google-research / sam

Apache License 2.0
565 stars 72 forks source link

Something wrong before the second loss backward #17

Open KeranLi opened 1 year ago

KeranLi commented 1 year ago

RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [64, 1]], which is output 0 of TBackward, is at version 2; expected version 1 instead. Hint: the backtrace further above shows the operation that failed to compute its gradient. The variable in question was changed in there or anywhere later. Good luck!