Kvíz spojený s průvodcem · Střední úroveň
Správa konfigurace pomocí Hydra Quiz
Skládejte konfigurace Hydra ze skupin, aplikujte přepsání a rozmítání a zkontrolujte vyřešená nastavení, ověření a hranice zdrojů.
Otázka 1 z 8
How does Hydra build an application's runtime configuration?
Náhled každé otázky8 otázky
Odpovědi a vysvětlení se zobrazují, jak budete v kvízu.
- How does Hydra build an application's runtime configuration?
- It hard-codes each run’s settings into separate scripts
- It loads one flat config and disallows command-line overrides
- It composes selected config pieces and command-line overrides
- It infers all settings from a model checkpoint
- What does a command-line override commonly change?
- Edits the source file permanently before execution
- A selected configuration value for a run
- Creates a new config group automatically
- Rewrites the defaults list on disk
- Which description distinguishes a Hydra config group from an ordinary scalar override?
- It replaces one existing scalar value
- It selects a different output directory
- It selects among related component configurations
- It logs a training metric
- Why inspect and save the fully composed config?
- It shows the actual settings used for that run
- It guarantees the model generalizes
- It eliminates data versioning
- It replaces source control
- What should happen before launching a costly sweep?
- Use the test set for every parameter combination
- Include secret tokens in each command-line override
- Estimate the run count and bound resources before dispatch
- Assume the launcher caps the sweep at a safe size
- How do ordinary overrides differ from adding a new config key?
- An ordinary override updates a value, while addition syntax introduces a missing key
- Ordinary overrides add missing keys while addition syntax removes them
- Both forms only select config groups and cannot change scalar values
- Both forms edit the source YAML file before execution
- What should be kept out of committed configuration files?
- Model architecture names
- Non-sensitive learning rates
- Dataset schema fields
- Passwords, access tokens and other secrets
- When is a configuration framework most useful?
- When an application has repeatable components and many meaningful variants
- For every one-line script
- When no parameters need to change
- Only after deployment is complete