Error in Calculating Flops

[Allhailankurgupta]

I was using this code below

class LeNet(nn.Module):
    def __init__(self):
        super(LeNet, self).__init__()
        self.conv1 = nn.Conv2d(1, 6, 5, padding=2)
        self.conv2 = nn.Conv2d(6, 16, 5)
        self.fc1   = nn.Linear(16*5*5, 120)
        self.fc2   = nn.Linear(120, 84)
        self.fc3   = nn.Linear(84, 10)

    def forward(self, x):
        x = F.max_pool2d(F.relu(self.conv1(x)), (2, 2))
        x = F.max_pool2d(F.relu(self.conv2(x)), (2, 2))
        x = x.view(-1, self.num_flat_features(x))
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x

    def num_flat_features(self, x):
        size = x.size()[1:]
        return np.prod(size)

new_model = LeNet()
new_model.to(DEVICE)
from pthflops import count_ops
false_inp = torch.rand(1,1,28,28).to(DEVICE)
# # Count the number of FLOPs
v = count_ops(new_model, false_inp)

and I got this error

It was working perfectly fine a few days back Could you please resolve this issue

[1adrianb]

Hi @Allhailankurgupta,

I tried running your code but I was unable to reproduce your error using pytorch 1.6/1.7 & pthflops 0.3.3 Could you please let me know what versions you are using?

[Allhailankurgupta]

I was running the code on colab so didn't pay that much attention to the versions But I checked I am using pytorch 1.7 And pthflops 0.3.3

[Allhailankurgupta]

I am facing the same error with the code given as illustration for resnet18

[1adrianb]

Unfortunately I am unable to reproduce it. I've also just triggered a build (https://travis-ci.com/github/1adrianb/pytorch-estimate-flops/builds/200193700) and as you can see the code run successfully.
Perhaps there is other code around that interferes?

[Allhailankurgupta]

It is kind of strange I just installed pthflops on my local machine It gave no error at all

But the exact same code is giving error on colab

Can you please check if the error comes on colab?

[Allhailankurgupta]

I just checked it is working with pytorch 1.6.0 but not with pytorch 1.7.0

[1adrianb]

Fixed in 0.3.4