nn.Functinal vs nn – Pytorch


while adding loss in pytorch I have same function in nn.Functional as well as nn. what is the difference ?

torch.nn.CrossEntropyLoss() and torch.nn.functional.cross_entropy