Caffe2 models in PyTorch for change or fine-tuning

I heard that Caffe2 was embedded to PyTorch. Does it mean that I can use pretrained model on Caffe2 in PyTorch?

For example, we have great repository: with a lot of pretrained models. Can I use it in PyTorch to fine-tuning or transfer learning without any problem?

If it is possible, give some example!