question

AntaraDas-4298 avatar image
1 Vote"
AntaraDas-4298 asked ramr-msft edited

Where are registered models saved in storage containers?

Have a small doubt. i could run a pipeline successfully and also register the model. I can locate the model on the AzureML UI .
Model.get_model_path() shows that it is located in azureml-models/model-name/..

But was wondering where exactly they are stored in storage account? Becasue i dont find and container azureml-model listed.

Any lead on this will be helpful

azure-machine-learning
· 4
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

@AntaraDas-4298 Thanks for the question. Are you using the Designer for model?

Here is the document to Download the entry script file and conda dependencies file.

The easy way is to use register_model from the run object like below

model = run.register_model(model_name='sklearn_mnist', model_path='outputs/sklearn_mnist_model.pkl')
print(model.name, model.id, model.version, sep='\t')
From https://docs.microsoft.com/en-us/azure/machine-learning/how-to-deploy-and-where#registermodel


0 Votes 0 ·

thanks for the response,

I am not using the designer. Registered the model with run.register_model()
What i am interested is , is where exactly on my storage account is the model saved.
Running Model.get_path() shows in azureml-models, but in storage account there is no container named azureml-models listed

0 Votes 0 ·

Thanks for the details. Are you getting any error. Can you please share output of the following model_root = Model.get_model_path('outputs/model_loc'). We have forwarded to the product team for the storage container.

The AZUREML_MODEL_DIR environment variable indicates a directory containing the model file you registered.

model_filename = 'sklearn_regression_model.pkl'
model_path = os.path.join(os.environ['AZUREML_MODEL_DIR'], model_filename)

model = joblib.load(model_path)


0 Votes 0 ·

HI,
output for Model.get_model_path is : 'azureml-models\\model.pkl\\5\\model.pkl'
From this we understand the path , but is azureml a container in the storage account? We are more interested in knowing if we can access it to perform LCM etc


0 Votes 0 ·

1 Answer

ramr-msft avatar image
0 Votes"
ramr-msft answered ramr-msft edited

@AntaraDas-4298 Thanks for the details. there is not an azureml-models container, run.register_model() copies the model files to the azureml container.

5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.