AzureML Model Register-CodePudding

I was trying to register a model using the Run Class like this:

model = run.register_model(
    model_name=model_name,
    model_path=model_path)

Errors with message: Could not locate the provided model_path ... in the set of files uploaded to the run...

CodePudding user response：

The only way I found to fix the issue was to use the Model Class instead:

        model = Model.register(
            workspace=ws,
            model_name=model_name,
            model_path=model_path,
            model_framework=Model.Framework.SCIKITLEARN,
            model_framework_version=sklearn.__version__,
            description='Model Deescription',
            tags={'Name' : 'ModelName', 'Type' : 'Production'},
            model_framework=Model.Framework.SCIKITLEARN,
            model_framework_version='1.0'
            )

CodePudding user response：

I think the problem is that you rely on AML background process to automatically upload content under ./outputs to AML workspace. But when the upload is not complete and we immediately call run.register_model which takes the content from AML workspace then the error will happen. To avoid that situation, you can do it like this:

Persist model (joblib.dump) to a custom folder other than outputs
Manually run upload_file to upload the model AML workspace. Name the destination same name with your model file.
Then run run.register_model.