Which of the following are use cases for ML Metadata? Select all that apply:

Correct! ML Metadata allows you to compare artifact outputs from multiple experiments. Correct! ML Metadata allows you to track structured data about artifact and execution records groups. This includes data such as the owner and changelist used for a workflow run, which experiments are associated with a pipeline, and which artifacts were a part of each pipeline run. Correct! An orchestrator like Kubeflow, Beam, or Airflow is able to read from the ML Metadata store to generate visualizations of pipeline DAGs by recursing through artifact dependencies to associate components visually.

Which of the following are examples of TFX pipeline metadata-powered functionality? Select all that apply:

Correct! Pipeline nodes enable you to import previously trained model weights to initialize model training in subsequent runs to improve model performance. Correct! ML Metadata powers linkages back between ML experiments and associated data artifacts and component executions. Correct! TFX pipeline caching is enabled by TFX ML Metadata to allow the pipeline to skip over artifact processing when inputs have not changed.

What standard TFX objects can be used for performing advanced TFX metadata operations? Select one.

Correct! Pipeline nodes are special-purpose classes for performing advanced metadata operations such as importing external artifacts into ML Metadata and performing queries to retrieve artifacts based on artifact properties and history in ML Metadata.

Which of the following elements of the ML Metadata data model are stored in an ML Metadata (MLMD) store? Select all that apply:

Correct! Executions are records of component runs and the associated runtime parameters. Correct! Events are records of the relationship between an Artifact and Executions. When an Execution happens, Events record every Artifact that was used by the Execution and every Artifact that was produced. Correct! Contexts are records of groups of Artifacts and Executions in a workflow and their structural properties. For example: projects, pipeline runs, experiments, owners.