Do learning rates adapt to the distribution of rewards?

Do learning rates adapt to the distribution of rewards?