Transfer Learning Examples train_vgg16 fine tune fine modified the output of vgg16 use trained parameters, modified the output from 1000 to 2 Also, modified the linear layer 4096 to save memory python3 no longer support / for integer to tensor use torch.true_divide or float or // instead