You are developing a process for training and running your custom model in production. You need to be able to show lineage for your model and predictions. What should you do?

Vertex AI managed datasets track lineage related to the data itself, but do not provide comprehensive lineage for the model training process or the predictions generated from it.

While Vertex AI Pipelines provide lineage tracking for components within a larger workflow, Vertex AI Experiments is the direct service for tracking individual model training runs and their metadata, forming the core lineage for the model itself.

Uploading a dataset to BigQuery is primarily a storage solution for data and does not inherently provide MLOps lineage tracking for the model training process or subsequent predictions.

Vertex AI Experiments is specifically designed to track and manage the metadata for machine learning development, including model training runs, parameters, metrics, and associated artifacts like datasets and models. By training your model using Vertex AI Experiments, you automatically capture the necessary lineage information for both the model and how it was produced, which can then be linked to its predictions.