Spandan-Madan
commented
7 years ago One possible reason is that I had written this on PyTorch 0.12, and you might be on 0.2 which was released recently. I tried looking up forums for the error you mentioned, but didn't find much. Basically what seems to be happening is that your input is the scalar 0, instead of a PyTorch variable.
First up, your input and target are not on cuda which doesn't make sense to me. Did you change the file/code, because even the error mentions a file main.py, which does not exist in this repository. If you did change the code, I think it's breaking on these lines and you can try debugging this part
HIN0209
Hello,
It is a great post. I used my custom 2-class dataset and worked well with resnet50 with CPU. However, with 1 or 2 GPU(s), there was a following error. Please advise.
Setting: Ubuntu 14.04, python 3.5.4 pyTorch 0.2 via conda (same error with 0.1.12_2 via pip) CUDA 8.0.61 NVIDIA GeForce 1080Ti x2
(15 , 2 ,.,.) = -1.5081 -1.4036 -1.4036 ... -1.6476 -1.6999 -1.6476 -1.6127 -1.5081 -1.5604 ... -1.5604 -1.5081 -1.6476 -1.6127 -1.6476 -1.6476 ... -1.5779 -1.5430 -1.5081 ... ⋱ ...
-1.6476 -1.6302 -1.7173 ... -0.9156 -1.1421 -1.2641 -1.6650 -1.6127 -1.6476 ... -0.5495 -0.9853 -1.2467 -1.4907 -1.6127 -1.4733 ... -1.6127 -1.2293 -1.5430 [torch.FloatTensor of size 16x3x224x224]
1 0 1 1 1 1 1 1 0 1 1 0 0 0 1 0 [torch.LongTensor of size 16]
Traceback (most recent call last): File "main.py", line 260, in
num_epochs=100)
File "main.py", line 177, in train_model
outputs = model(inputs)
File "/home/owner/anaconda3/envs/pytorch_ssd/lib/python3.5/site-packages/torch/nn/modules/module.py", line 206, in call
result = self.forward(*input, *kwargs)
File "/home/owner/anaconda3/envs/pytorch_ssd/lib/python3.5/site-packages/torchvision/models/resnet.py", line 139, in forward
x = self.conv1(x)
File "/home/owner/anaconda3/envs/pytorch_ssd/lib/python3.5/site-packages/torch/nn/modules/module.py", line 206, in call
result = self.forward(input, **kwargs)
File "/home/owner/anaconda3/envs/pytorch_ssd/lib/python3.5/site-packages/torch/nn/modules/conv.py", line 237, in forward
self.padding, self.dilation, self.groups)
File "/home/owner/anaconda3/envs/pytorch_ssd/lib/python3.5/site-packages/torch/nn/functional.py", line 40, in conv2d
return f(input, weight, bias)
TypeError: argument 0 is not a Variable