Deploying directly from model artifacts

Question

I want to deploy a pretrained neural network as an endpoint at Sagemaker. All my model artifacts are stored in S3.

In the Sagemacer documentation (here) I found this