I’m having trouble getting Atari games to work properly with OpenAI Gym on my Windows 10 system. I found a tutorial about training a neural network to play Atari games using policy gradients, and I copied the code to try it out.
The main issue is that when I try to test the Atari environment with this simple code:
import gym
game_env = gym.make('Breakout-v0')
game_env.reset()
for step in range(1000):
game_env.step(game_env.action_space.sample())
game_env.render('human')
The program runs for about 30 seconds and then crashes completely. I’m not sure what’s causing this problem. When I run the full training code from the tutorial, it works but doesn’t show the actual game graphics, just the training results.
I’m using Windows 10 32-bit with Python 2.7.6 installed. Has anyone else experienced similar crashes with Atari environments on Windows? Any suggestions on how to fix this would be really helpful.