A foundation model fine-tuning job is extremely expensive, and retraining the same model separately in staging and production would add major cost and delay. In this case, when might promoting a model version be more appropriate than promoting code alone?
Answer & explanation
Correct answer: B
WHY B: Databricks notes that although code promotion is preferred by default, promoting model versions can be appropriate when retraining across environments is too expensive or impractical.
WHY NOT A: Unity Catalog can still govern promoted models; model promotion is not reserved for models outside a governed registry.
WHY NOT C: Missing tests are not a sound reason to abandon code promotion discipline.
WHY NOT D: Lack of a signature is not the rationale for promoting a model version.
WHY NOT E: Alias management can still be useful after model promotion.