Make a reward per episode over time ticker

[GoogleCodeExporter]

If it's not too hard, make a little graphic that just plots reward in the
last k episodes, maybe smoothed, that we can easily plop down on any
visualizer.

Maybe look at the code from Sam Sarjant because I think he does something
like this in his Tetris trainer program.

Original issue reported on code.google.com by brian.ta...@gmail.com on 7 Nov 2008 at 6:08