Google Cloud Pipeline Components are released in two stages:
Generally available (v1.0 and later)
Preview
For more information, see the launch stage descriptions.
Generally available (1.0 and later) components
Generally available (GA) Google Cloud Pipeline Components are available in the v1
namespace of the Google Cloud Pipeline Components SDK. These components have a stable interface and are ready to be used in a production environment.
AutoML components
AutoML Forecasting components
The Google Cloud Pipeline Components SDK includes the following operators related to AutoML forecasting:
AutoML Tabular components
The Google Cloud Pipeline Components SDK includes the following operators related to AutoML tabular:
-
CvTrainerOp
-
EnsembleOp
-
FinalizerOp
-
InfraValidatorOp
-
SplitMaterializedDataOp
-
Stage1TunerOp
-
StatsAndExampleGenOp
-
TrainingConfiguratorAndValidatorOp
-
TransformOp
AutoML Training Job components
The Google Cloud Pipeline Components SDK includes the following operators related to AutoML training jobs:
-
AutoMLForecastingTrainingJobRunOp
-
AutoMLImageTrainingJobRunOp
-
AutoMLTabularTrainingJobRunOp
-
AutoMLTextTrainingJobRunOp
-
AutoMLVideoTrainingJobRunOp
Batch Prediction components
The Google Cloud Pipeline Components SDK includes the following operators related to Batch Prediction:
BigQuery ML components
The Google Cloud Pipeline Components SDK includes the following operators for creating and executing machine learning models using BigQuery:
-
BigqueryCreateModelJobOp
-
BigqueryDetectAnomaliesModelJobOp
-
BigqueryDropModelJobOp
-
BigqueryEvaluateModelJobOp
-
BigqueryExplainForecastModelJobOp
-
BigqueryExplainPredictModelJobOp
-
BigqueryExportModelJobOp
-
BigqueryForecastModelJobOp
-
BigqueryMLAdvancedWeightsJobOp
-
BigqueryMLArimaCoefficientsJobOp
-
BigqueryMLArimaEvaluateJobOp
-
BigqueryMLCentroidsJobOp
-
BigqueryMLConfusionMatrixJobOp
-
BigqueryMLFeatureImportanceJobOp
-
BigqueryMLFeatureInfoJobOp
-
BigqueryMLGlobalExplainJobOp
-
BigqueryMLPrincipalComponentInfoJobOp
-
BigqueryMLPrincipalComponentsJobOp
-
BigqueryMLRecommendJobOp
-
BigqueryMLReconstructionLossJobOp
-
BigqueryMLRocCurveJobOp
-
BigqueryMLTrainingInfoJobOp
-
BigqueryMLTrialInfoJobOp
-
BigqueryMLWeightsJobOp
-
BigqueryPredictModelJobOp
-
BigqueryQueryJobOp
Custom Job components
The Google Cloud Pipeline Components SDK includes the following operators for running KFP components as Vertex AI Custom Training jobs:
Dataflow components
The Google Cloud Pipeline Components SDK includes the following operators related to Dataflow:
Dataproc Serverless components
The Google Cloud Pipeline Components SDK includes the following operators related to Dataproc Serverless:
Dataset components
The Google Cloud Pipeline Components SDK includes the following operators for AutoML dataset
resource management:
-
GetVertexDatasetOp
-
ImageDatasetCreateOp
-
ImageDatasetExportDataOp
-
ImageDatasetImportDataOp
-
TabularDatasetCreateOp
-
TabularDatasetExportDataOp
-
TextDatasetCreateOp
-
TextDatasetExportDataOp
-
TextDatasetImportDataOp
-
TimeSeriesDatasetCreateOp
-
TimeSeriesDatasetExportDataOp
-
VideoDatasetCreateOp
-
VideoDatasetExportDataOp
-
VideoDatasetImportDataOp
Endpoint components
The Google Cloud Pipeline Components SDK includes the following operators related to model serving endpoints:
Forecasting components
The Google Cloud Pipeline Components SDK includes the following operators related to forecasting:
Hyperparameter Tuning components
The Google Cloud Pipeline Components SDK includes the following operators related to hyperparameter tuning jobs:
Model components
The Google Cloud Pipeline Components SDK includes the following operators for managing models through the Vertex AI Model Registry:
Model evaluation components
The Google Cloud Pipeline Components SDK includes the following operators related to model evaluation:
Email notification components
The Google Cloud Pipeline Components SDK includes the following operators for managing email notifications:
Components related to wait on resources
The Google Cloud Pipeline Components SDK includes the following operator related to waiting on resource creation:
Preview components
Google Cloud Pipeline Components in the Preview stage are available in the preview
namespace of the Google Cloud Pipeline Components SDK.
AutoML components
AutoML Forecasting components
The Google Cloud Pipeline Components SDK includes the following preview operators related to AutoML forecasting:
AutoML Tabular components
The Google Cloud Pipeline Components SDK includes the following preview operators related to AutoML tabular:
-
AutoFeatureEngineeringOp
-
DistillationStageFeatureTransformEngineOp
-
FeatureSelectionOp
-
FeatureTransformEngineOp
-
TabNetHyperparameterTuningJobOp
-
TabNetTrainerOp
-
WideAndDeepHyperparameterTuningJobOp
-
WideAndDeepTrainerOp
-
XGBoostHyperparameterTuningJobOp
-
XGBoostTrainerOp
Custom Job components
The Google Cloud Pipeline Components SDK includes the following preview operators for running KFP components as Vertex AI Custom Training jobs:
Dataflow components
The Google Cloud Pipeline Components SDK includes the following preview operators related to Dataflow:
Large-language model (LLM) components
The Google Cloud Pipeline Components SDK includes the following preview operators related to LLMs:
Model evaluation components
The Google Cloud Pipeline Components SDK includes the following preview operators related to model evaluation:
What's next
See the following reference pages for specific Google Cloud Pipeline Components information: