Azure Machine Learning: Creating ML Pipeline from YAML fails: TabularDataset does not support mount. Only FileDataset supports mount

Question

I'm new to Azure Machine Learning, and trying to create a simple ML pipeline. AzureML supports YAML to define ML pipeline, and it's described here (https://learn.microsoft.com/en-us/azure/machine-learning/reference-pipeline-yaml).

An error I faced is that, when I create a pipeline from "az ml pipeline create" with YAML file, it returns the message below even if I specify "download" for bind_mode of data_references.

Messeage: "<class azureml.data.tabular_dataset.TabularDataset'> does not support mount. Only FileDataset supports mount"

Environment:
OS: Windows 10
Azure CLI: 2.11.1

It seems that bind_mode of Tabular dataset is not working or I miss something. The reason I'm confused is that, as you can see in the sample yaml file described in the link above, dataset with "bind_mode: download" should work.

Sample YAML is below with a defined dataset called "dataset1" of Tabular format.

Sample YAML:

pipeline:
    name: "Sample ML pipeline YAML"
    data_references:
        sampleDS:
            dataset_name: dataset1
            bind_mode: download
    default_compute: compute-name
    steps:
        SampleStep:
            type: PythonScriptStep
            name: SampleProcessing
            script_name: processing.py
            allow_reuse: True
            source_directory: ".\\src\\pipeline\\steps"
            inputs:
                input_ds:
                    source: sampleDS

When data_references is changed to the following (specify the path in a datastore directly, not via registered dataset), it works.

    name: "Sample ML pipeline YAML"
    data_references:
        sampleDS:
            datastore: workspaceblobstore
            path_on_datastore: path/of/sampeDS/sample.csv

Answer 1

Yes.You are right. TabularDataset does not support download or mount. You can create&register a Filedataset and the code sample will work. Learn more about dataset type here