AWS Contributes to Milestone 1.0 Release and Adds Model Serving Capability for Apache MXNet

Post by Dr. Matt Wood

Today AWS announced contributions to the milestone 1.0 release of the Apache MXNet deep learning engine including the introduction of a new model-serving capability for MXNet. The new capabilities in MXNet provide the following benefits to users:

1) MXNet is easier to use: The model server for MXNet is a new capability introduced by AWS, and it packages, runs, and serves deep learning models in seconds with just a few lines of code, making them accessible over the internet via an API endpoint and thus easy to integrate into applications. The 1.0 release also includes an advanced indexing capability that enables users to perform matrix operations in a more intuitive manner.

• Model Serving enables set up of an API endpoint for prediction: It saves developers time and effort by condensing the task of setting up an API endpoint for running and integrating prediction functionality into an application to just a few lines of code. It bridges the barrier between Python-based deep learning frameworks and production systems through a Docker container-based deployment model.
• Advanced indexing for array operations in MXNet: It is now more intuitive for developers to leverage the powerful array operations in MXNet. They can use the advanced indexing capability by leveraging existing knowledge of NumPy/SciPy arrays. For example, it supports MXNet NDArray and Numpy ndarray as index, e.g. (a[mx.nd.array([1,2], dtype = ‘int32’]).

2) MXNet is faster: The 1.0 release includes implementation of cutting-edge features that optimize the performance of training and inference. Gradient compression enables users to train models up to five times faster by reducing communication bandwidth between compute nodes without loss in convergence rate or accuracy. For speech recognition acoustic modeling like the Alexa voice, this feature can reduce network bandwidth by up to three orders of magnitude during training. With the support of NVIDIA Collective Communication Library (NCCL), users can train a model 20% faster on multi-GPU systems.

• Optimize network bandwidth with gradient compression: In distributed training, each machine must communicate frequently with others to update the weight-vectors and thereby collectively build a single model, leading to high network traffic. Gradient compression algorithm enables users to train models up to five times faster by compressing the model changes communicated by each instance.
• Optimize the training performance by taking advantage of NCCL: NCCL implements multi-GPU and multi-node collective communication primitives that are performance optimized for NVIDIA GPUs. NCCL provides communication routines that are optimized to achieve high bandwidth over interconnection between multi-GPUs. MXNet supports NCCL to train models about 20% faster on multi-GPU systems.

3) MXNet provides easy interoperability: MXNet now includes a tool for converting neural network code written with the Caffe framework to MXNet code, making it easier for users to take advantage of MXNet’s scalability and performance.

• Migrate Caffe models to MXNet: It is now possible to easily migrate Caffe code to MXNet, using the new source code translation tool for converting Caffe code to MXNet code.

MXNet has helped developers and researchers make progress with everything from language translation to autonomous vehicles and behavioral biometric security. We are excited to see the broad base of users that are building production artificial intelligence applications powered by neural network models developed and trained with MXNet. For example, the autonomous driving company TuSimple recently piloted a self-driving truck on a 200-mile journey from Yuma, Arizona to San Diego, California using MXNet. This release also includes a full-featured and performance optimized version of the Gluon programming interface. The ease-of-use associated with it combined with the extensive set of tutorials has led significant adoption among developers new to deep learning. The flexibility of the interface has driven interest within the research community, especially in the natural language processing domain.

Getting started with MXNet
Getting started with MXNet is simple. To learn more about the Gluon interface and deep learning, you can reference this comprehensive set of tutorials, which covers everything from an introduction to deep learning to how to implement cutting-edge neural network models. If you’re a contributor to a machine learning framework, check out the interface specs on GitHub.

To get started with the Model Server for Apache MXNet, install the library with the following command:

$ pip install mxnet-model-server

The Model Server library has a Model Zoo with 10 pre-trained deep learning models, including the SqueezeNet 1.1 object classification model. You can start serving the SqueezeNet model with just the following command:

$ mxnet-model-server \
  --models squeezenet=https://s3.amazonaws.com/model-server/models/squeezenet_v1.1/squeezenet_v1.1.model \
  --service dms/model_service/mxnet_vision_service.py

Learn more about the Model Server and view the source code, reference examples, and tutorials here: https://github.com/awslabs/mxnet-model-server/

-Dr. Matt Wood

Modified 10/24/2020 – In an effort to ensure a great experience, expired links in this post have been updated or removed from the original post.