initial commit for reintroducing PyQT rendering

examples/play_model.py

@@ -7,6 +7,7 @@ import torch
 import random
 import numpy as np
 import matplotlib.pyplot as plt
+import redis
 
 
 def main():
@@ -32,6 +33,8 @@ def main():
                 number_of_agents=1)
     env_renderer = RenderTool(env)
     plt.figure(figsize=(5,5))
+    # fRedis = redis.Redis()
+
     handle = env.get_agent_handles()
 
     state_size = 105
@@ -98,6 +101,7 @@ def main():
                 score += all_rewards[a]
 
             env_renderer.renderEnv(show=True, frames=True, iEpisode=trials, iStep=step)
+            sEnv = fR.set("RailEnv0")
 
             obs = next_obs.copy()
             if done['__all__']:
@@ -115,10 +119,8 @@ def main():
             '\tEpsilon: {:.2f} \t Action Probabilities: \t {}'.format(
                 env.number_of_agents,
                 trials,
-                np.mean(
-                    scores_window),
-                100 * np.mean(
-                    done_window),
+                np.mean(scores_window),
+                100 * np.mean(done_window),
                 eps, action_prob/np.sum(action_prob)),
             end=" ")
         if trials % 100 == 0: