Use latest version of ALE Breakout (#142)

Changing to v4, seems to fix the issue. Tested on Python 3.6.3, gym (0.9.4) and atari-py (0.1.1)
keras-rl · Nov 10, 2017 · 5981750 · 5981750
1 parent 67edfb4
commit 5981750
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/examples/dqn_atari.py b/examples/dqn_atari.py
@@ -42,7 +42,7 @@ def process_reward(self, reward):
 
 parser = argparse.ArgumentParser()
 parser.add_argument('--mode', choices=['train', 'test'], default='train')
-parser.add_argument('--env-name', type=str, default='BreakoutDeterministic-v3')
+parser.add_argument('--env-name', type=str, default='BreakoutDeterministic-v4')
 parser.add_argument('--weights', type=str, default=None)
 args = parser.parse_args()