Bring Your Own Container With Amazon SageMaker

In the past I've talked about how to train a custom TensorFlow model on Amazon SageMaker. This is made easy because SageMaker manages containers for popular frameworks such as TensorFlow, PyTorch, HuggingFace, and more. This allows for developers to use these provided containers and focus on providing a Script for training and/or inference in a method known as Script Mode. Now let's say the framework you're working with is not supported by SageMaker's existing Deep Learning containers. This is a real possibility as there's many existing ML frameworks that are being launched with every coming week.