這個是因為沒有將requires_grad設為True，l=LOSS（out，label）中的l默認是requires_grad為false，這個l其實也是一個張量Tensor類型，將其的requires_grad改為True后，使用backward函數就可以得到requires_grad為True的所有參數的梯度。
上面的loss = loss.requires_grad_()，是我添加的部分。
本錯誤糾正的參考的內容如下：
免責聲明！
本站轉載的文章為個人學習借鑒使用，本站對版權不負任何法律責任。如果侵犯了您的隱私權益，請聯系本站郵箱yoyou2525@163.com刪除。
猜您在找
#報錯記錄#RuntimeError: Expected tensor for 'out' to have the same device as tensor for argument #2 'mat1'; but device 1 does not equal 0 (while checking arguments for addmm)
RuntimeError: invalid argument 0: Sizes of tensors must match except in dimension 1 #805
RuntimeError:Expected to have finished reduction in the prior iteration
DateWindow does not have UPDATE capability
git 報錯does not have a commit checked out
張量與數組運算報錯(Use Tensor.cpu() to copy the tensor to host memory first;RuntimeError: Expected all tensors to be on the same device)
HasOutput(framework::GradVarName"> RuntimeError: (NotFound) No Output(X@GRAD) found for BatchNormGrad operator. [Hint: Expected ctx->HasOutput(framework::GradVarName("X")) == true, but received ctx->HasOutput(framework::GradVarName
Load balancer does not have available server for client: xxx
User xxx does not have privileges for CREATETABLE的解決辦法
tomcat下出現The file is absent or does not have execute&