detailed steps on how to test pre-trained net for image captioning

[top-on]

I was interested in testing the image captioning example with a pre-trained network. While doing that, I wrote down the practical steps to do that and extended the README.

Maybe you would like to have these steps documented in your repo, too?