I think it's supposed to be dL/dw1. As in the loss function derived wrt w1.