How can I get each CUDA thread’s dynamic instruction sequence and num?

How can I get each CUDA thread’s dynamic instruction sequence and num?
I have tried to search for the proper metrics in Nsight Compute but failed.