[AutoTVM] How AutoTVM judge the effective parameters?


#1

Hi, I am a beginner of TVM. I am trying to understand the working mechanism of AutoTVM.


My understanding of this framework is that when when AutoTVM change the parameters, the Exploration Module will get some experiment feedback. According the feedback, AutoTVM will judge whether the parameter adjustment is good or bad.
I want to know what’s the content of the experiment feedback. Can I extract it?