Could not correctly profile python tasks [No CUDA events collected. Does the process use CUDA?]

I do need some help for my first nsight experiment on my windows pc. I wish to profile the training process of my dnn task which used pytorch. However, I can not run it correctly. (No CUDA events collected. Does the process use CUDA?) I have the RTX 3090 as my GPU. Please help!

Okay, to start with, what version of Nsight Systems are you using?

Are you using the GUI or the CLI? What options did you use?