Pytorch pretrained models and caffe pretrained models

Why people did not use pytorch pretrained models (vgg, googlelenet) which are easy to use instead of building caffe library and use its pretrained models

Pretrained torchvision models are used in a lot of scripts.
Where did you see Caffe models in PyTorch?

thank for answering my question. I see people using it in kaggle and some github repo which use caffe as feature extractor