Pytorch C++ API - Int Tensor?

1

If I do torch::zeros({inputVector.size()}).cuda();

I get a tensor of type Float32. However, I need it to be of type Int32. How can I do this in the C++ API?