Adjusting strategy of learning rate

In pytorch, there are multiple strategies to adjust learning rate. How do I choose from them? To achieve good performance, I have to make much attempt.