pub fn to_ndarray_int<const D: usize>( t: &Tensor<AppBackend, D, Int>, ) -> Tensor<NdArray, D, Int>
Same idea for int tensors, if you ever need it.