mammoth-rainbow-82717
04/14/2023, 2:52 PMforeach
and dynamic resource configurations for runs on Kubernetes. For example, it is quite a common use case for us to have a foreach
branch for multiple model training steps for a different setting, e.g., training a different model for a range of different countries.
Typically the resources required for these different models varies dramatically. As such, we would like to allocate different amounts of resource to each of the different steps coming out of the foreach
branching.
Is there some way to achieve this behaviour with Metaflow?