Loading fastai 1 model into pytorch


For production purposes, I want to load fastai resnet18 model(bear/cat image classification data) into pytorch. In pytorch how do you create the same architecture and then load the weights?
import torch
import torchvision.models as models
obj = torch.load("/content/model_resnet18_v1_save_test.pth",map_location=‘cpu’)
mod = models.resnet18()
how to extend this architecture in pytorch to become a bear/cat classification? and how to load the weights into this model in pytorch?

Thanks in advance