Tianxiaomo / pytorch-YOLOv4

PyTorch ,ONNX and TensorRT implementation of YOLOv4
Apache License 2.0
4.47k stars 1.49k forks source link

I can not run train because the gradient can not be counted #432

Open qam1314 opened 3 years ago

qam1314 commented 3 years ago

RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.FloatTensor [1, 3, 19, 19, 121]], which is output 0 of AsStridedBackward, is at version 6; expected version 3 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!

JiaweiLian commented 3 years ago

Is this problem solved? I got the same problem

JustWon commented 2 years ago

@qam1314 @JiaweiLian Check that you are using pytorch version 1.4.0 as illustrated in requirement.txt. That solved the problem in my case.