You want the best learning rate and batch size automatically. What do you run?
Why A is the answer
Hyperparameter tuning
Tuning searches for good settings.
In detail
- Hyperparameter tuning searches for good values.
- Set a sensible budget.
Use case
An illustrative scenario, not a real incident.
A team runs hyperparameter tuning over learning rate and batch size, and the best trial wins.
See it in different ways
stateDiagram-v2 [*] --> FitsInMemory FitsInMemory --> DataParallelism: yes, but too slow FitsInMemory --> ModelParallelism: no DataParallelism --> Train ModelParallelism --> Train Train --> Tune: hyperparameter tuning Tune --> [*]
flowchart LR S0["Set ranges"] --> S1["Run trials"] --> S2["Compare"] --> S3["Use the best"]
Key point: Tuning searches for good settings.
Why the other options are wrong
- B. A canary deployment
- A canary tests a deployment.
- C. Model Monitoring
- Monitoring watches production.
- D. A data parallelism test on one device
- That does not search values.