Tensorflow 2.x CPU and GPU profiling and execution

Question

I am training a DNN using tf 2.3 in python 3.7 and have noticed that many of the tf operations are executed in CPU as opposed to GPU, without a clear distinction of how tf d