Hi, is it possible to store different versions of the calculated asset output and configure a job to use an older output version for downstream assets? I've looked into the provided io managers but didn't find something suitable. In my use case I want to train a ml-model and archive the different model versions. Later, I want be able to use older model versions for prediction in my job.

You could probably do something smart with partitions 🙂

Hi Tobias, We don’t have this kind of “asset history” functionality baked in (yet), but I think that you might be able to use the experimental Dynamic partitions API to achieve what you’re after.

Thanks, I will give it a try