MLA-C01 Practice Question 89: Training Instance Quota
AWS Certified Machine Learning Engineer – Associate (MLA-C01) · Test 9 · question 9 of 10
A startup's new AWS account will fine-tune a model on eight ml.p4d.24xlarge instances. Every CreateTrainingJob call fails right away with a ResourceLimitExceeded error saying the account-level limit for that instance type for training job usage is 0. The engineer's IAM role has full SageMaker AI permissions, and smaller CPU training jobs run successfully in the account.
What should the engineer do?
Show answer and explanation
Answer:
SageMaker AI enforces per-account quotas, including the number of instances of each type that training jobs can use. The error says the account's limit for this instance type is 0, which is common for large GPU instances in a new account. The Service Quotas console shows your SageMaker AI quotas and lets you request increases for adjustable ones. Once the quota covers eight instances, the job can start (option A).
- B is wrong because the error is about a quota, not authorization. A missing permission returns an access denied error, and the role already has full SageMaker AI permissions. - C is wrong because managed spot training has its own per-instance-type quotas, which are also low or zero for large GPU instances in new accounts. It does not bypass limits. - D is wrong because a limit of 0 is an account setting, not a temporary shortage of capacity. Retrying will fail every time until the quota is raised.
Reference: Supported Regions and Quotas - Amazon SageMaker AI