How can i inference with custom op which takes a span as input

[HarryHang]

I am trying to write a custom op and inference a model with it in python. I've read this example https://github.com/microsoft/onnxruntime/blob/rel-1.16.0/onnxruntime/test/testdata/custom_op_library/cpu/cpu_ops.cc and run a model with CustomOpOne successfully, but i notice this function below takes ort::runtime::span as input instead of ort::runtime::tensor(in the case of CustomOpOne).

I tried to add this op in python and inference with it but it came with this error:

[E:onnxruntime:, sequential_executor.cc:514 ExecuteKernel] Non-zero status code returned while running Fuse node. Name:'/Add' Status Message: invalid shape while trying to get a span out of Ort::Custom::Tensor

SO my question is:
1、how can i inference with this op?
2、am i following the right example to implement a custom op?
Thank you.