Can I use the profiler in graph execution mode instead of eager execution mode?

[firejq]

My tensorflow version is 2.6, but I use the api under compat.v1, i.e. it is the legcacy session-based graph execution mode.

After reading the profiler's usage documentation, I found that all usage methods seem to be aimed at eager mode. So is there any way to use this project in graph execution mode? Thanks for your help!