Loss is going down but Root Mean Squared Error is going up

Question

I am using tensorflow with keras and the keras tuner. As each epoch goes by, my loss goes down by my root mean squared error goes up. How do I correct this so that I can min