I was deeply impressed with Gorners tutorial on Youtube: https://www.youtube.com/watch?v=u4alGiomYP4 He has a clear explanation and nice visuals to see what he means.
I downloaded his code from GitHub (https://github.com/martin-gorner/tensorflow-mnist-tutorial) to get a better understanding of what he was doing. I found out that in his code the visuals and the tensorflow code mix up.
For me this was a problem as I don't know much about tensorflow AND the matplotlib.animation lib is new to me. I want to learn about tensorflow so I decided to separate the tensorflow code from the visuals.
Maybe somebody else wants to have a look at the pure tensorflow part too, that's why I put it on GitHub. (And by the way, there is a DataFrame plot at the end, so there is a little bit of visual output to compare to what Martin Gorner shows in his tutorial.