DEV Community


AWS re:Invent 2019 — AI/ML recap — Part 3: Frameworks & Infrastructure

Julien Simon
Global Evangelist, AI & Machine Learning, Amazon Web Services
Originally published at Medium on ・3 min read

AWS re:Invent 2019 — AI/ML recap — Part 3: Frameworks & Infrastructure

In the last two previous posts, I introduced you to our new high-level services, and to the new capabilities we added to Amazon SageMaker. Thought I was done, didn’t ya? Nope, there’s more. There’s ALWAYS more.

In this final post, let’s talk about frameworks and infrastructure.

As always, happy to answer questions here or on Twitter.

Amazon DeepComposer

Not sure when this one fits! AWS DeepComposer is the world’s first musical keyboard combined with a generative AI service.



Amazon EC2 Inferentia (Inf1)

Inf1 instances are powered by AWS Inferentia chips, and are designed to provide you with fast, low-latency inferencing.

Using the Neuron SDK, which is pre-integrated into popular machine learning frameworks like TensorFlow, MXNet and Pytorch, you can compile and load your models on Inf1.


Neuron SDK:

TensorFlow support

TensorFlow 1.15 is now supported in Deep Learning AMIs, Deep Learning containers and SageMaker.

TensorFlow 2.0 is also available in the Deep Learning AMIs. I hear that it’s coming soon to Deep Learning containers and SageMaker as well, but shhhh.

By the way, did you know that 85% of all TensorFlow workloads in the cloud run on AWS? Check out the report for more details (PDF).

Deep Graph Library

The Deep Graph Library is an open source library built for easy implementation of graph neural networks, and it is now available on Amazon SageMaker. Weird and fascinating stuff!


Documentation and examples:

Deep Java Library

The Deep Java Library is an open source library to develop Deep Learning models in Java. Does this mean that Java developers don’t have to learn Python anymore? ;)

Documentation and examples:

Discussion (0)