multi-model-server 1.1.0b20191213

Multi Model Server is a tool for serving neural net models for inference

Multi Model Server (MMS) is a flexible and easy to use tool for serving deep learning models exported from MXNet or the Open Neural Network Exchange (ONNX).

Use the MMS Server CLI, or the pre-configured Docker images, to start a service that sets up HTTP endpoints to handle model inference requests.

Detailed documentation and examples are provided in the docs folder.

Prerequisites

• java 8: Required. MMS use java to serve HTTP requests. You must install java 8 (or later) and make sure java is on available in $PATH environment variable before installing MMS. If you have multiple java installed, you can use $JAVA_HOME environment vairable to control which java to use.

• mxnet: mxnet will not be installed by default with MMS 1.0 any more. You have to install it manually if you use MxNet.

For ubuntu:

sudo apt-get install openjdk-8-jre-headless

For centos

sudo yum install java-1.8.0-openjdk

For Mac:

brew tap caskroom/versions
brew update
brew cask install java8

Install MxNet:

pip install mxnet

MXNet offers MKL pip packages that will be much faster when running on Intel hardware. To install mkl package for CPU:

pip install mxnet-mkl

or for GPU instance:

pip install mxnet-cu92mkl

Installation

pip install multi-model-server

Development

We welcome new contributors of all experience levels. For information on how to install MMS for development, refer to the MMS docs.

Important links

• Official source code repo

• Download releases

• Issue tracker

Source code

You can check the latest source code as follows:

git clone https://github.com/awslabs/multi-model-server.git

Testing

After installation, try out the MMS Quickstart for

• Serving a Model

• Create a Model Archive.

Help and Support

• Documentation

• Forum

Citation

If you use MMS in a publication or project, please cite MMS: https://github.com/awslabs/multi-model-server