If users, for example, train several ML models on the same data, we could do this in parallel instead of sequentially.
Desired solution
Add new interface methods to the runner to tell it that operations may be run in parallel. The runner can then decide whether to do so.
At this stage, it might be best to split the Python generator into two, one with and one without runner generation, since the generated code would become very dissimilar.
Is your feature request related to a problem?
If users, for example, train several ML models on the same data, we could do this in parallel instead of sequentially.
Desired solution
Add new interface methods to the runner to tell it that operations may be run in parallel. The runner can then decide whether to do so.
At this stage, it might be best to split the Python generator into two, one with and one without runner generation, since the generated code would become very dissimilar.
Possible alternatives (optional)
No response
Screenshots (optional)
No response
Additional Context (optional)
No response