Custom entrypoints

For some images, Deep Learning Containers use a custom entrypoint script. If you want to use your own entrypoint, you can override the entrypoint as follows.

Modify the entryPoint parameter in the JSON file that includes your task definition. Include the file path to your custom entry point script. An example is shown here.

"entryPoint":[
           "sh",
           "-c",
           "/usr/local/bin/multi-model-server --start --foreground --mms-config /home/model-server/config.properties --models densenet=https://dlc-samples.s3.amazonaws.com/pytorch/multi-model-server/densenet/densenet.mar"],