Help Needed

[FranzPfeifroth]

I am trying to implement the German Schafkopf game using RLCard. I got a running implementation but the agents are NOT learning. So I am looking for support. Schafkopf is a rather complicated game (in comparision to Poker) with 4 Players and 10 different game types. My assumption is that it requires some effort on your side to understand the problems. If you are willing to spend this effort, please give me a notice.

Thanks in advance Franz Pfeifroth

[daochenzha]

@FranzPfeifroth To my experience, whether it will work mainly depend on the feature designs. My suggestion is to mainly base on DMC in https://github.com/datamllab/rlcard/blob/master/examples/run_dmc.py, which supports multiple actors so the training will be fast, and then iterate on the features to see whether there is any improvement.