Azure ML Studio. UI. How to retrive custom model in custom component?

Anna Maslova 5

Hello.

I work in Designer for create Pipeline. I am using my custom components in Python. Also I registered Custom model in block Models.

I designer I connect to blocks my model and components (out-in)

I use args for reading outposts.

like path_model = args.models_output

kmeans=joblib.load(path_model)

But Error model does exist.

How can I read my model correct?

romungi-MSFT 48,906 Reputation points Microsoft Employee Moderator

2023-05-16T08:42:38.8433333+00:00

@Anna Maslova Are you planning to use the SDK v2 to use your model? If Yes, then try to check this sample to set the model path and using it in a job.

If you are using v1 SDK then please check this page for details on deploying the same.
Anna Maslova 5 Reputation points

2023-05-16T15:41:35.4866667+00:00
@romungi-MSFT Thank you very much for your answer.

I was "there" many times. I still didn't get. I am using SDKv1.

I register custom component using my "code", but I'm using UI Designer for creation pipeline.

I am not deploying model yet. I only registered model in Model list and I selected this model in UI Designer as component and link output as Input for my custom component. (type of input =custom_model)

In Model list a have Kmean model- saved with one pickle file.

I cannot read my model correctly. What I forgot?

When I print my input

print(args.input_model)_

I see link "azureml//... datastores/workspaceartifactsstore/paths/LocalUpLoaad/.. nameOfFileMyModel

I try to read like pickle.load(open(args.inputmodel_),'rb)_

I try model = joblib.load(args.inputmodel_)_

Answer - file not found - no such file or directory. What I forgot?

Your answer