If you are like me, you might be wondering what training times others are observing? Is my tutorial/setup too slow? Perhaps my hardware is different than others.
What do you folks think of instrumenting our code simply like this? And we can compare using the lesson/groups names.
Please see this. Seems easy to include.
mlinstrumentation.start() <my code> mlinstrumentation.stop()
Likely, we can also do this during inference/serving time. Thoughts/feedback?