The container images are published to Docker Hub and are pre-configured and optimized for both CPU hosts (EC2 C5.2xlarge instance) and multi-GPU hosts (EC2 P3.8xlarge instance). MMS also provides tooling to package MXNet and ONNX neural network models into a single “model archive,” which includes all of the artifacts needed to serve the model.
To learn more about MMS, visit the model zoo and documentation.
Advertisements
