引導式測驗 · 中等 等級
機器學習測驗的資料沿襲
追蹤記錄的數據和工作關係,同時將譜系事件與完整的歷史和再現性區分開來。
問題 1 的 8
What core entities does OpenLineage use to describe data processing?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- What core entities does OpenLineage use to describe data processing?
- Models, prompts, and users only
- Jobs, Runs, and Datasets
- Tables and dashboards only
- Checksums and backups only
- What does a RunEvent primarily describe?
- A guarantee that a dataset is correct
- A full copy of every source record
- A contract signed by all consumers
- An observed state of a job run and associated input/output datasets
- When can a lineage graph be incomplete?
- When systems do not emit events or required facets are not enabled
- Only when a job has no output
- When a dataset has a schema
- Never, because the standard fills gaps
- Is OpenLineage’s Python source-code-location facet enabled by default?
- Yes, in every run automatically
- Only for Delta tables
- No; the documented default is disabled unless explicitly enabled
- It is required for all events
- What does a recorded lineage edge prove?
- That the transformation was correct
- That the event producer reported a relationship between jobs or datasets
- That the input data were unbiased
- That the data will be retained forever
- What additional records help reproduce a historical ML run?
- Dataset snapshot/version, code revision, parameters, and environment
- Only the model display name
- A current dataset without a version
- Only a lineage graph image
- How does declared job lineage differ from observed run lineage?
- They are always identical
- Observed lineage is a legal contract
- Declared lineage contains data values
- Declared lineage describes expected job relationships; run lineage records what was observed for an execution
- What should a team verify if lineage names an old dataset version?
- That a graph still renders
- That current data have the same row count
- That the version is retained and can actually be reloaded
- Nothing, because lineage stores the data