Question 180
A Python training job fails sporadically because it assumes a working directory not included in the uploaded code assets. For acceptance, the reviewer must confirm that the remote command resolves every expected file without local machine paths. Which approach is best supported by the observed results?
Explanation
Correct Answer
Azure ML uploads only the declared code and input paths into the remote execution context. A test run should show that the remote command resolves every expected file without local machine paths.
Incorrect Answers
Remote job environments do not inherit local developer package installations.
Command jobs package and execute training scripts on declared Azure ML targets.
Azure ML resolves declared inputs into the command line supplied to the container.