ClearML pipelines can be build from tasks, functions, and decorated functions, according to the examples in None . I am guessing there was a certain way of doing things, that allgegro.ai started with and then there was a pain that the other approaches solved. So, what are the rules-of-thumb when to use which approach?
My current approach with pipelines basically looks like a GH CICD yaml config btw, so I give the user a lot of control on which steps to run, why, and how, and the default simply caches all results so as to minimize the number of reruns.
The user can then override and choose exactly what to do (or not do).
