Open newhardwarefound opened 1 year ago
Thank you for opening this issue, we will look into it.
Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @azureml-github.
Author: | newhardwarefound |
---|---|
Assignees: | - |
Labels: | `Service Attention`, `Machine Learning`, `customer-reported` |
Milestone: | - |
It is indeed very frustrating. For me it's the only blocking point to migrate my scripts to V2.
Are there any updates on this?
@tdegamboda @newhardwarefound I made a tool on my own to get something similar, to create a worflow of steps (command/node in V2 namming) without worring about the Inputs/Outputs. It will run your steps in the order you gave in a graph.
Related command
az ml job create
Is your feature request related to a problem? Please describe.
I am frustrated when trying to get one AML step to start after another AML step in a pipeline, when there are no input/output files between the 2 steps. It's a useful v1 feature (https://learn.microsoft.com/en-us/python/api/azureml-pipeline-core/azureml.pipeline.core.stepsequence?view=azure-ml-py) that is removed in v2. Airflow also supports it naturally. Right now I have to modify like 6 different files to add a placeholder parameter for the 2 successive steps.
Describe the solution you'd like
Add a field called prerequisites in the pipelinecommandjob yaml so that i can specify which steps need to complete before starting the current step.
Describe alternatives you've considered
Additional context