I have a related question: I read here that 4GB is a http limitation and ClearML will not chunk single files. I take from that, that ClearML did not want/there was no need to implement an own solution so far. But what about models that are larger than 4GB?