Run Unity build in inference mode while interacting with mlagents.envs

Hello,

what’s the equivalent approach to run a Unity build in inference mode, while the interaction is done via mlagents.envs.

In v0.10.0 this was done by setting an argument during reset env.reset(train_mode=not self.realtime_mode)[brain_name].
Now I cannot seem to find a solution in v.0.13.1.

I tried to initialize the environment with no_graphics = False and setting the target frame rate inside the engine configuration to 60.

Hi MarcoMeter,

Please have a look here ml-agents/docs/Python-API.md at 0.13.1 · Unity-Technologies/ml-agents · GitHub regarding this issue. I also suggest upgrading to the latest version of ML-Agents.