search

eleurent / rl-agents

Implementations of Reinforcement Learning and Planning algorithms
MIT License
553 stars 149 forks source link

Environment window resizing #111

Open Louise599 opened 3 months ago

Louise599 commented 3 months ago

Hi, after I ran

python experiments.py evaluate configs/IntersectionEnv/env.json \ configs/IntersectionEnv/agents/DQNAgent/baseline.json \ After --train --episodes=4000 --name-from-config,

I found that the window of IntersectionEnv is very small, may I know where to adjust the parameters of the window? Looking forward to your reply, thank you very much!

eleurent commented 3 months ago

Hi,

When creating a highway-env environment, you can edit the window size with

env = gym.make('intersection-v0', config={"screen_width": 600, "screen_height": 600})

But since you're using the experiments.py script which handles environment creation internally, you should instead edit the env config file, configs/IntersectionEnv/env.json in your case, and add the "screen_width": 600, "screen_height": 600 options there