A production computer-vision team runs Ray Tune with 64 concurrent trials on a Databricks GPU cluster. GPU utilization spikes, several trials fail with out-of-memory errors, and the best completed result is unstable. What is the best first adjustment?
Your Answer
Explanation
FREE AI FEEDBACK
Guest checks show correctness and the answer key. Log in to save history and unlock evaluator notes.
i
Review the explanation and try similar questions to strengthen your understanding.
Issue reporting
Report an issue
Clara
How can I help?
Search Databricks AI Certs Reviewer
Search across lessons, syllabus topics, provider capabilities, and certification questions.