The reasons why PyTorch implements different variants of the cross entropy loss are convenience and computational efficiency.
確定! 回上一頁