The remove_tags() function allows you to remove tags from various ZenML resources including pipelines, runs, artifacts, and more.
Signature def remove_tags ( tags : Union[ str , List[ str ]], # Pipeline parameters pipeline : Optional[Union[ UUID , str ]] = None , # Run parameters run : Optional[Union[ UUID , str ]] = None , # Run template parameters run_template : Optional[Union[ UUID , str ]] = None , # Snapshot parameters snapshot : Optional[Union[ UUID , str ]] = None , # Deployment parameters deployment : Optional[Union[ UUID , str ]] = None , # Artifact parameters artifact : Optional[Union[ UUID , str ]] = None , # Artifact version parameters artifact_version_id : Optional[ UUID ] = None , artifact_name : Optional[ str ] = None , artifact_version : Optional[ str ] = None , infer_artifact : Optional[ bool ] = None , ) -> None
Parameters tags
Union[str, List[str]]
required
The tag name(s) to remove. Can be a single string or a list of strings.
The ID or name of the pipeline.
The ID, name, or prefix of the pipeline run.
The ID or name of the run template.
The ID or name of the deployment.
The ID or name of the artifact.
The ID of a specific artifact version.
The name of the artifact (used with artifact_version or infer_artifact).
The version of the artifact.
Whether to infer the artifact from the step context.
Examples from zenml import step, remove_tags @step def cleanup_tags () -> None : # Remove experimental tags from the run remove_tags([ "experimental" , "draft" ])
from zenml import step, remove_tags import pandas as pd @step def update_artifact_tags () -> pd.DataFrame: df = pd.DataFrame({ "data" : range ( 100 )}) # Remove draft tag, mark as finalized remove_tags( tags = [ "draft" , "unreviewed" ], infer_artifact = True ) return df
from typing import Tuple, Annotated from zenml import step, remove_tags import pandas as pd @step def update_datasets () -> Tuple[ Annotated[pd.DataFrame, "train" ], Annotated[pd.DataFrame, "test" ], ]: train = pd.DataFrame({ "x" : range ( 800 )}) test = pd.DataFrame({ "x" : range ( 200 )}) # Remove old version tags remove_tags( tags = [ "v1" , "deprecated" ], artifact_name = "train" , infer_artifact = True ) return train, test
from zenml import remove_tags # Remove tags from a specific artifact version remove_tags( tags = [ "testing" , "unvalidated" ], artifact_name = "training_data" , artifact_version = "v5" )
from zenml import remove_tags # Remove tags from a pipeline remove_tags( tags = [ "experimental" ], pipeline = "training_pipeline" )
from zenml import remove_tags # Remove tags from a completed run remove_tags( tags = [ "in-progress" , "running" ], run = "training_pipeline-2024_01_15" )
Conditional Tag Removal from zenml import step, remove_tags, add_tags import pandas as pd @step def validate_data () -> pd.DataFrame: df = pd.read_csv( "data.csv" ) # Check data quality quality_score = check_quality(df) if quality_score > 0.9 : # Remove pending tag, add validated tag remove_tags([ "pending-validation" ]) add_tags([ "validated" , "high-quality" ]) else : # Remove any approval tags remove_tags([ "validated" , "approved" ]) add_tags([ "needs-review" ]) return df
from zenml import remove_tags # Remove all experimental tags from an artifact remove_tags( tags = [ "experimental" , "draft" , "test" , "debug" , "temporary" ], artifact_name = "model_weights" , artifact_version = "15" )
Remove Single Tag from zenml import step, remove_tags @step def finalize_run () -> None : # Remove the draft tag remove_tags( "draft" )
from zenml import pipeline, step, remove_tags @step def process_data () -> dict : return { "status" : "complete" } @step def cleanup () -> None : # Remove temporary tags from the run remove_tags([ "in-progress" , "processing" , "temporary" ]) @pipeline def data_pipeline (): result = process_data() cleanup()
from zenml import remove_tags # Remove staging tag after promoting to production remove_tags( tags = [ "staging" , "pre-production" ], deployment = "ml-model-deployment" )
Safe Tag Removal from zenml import remove_tags try : remove_tags( tags = [ "old-tag" ], artifact_name = "my_artifact" , artifact_version = "3" ) print ( "Tags removed successfully" ) except Exception as e: print ( f "Tag removal failed: { e } " ) # Tag might not exist, which is fine
Use Cases
Clean up experimental tags after finalizing a model or datasetUpdate status tags as resources move through workflows (e.g., remove “draft”, add “approved”)Remove outdated labels when reprocessing or updating artifactsManage deployment tags when promoting or demoting modelsFix tagging mistakes by removing incorrect tagsImplement tag lifecycle by removing tags when conditions changeMaintain tag hygiene by cleaning up temporary or debugging tags
Important Notes
When called inside a step without parameters, tags are removed from the pipeline run
If a tag doesn’t exist on the resource, the operation completes silently (no error)
Tag names are case-sensitive
Removing a tag from one resource doesn’t affect the same tag on other resources
The tag definition persists in ZenML even if removed from all resources
from zenml import step, add_tags, remove_tags @step def promote_model () -> None : # Update tags to reflect new status remove_tags([ "staging" , "testing" ]) add_tags([ "production" , "active" ])
add_tags Add tags to resources
Tag Learn about Tag objects
@pipeline Configure tags in pipelines
