You have an existing TensorFlow training script (train.py) that reads static training data stored as TFRecord files. You need a scalable storage solution for training on Amazon SageMaker with the least amount of code changes. Which approach minimizes development effort?

The correct answer minimizes development effort because SageMaker's script mode allows you to run existing TensorFlow scripts with minimal changes. By uploading the TFRecord files to Amazon S3, SageMaker can directly access this data during training without requiring any reformatting or modifications to your train.py script. SageMaker automatically handles the data transfer from S3 to the training instance. The other options are incorrect because: Pointing to a local filesystem path is not scalable and doesn't leverage SageMaker's distributed capabilities. Converting TFRecords to protobuf or reformatting data into a SageMaker-supported schema introduces unnecessary development effort and code changes, which goes against the requirement of minimizing effort.