Baselines version issues 0.1.4

[Kjell-K]

First of all thanks a lot for sharing this repository. Before I adapt for Multirotor, there are some issues when trying to run the gocar.py.

1. mlp() got an unexpected keyword argument 'layer_norm' model = deepq.models.mlp([64], layer_norm = True) Seems strange since the OpenAI code allows layer_norm input.

2. learn() got an unexpected keyword argument 'param_noise' When I remove layer_norm argument to proceed I get this error. Again actually an argument that should be valid.

It goes on like this. Is there anything I am missing to set things up?

[Kjell-K]

It is definitively trouble with baselines. I am successfully running other gym integrations like keras-rl.

[kaihuchen]

I noticed from the repository history of gym and baselines that both layer_norm and param_noise were added on July 27. You might want to check to see if you have older versions of gym/baseline on your computer

[Kjell-K]

@kaihuchen That seems like a lead, but I have the newest version, baselines (0.1.4). Their github: version="0.1.4

Which system are you using?

[kaihuchen]

I installed everything about a week ago, so I guess I have the latest of Gym/Baselines/AirSim from Github.

As for my experiments I did them on Windows 10, solely for the reason that building the AirSim Linux binaries turned out to be not very straightforward, so I took the shortcut of just downloading the pre-built Windows binaries.

[Kjell-K]

So we have the same set up and I am testing also with the pre-built Windows binaries. Strange. I will try to look further into it. For now I will work with the running keras-rl then.