Torch.hub.load vs torchvision.model

Greetings! These two commands result in the “same” model if I dump them with torchsummary, but end up causing different downloads (cached files).

import torch.hub
hub_model = torch.hub.load(‘pytorch/vision’, ‘resnet18’, pretrained=True)

import torchvision
made_model = torchvision.models.resnet18(pretrained=True)

I am preloading caches for a container for distribution and I want to undestand why they are different and how. Also, how to determine the download name from each call would be useful.

Thanks for any help!