Azure Machine learning and attached compute instance

Petz Tamás 6 Reputation points
2021-04-08T13:28:35.36+00:00

Hello,
My initial one-month free credits have expired and I registered my credit card. In the first year there are several free resources which can be used free of charge for a limited capacity.
I have created a Standard B2s (2 vcpus, 4 GiB memory) VM and a Machine Learning resource in the same resource group.
I have successfully attached this VM to the ML computing resources as you can see on the photo:

85778-ml-attach.png

However whether in the notebook part if I create a new notebook or in the pipeline part if I create a new pipeline I can't select this attached compute resource to be the compute target.

85804-target.png

Can you please help how to manage this situation?
I am in the learning phase of Azure ML and I would like to use the free resource until I have enough experience to create the production pipeline.
Thanks in advance

Azure Machine Learning
Azure Machine Learning
An Azure machine learning service for building and deploying models.
2,715 questions
Azure Virtual Machines
Azure Virtual Machines
An Azure service that is used to provision Windows and Linux virtual machines.
7,558 questions
0 comments

1 answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. romungi-MSFT 43,691 Reputation points Microsoft Employee
    2021-04-09T08:24:08.453+00:00

    @Petz Tamás From the screen shot above you seem to have created an attached compute which is primarily compute that is not managed by Azure ML. This type of compute needs additional steps to be used by your experiments in the workspace.

    In your scenario to use the notebook and designer experiments you need to create a compute instance from the compute instance tab for the notebook. This type of compute allows a notebook to pickup a kernel and run Jupyter notebooks on cloud with some extended capabilities.

    For designer experiments with compute and inference scenarios the compute clusters and inference clusters are used as part of the experiment or the pipeline that is deployed from these experiments. All these types of compute are managed from Azure ML workspace so you have the ability to use them for different experiments and scale them down to 0 when not required or stop the compute for notebooks.

    You can utilize free resources for all the above scenarios since Azure ML only charges for the compute and other resources that are created as part of your experiments. I would recommend to delete the attached compute that is already created and use the compute instance for notebook if you are new to Azure ML. I hope this helps.