Benjamin Klebel-Knobloch

@RaphaelMeudec Thank you! It is very good and heartening to know that there is some attention to this topic.

@RaphaelMeudec The nonzero window length option works now, however with an ugly workaround for a rather weird problem: lines 48-50 (in processors.py) exist, because at a nonzero window length, the...

Thanks - I think I solved it in the commits after my comment, but I am not entirely sure. I have tested the dqn_cartpole example, rewriting everything to dictionaries, and...

In this [gist](https://gist.github.com/bklebel/e3bd43ce228a53d27de119c639ac61ee), the there is the code for the keras-rl cartpole example, rewritten for dictionary inputs. The corresponding results are displayed, once for the whole observation put into one...

I think with the pull-request #195 this can be amended: if your image, and vectors (and whatever more input) are named variables, it is rather easy to put them into...

in case you have more actions you can of course change the +1 to a variable....after all, the only problem you have is the assertion check, as you have two...

yes, that's true, as long as ddpg remains unchainged, with #195 it would not learn, and possibly crash anyways. Well, I guess my analysis of the situation was not that...