Define errors associated with data during fit

Hey, I'm looking to define the errors associated with each data point that I'm fitting. I was wondering if there was any way to do something like fit(x, y, dy) since the error in y is not as simple as sqrt(y). Thanks for any replies.