Convolution algorithm doesn't work with sparse tensor

Hi guys,

Just wondering if I made a mistake or if it’s not currently implemented, when I try to use a conv2d layer on a sparse tensor I get the following error :

RuntimeError: sparse tensors do not have is_contiguous

I used this code :

import torch
class Net(torch.nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = torch.nn.Conv2d(1, 6, 3)

    def forward(self, x):
        x = self.conv1(x)
        return x
    
inputs = torch.sparse.FloatTensor(1, 1, 500, 500)
model = Net()
model(inputs)