easy5 pts · first answer countsdeep-learningloss-functionspytorch

Up-weighting a rare class

In a 3-class problem, class 2 is rare. How can you make its errors count more in nn.CrossEntropyLoss?

A.Set reduction="sum" on the lossB.Use label_smoothing=0.5 in the lossC.Pass weight=torch.tensor([1., 1., 5.])D.Call loss.backward() five times per step

Discussion 0

Join free to post in the discussion.

Up-weighting a rare class

Discussion 0

Related questions