@Olivier Cruchant, then how will I get model_artifacts inside the docker container if I implement the transform functionality using SageMaker training job? Because I think SageMaker automatically download model_artifacts from S3 bucket to /opt/ml/model path inside the docker container for a transform job.