Canal-camera barge-count images take fourteen hours on one instance. The code already uses a supported framework and the set is large enough to shard. Which official method cuts that wall-clock time?
Select an answer to reveal the explanation.
Short Explanation
Think of barge-count images taking fourteen hours on one instance, with a supported framework and a set large enough to shard. SageMaker distributed training, data parallel. Adding epochs on one host does not shard the set.
Full Explanation
SageMaker distributed training with data parallel across instances is the official method to cut wall-clock time when the framework is supported and the set is large enough to shard. Adding epochs on one host does not shard. Translate and Fraud Detector are not distributed trainers.