Max running time for one iteration

[joistick11]

Hi

I am using env = normalize(GymEnv("CartPole-v0")) to create an environment. It works well, but all the visualizations lasts maximum for 5 seconds and after that they stop even if the pole was in the vertical state.
Can I change this behaviour somehow?

Thanks.

[dementrock]

By default, visualization only happens when video is being recorded. What behavior do you want to change it to? Always showing video or never?

[joistick11]

I think my question is deeper. I am okay that video is being shown every "perfect cube number". I actually want to change the maximum reward the agent may achieve (by default it is only 200). So in that way the episodes will be longer.
It would be perfect if I can set the goal of balancing for 20 seconds to complete the learning. Please note that I am using gym environment.

[dementrock]

@joistick11 In that case, you will probably need to modify attributes of the constructed gym environment. By default it's the cartpole env wrapped in a TimeLimit class (see https://github.com/openai/gym/blob/d337f4e5719bb8b49f2f9fba1501fe7749fa4d5c/gym/wrappers/time_limit.py)