AI-300 Practice Question 87: Developer Tier Evaluation
Microsoft Machine Learning Operations Engineer Associate (AI-300) · Test 9 · question 7 of 10
A team fine-tuned five candidate models and wants reviewers to test each one for a few hours to pick a winner. The candidates will not serve production traffic, and the team wants to avoid an hourly hosting fee while they are compared. Which deployment type should they use?
Show answer and explanation
Answer:
Developer deployments of fine-tuned models offer an experience similar to Global Standard, without an hourly hosting fee and without an availability SLA. They are designed for evaluating model candidates, not for production, and each deployment lasts 24 hours, which fits a few hours of review (option D).
- A is wrong because a Standard fine-tuned deployment is billed hourly while it exists, which is the cost the team wants to avoid. - B is wrong because provisioned throughput reserves capacity and is billed for the reservation, which is far more than an evaluation needs. - C is wrong because batch is asynchronous with a target of 24 hours, so reviewers could not interactively test the outputs.
Reference: Deploy a fine-tuned model