Skip to content

components evaluate_model

github-actions[bot] edited this page Dec 20, 2023 · 19 revisions

Evaluate Model

evaluate_model

Overview

Evaluate MLFlow models for supported task types.

Version: 0.0.19

Tags

type : evaluation sub_type : evaluate_model

View in Studio: https://ml.azure.com/registries/azureml/components/evaluate_model/version/0.0.19

Inputs

Name Description Type Default Optional Enum
task Task type string tabular-classification False ['tabular-classification', 'tabular-classification-multilabel', 'tabular-regression', 'tabular-forecasting', 'text-classification', 'text-classification-multilabel', 'text-named-entity-recognition', 'text-summarization', 'question-answering', 'text-translation', 'fill-mask', 'text-generation', 'image-classification', 'image-classification-multilabel', 'chat-completion', 'image-object-detection', 'image-instance-segmentation']
test_data Test Data as JSON Lines URI_FILE uri_folder False
evaluation_config Additional parameters required for evaluation. uri_file True
test_data_label_column_name Column name of target values string False
test_data_input_column_names Comma separated values of feature column names string True
mlflow_model Mlflow Model - registered model or output of a job with type mlflow_model in a pipeline mlflow_model False
device string auto False ['auto', 'cpu', 'gpu']
batch_size integer True
evaluation_config_params JSON Serialized string of evaluation_config string True

Outputs

Name Description Type
evaluation_result uri_folder

Environment

azureml://registries/azureml/environments/model-evaluation/versions/16

Clone this wiki locally