sagemaker-batch-inference-tensorflow-simple-SageMakerModelComplete.scala

import com.amazonaws.services.sagemaker.sparksdk.SageMakerModel
import java.util.UUID

val REGION="us-east-1"
val TFS_VERSION="2.1.0"
val PROCESSOR_TYPE="gpu"
val IMAGE=s"763104351884.dkr.ecr.$REGION.amazonaws.com/tensorflow-inference:$TFS_VERSION-$PROCESSOR_TYPE"
val MODEL_PATH = "s3://bucket/path/model.tar.gz"

val id = UUID.randomUUID().toString

val model = SageMakerModel.fromModelS3Path(
    endpointInstanceType = "ml.p3.2xlarge",
    endpointInitialInstanceCount = 1,
    modelImage = IMAGE,
    modelPath = MODEL_PATH,
    uid = s"saty-nn-test-${id}",
    requestRowSerializer = new TFRequestRowSerializer(featureLength=10),
    responseRowDeserializer = new TFResponseRowDeserializer(),
    modelExecutionRoleARN = "aws role for sagemaker"
)