How to pass a Tensor list to cuda or c++ kernel

1

I can successfully pass a Tensor from the python front end to a cuda kernel’s argument following the tutorial at https://pytorch.org/tutorials/advanced/cpp_extension.html#writing-a-mixed-c-cuda-extension

However, I am wondering how can I pass a Tensor List from python to cuda, as there is no such c or cuda library for a “Tensor List”.

For example, if I have a Tensor List tensor_list = [tensor1, tensor2], and a cuda kernel add_cu,
if I want to be able to call add_cu by tensor3 = add_cu(tensor_list), then how should I declare this cuda kernel?