RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation

[qiaokang0]

I try the code and get the runtime error down,any idea?

Training...
Epoch-0-: 0%| | 0/391 [00:00<?, ?it/s]Traceback (most recent call last):
File "main.py", line 46, in
main()
File "main.py", line 34, in main
trainer.train()
File "/home/qiaokang/MobileNet-V2-master/train.py", line 63, in train
cur_loss.backward()
File "/usr/local/lib/python3.5/dist-packages/torch/tensor.py", line 93, in backward
torch.autograd.backward(self, gradient, retain_graph, create_graph)
File "/usr/local/lib/python3.5/dist-packages/torch/autograd/init.py", line 89, in backward
allow_unreachable=True) # allow_unreachable flag
RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation

[ddhgiang]

also got the same problem

[nuptwuchen]

same problem

[MG2033]

Which Pytorch version do you use?

[nekulkarni]

I got the same error with Pytorch 0.4.1.

Training...
Epoch-0-: 0%| | 0/391 [00:00<?, ?it/s]Traceback (most recent call last):
File "main.py", line 45, in
main()
File "main.py", line 33, in main
trainer.train()
File "/home/grafty/external_projects/pruning/MobileNet-V2/train.py", line 63, in train
cur_loss.backward()
File "/usr/local/lib/python3.5/dist-packages/torch/tensor.py", line 93, in backward
torch.autograd.backward(self, gradient, retain_graph, create_graph)
File "/usr/local/lib/python3.5/dist-packages/torch/autograd/init.py", line 90, in backward
allow_unreachable=True) # allow_unreachable flag
RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation

[nekulkarni]

Update: worked for me when I used pytorch 0.3.0. The error no longer came up.

[feixuedudiao]

what the checkpoint name,and where can download it?

[jehovahxu]

it worked when I used pytorch 0.3.1 but i was not found the change in pytorch 0.4.x

[lcj1105]

i also got the same problem in pyTorch 0.4.0

[jehovahxu]

i also got the same problem in pyTorch 0.4.0

changing your dropout or relu6 to out-of-place or use the model of https://github.com/tonylins/pytorch-mobilenet-v2

[lcj1105]

i also got the same problem in pyTorch 0.4.0

changing your dropout or relu6 to out-of-place or use the model of https://github.com/tonylins/pytorch-mobilenet-v2

thanks ! When I did this, the problem was solved.

[BlueBubbleWei]

search the keyword "inplace" in train.py and delete the args like inplace=True,or change inplace=False,I solved the problem,it's a pytorch version problem,the author used version 0.4,and yours in version 0.4.