components question_answering_model_import

Question Answering Model Import

question_answering_model_import

Overview

Component to import PyTorch / MLFlow model. See docs to learn more.

Version: 0.0.65

View in Studio: https://ml.azure.com/registries/azureml/components/question_answering_model_import/version/0.0.65

Inputs

huggingface id

NOTE The pytorch_model_path or mlflow_model_path takes precedence over huggingface_id

Name	Description	Type	Default	Optional	Enum
huggingface_id	The string can be any valid Hugging Face id from the Hugging Face models webpage. Models from Hugging Face are subject to third party license terms available on the Hugging Face model details page. It is your responsibility to comply with the model's license terms.	string		True

PyTorch model as input This is nothing but huggingface model folder. Here's the link to the example model folder - bert-base-uncased. Additionally, the model folder MUST contain the file finetune_args.json with model_name_or_path as one of the keys of the dictionary

Name	Description	Type	Default	Optional	Enum
pytorch_model_path	Pytorch model asset path	custom_model		True

MLflow model as an input This is also a huggingface model folder expect that the folder structure is slightly different. You could invoke a model import pipeline to convert the standard huggingface model into MLflow format. Please refer to this notebook for steps to do the same NOTE The pytorch_model_path take priority over mlflow_model_path, in case both inputs are passed

Name	Description	Type	Default	Optional	Enum
mlflow_model_path	MLflow model asset path	mlflow_model		True

Output of validation component

Name	Description	Type	Default	Optional	Enum
validation_output	Validation status.	uri_file		True

Outputs

Name	Description	Type
output_dir	Path to output directory which contains the component metadata and the model artifacts folder	uri_folder

Environment

azureml://registries/azureml/environments/acft-hf-nlp-gpu/versions/80