My learning rate finder plot looks as follows after unfreeze.
I am using an unet_learner with pretrained resnet34
I have seen that many people uses tinetune() method instead of unfreeze and fit_one_cycle. What is the difference between these two options???