks.

Though Facebook, Microsoft and now AWS has announced support for ONNX, Google has not joined the army yet. It recently announced a lighter version of its open source deep learning framework called TensorFlow Lite which is in a developer preview.

AWS has also announced collaborating on ONNX support along with Facebook, Microsoft and the other deep learning communities to develop ONNX for making it more accessible and useful for the AI application developers.

Share This Post

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>