(A doubt NOT an issue) Gradient of loss function having different sign after performing first-degree differentiation of LogisticLoss

I differentiated the loss function in the LogisiticLoss() class used in XGboost to get the gradient. I am getting (y - p) but turns out that it is -(y - p). For more reference, here is the link to the exact class and function, I am referring to. Am I missing something?