I got a simple question on Pairwise loss. I think that pairwise loss is used to measure relative rank between in training set. Sometimes we want to train each entities\' rank re