Stroia Dacian

Dilated layer takes more than `k` neightbours

5

The `Dilated` layer doesn't take into account `k`. This can lead to taking more neighbours than intended. ```python t = torch.tensor([ [0, 0, 0, 0, 0, 0, 1, 1, 1,...

DilatedKnnGraph can accept a badly shaped batch parameter when using "matrix" mode. This happens because only batch_size is used. Maybe an assertion / error should be raised as in Pytorch...