tfx.components.Evaluator

A TFX component to evaluate models trained by a TFX Trainer component.

Inherits From: BaseComponent

tfx.components.Evaluator(
    examples: tfx.types.Channel = None,
    model: tfx.types.Channel = None,
    baseline_model: Optional[tfx.types.Channel] = None,
    feature_slicing_spec: Optional[Union[evaluator_pb2.FeatureSlicingSpec, Dict[Text, Any]]] = None,
    fairness_indicator_thresholds: Optional[List[Union[float, tfx.orchestration.data_types.RuntimeParameter]]] = None,
    example_splits: Optional[List[Text]] = None,
    output: Optional[tfx.types.Channel] = None,
    model_exports: Optional[tfx.types.Channel] = None,
    instance_name: Optional[Text] = None,
    eval_config: Optional[tfma.EvalConfig] = None,
    blessing: Optional[tfx.types.Channel] = None,
    schema: Optional[tfx.types.Channel] = None,
    module_file: Optional[Text] = None
)

Used in the notebooks

Used in the tutorials

The Evaluator component performs model evaluations in the TFX pipeline and the resultant metrics can be viewed in a Jupyter notebook. It uses the input examples generated from the ExampleGen component to evaluate the models.

Specifically, it can provide:

  • metrics computed on entire training and eval dataset
  • tracking metrics over time
  • model quality performance on different feature slices

Exporting the EvalSavedModel in Trainer

In order to setup Evaluator in a TFX pipeline, an EvalSavedModel needs to be exported during training, which is a special SavedModel containing annotations for the metrics, features, labels, and so on in your model. Evaluator uses this EvalSavedModel to compute metrics.

As part of this, the Trainer component creates eval_input_receiver_fn, analogous to the serving_input_receiver_fn, which will extract the features and labels from the input data. As with serving_input_receiver_fn, there are utility functions to help with this.

Please see https://www.tensorflow.org/tfx/model_analysis for more details.

Example

  # Uses TFMA to compute a evaluation statistics over features of a model.
  model_analyzer = Evaluator(
      examples=example_gen.outputs['examples'],
      model=trainer.outputs['model'],
      eval_config=tfma.EvalConfig(...))

examples A Channel of type standard_artifacts.Examples, usually produced by an ExampleGen component. required
model A Channel of type standard_artifacts.Model, usually produced by a Trainer component.
baseline_model An optional channel of type 'standard_artifacts.Model' as the baseline model for model diff and model validation purpose.
feature_slicing_spec Deprecated, please use eval_config instead. Only support estimator. evaluator_pb2.FeatureSlicingSpec instance that describes how Evaluator should slice the data. If any field is provided as a RuntimeParameter, feature_slicing_spec should be constructed as a dict with the same field names as FeatureSlicingSpec proto message.
fairness_indicator_thresholds Optional list of float (or RuntimeParameter) threshold values for use with TFMA fairness indicators. Experimental functionality: this interface and functionality may change at any time. to additional documentation for TFMA fairness indicators here.
example_splits Names of splits on which the metrics are computed. Default behavior (when example_splits is set to None or Empty) is using the 'eval' split.
output Channel of ModelEvalPath to store the evaluation results.
model_exports Backwards compatibility alias for the model argument.
instance_name Optional name assigned to this specific instance of Evaluator. Required only if multiple Evaluator components are declared in the same pipeline. Either model_exports or model must be present in the input arguments.
eval_config Instance of tfma.EvalConfig containg configuration settings for running the evaluation. This config has options for both estimator and Keras.
blessing Output channel of 'ModelBlessingPath' that contains the blessing result.
schema A Schema channel to use for TFXIO.
module_file A path to python module file containing UDFs for Evaluator customization. The module_file can implement following functions at its top level. def custom_eval_shared_model( eval_saved_model_path, model_name, eval_config, **kwargs, ) -> tfma.EvalSharedModel: def custom_extractors( eval_shared_model, eval_config, tensor_adapter_config, ) -> List[tfma.extractors.Extractor]:

component_id DEPRECATED FUNCTION
component_type DEPRECATED FUNCTION
downstream_nodes

exec_properties

id Node id, unique across all TFX nodes in a pipeline.

If instance name is available, node_id will be: . otherwise, node_id will be:

inputs

outputs

type

upstream_nodes

Child Classes

class DRIVER_CLASS

class SPEC_CLASS

Methods

add_downstream_node

View source

add_downstream_node(
    downstream_node
)

Experimental: Add another component that must run after this one.

This method enables task-based dependencies by enforcing execution order for synchronous pipelines on supported platforms. Currently, the supported platforms are Airflow, Beam, and Kubeflow Pipelines.

Note that this API call should be considered experimental, and may not work with asynchronous pipelines, sub-pipelines and pipelines with conditional nodes. We also recommend relying on data for capturing dependencies where possible to ensure data lineage is fully captured within MLMD.

It is symmetric with add_upstream_node.

Args
downstream_node a component that must run after this node.

add_upstream_node

View source

add_upstream_node(
    upstream_node
)

Experimental: Add another component that must run before this one.

This method enables task-based dependencies by enforcing execution order for synchronous pipelines on supported platforms. Currently, the supported platforms are Airflow, Beam, and Kubeflow Pipelines.

Note that this API call should be considered experimental, and may not work with asynchronous pipelines, sub-pipelines and pipelines with conditional nodes. We also recommend relying on data for capturing dependencies where possible to ensure data lineage is fully captured within MLMD.

It is symmetric with add_downstream_node.

Args
upstream_node a component that must run before this node.

from_json_dict

View source

@classmethod
from_json_dict(
    dict_data: Dict[Text, Any]
) -> Any

Convert from dictionary data to an object.

get_id

View source

@classmethod
get_id(
    instance_name: Optional[Text] = None
)

Gets the id of a node.

This can be used during pipeline authoring time. For example: from tfx.components import Trainer

resolver = ResolverNode(..., model=Channel( type=Model, producer_component_id=Trainer.get_id('my_trainer')))

Args
instance_name (Optional) instance name of a node. If given, the instance name will be taken into consideration when generating the id.

Returns
an id for the node.

to_json_dict

View source

to_json_dict() -> Dict[Text, Any]

Convert from an object to a JSON serializable dictionary.

Class Variables

  • EXECUTOR_SPEC