The BatchPredictionJob resource lets you run an asynchronous
prediction request. Request batch predictions directly from the model
resource. You don't need to deploy the model to an endpoint. For data types
that support both batch and online predictions you can use batch predictions.
This is useful when you don't require an immediate response and want to process
accumulated data by using a single request.
To make a batch prediction, specify an input source and an output location
for Vertex AI to store predictions results. The inputs and outputs
depend on the model type that you're working with. For example, batch
predictions for the AutoML image model type require an input
JSON Lines
file and the name of a Cloud Storage bucket to store the output.
For more information about batch prediction, see
Get batch predictions.
You can use the ModelBatchPredictOp component to access this resource through Vertex AI Pipelines.
API reference
- For component reference, see the Google Cloud SDK reference for Batch prediction components.
- For Vertex AI API reference, see the
BatchPredictionJobresource page.
Tutorials
Version history and release notes
To learn more about the version history and changes to the Google Cloud Pipeline Components SDK, see the Google Cloud Pipeline Components SDK Release Notes.
Technical support contacts
If you have any questions, reach out to kubeflow-pipelines-components@google.com.