The Model class represents a namespace for organizing artifacts, metadata, and other resources related to a machine learning model.
Signature Model( name: str , license : Optional[ str ] = None , description: Optional[ str ] = None , audience: Optional[ str ] = None , use_cases: Optional[ str ] = None , limitations: Optional[ str ] = None , trade_offs: Optional[ str ] = None , ethics: Optional[ str ] = None , tags: Optional[List[ str ]] = None , version: Optional[Union[ModelStages, int , str ]] = None , save_models_to_registry: bool = True , )
Parameters The license under which the model is created.
The description of the model.
The target audience of the model.
The use cases of the model.
The known limitations of the model.
The tradeoffs of the model.
The ethical implications of the model.
Tags associated with the model.
version
Union[ModelStages, int, str]
The version name, version number, or stage. If skipped, a new version will be created. Supports placeholders like {date} and {time}, plus custom placeholders passed as substitutions.
Whether to save all ModelArtifacts to Model Registry if available in the active stack.
Properties The unique identifier of the model version.
The unique identifier of the model.
The version number of this model version.
The current stage of this model version (e.g., staging, production).
Methods load_artifact def load_artifact ( name : str , version : Optional[ str ] = None ) -> Any
Load an artifact linked to this model version.
Parameters:
name (str): Name of the artifact to loadversion (Optional[str]): Version of the artifact to load
Returns: The loaded artifactget_artifact def get_artifact ( name : str , version : Optional[ str ] = None ) -> Optional[ArtifactVersionResponse]
Get the artifact response linked to this model version.
set_stage def set_stage ( stage : Union[ str , ModelStages], force : bool = False ) -> None
Set this model version to a specific stage.
Parameters:
stage: The target stage (e.g., “staging”, “production”)force: Whether to force archiving of the current version in the target stage
def log_metadata ( metadata : Dict[ str , MetadataType] ) -> None
Log metadata for this model version.
Parameters:
metadata: Dictionary of metadata key-value pairs
Examples Basic Model Usage from zenml import Model, pipeline, step model = Model( name = "iris_classifier" , version = "1.0.0" , description = "Iris species classifier" , tags = [ "classification" , "iris" ] ) @pipeline ( model = model) def training_pipeline (): # Pipeline steps pass
Using Model Stages from zenml import Model, ModelStages # Load production model production_model = Model( name = "iris_classifier" , version = ModelStages. PRODUCTION ) # Load specific version specific_model = Model( name = "iris_classifier" , version = "1.2.0" )
Loading Artifacts from Model from zenml import Model model = Model( name = "iris_classifier" , version = "production" ) # Load a specific artifact training_data = model.load_artifact( "training_data" ) model_weights = model.load_artifact( "model" , version = "3" )
from zenml import step, Model @step ( model = Model( name = "my_model" )) def train_model (): # Training code accuracy = 0.95 from zenml import get_step_context context = get_step_context() context.model.log_metadata({ "accuracy" : accuracy, "framework" : "sklearn" , "dataset_size" : 1000 })
from zenml import Model, ModelStages model = Model( name = "iris_classifier" , version = "1.5.0" ) # Promote to staging model.set_stage(ModelStages. STAGING ) # Promote to production (force replace current production model) model.set_stage(ModelStages. PRODUCTION , force = True )
@pipeline Configure model in pipelines
@step Configure model in steps
log_model_metadata Log model metadata
link_artifact_to_model Link artifacts to models
