Minor fixes

Author: gabrielplicks

sampler.py

@@ -8,6 +8,8 @@
 
 class Agent:
     def __init__(self, env=Environment(), output_path=None):
+        self.output_path = output_path
+        
         # Training
         self.n_steps = 100000  # 100k
         self.memory = list()
@@ -23,20 +25,26 @@ def __init__(self, env=Environment(), output_path=None):
     def sample(self):
         state = self.env.reset()
 
+        import time
+        start = time.time()
+
         for step in range(self.n_steps):
             action = random.randrange(self.n_actions)
 
             next_state, reward, done, _ = self.env.step(action)
 
-            self.memory.append((state, [action], [reward], next_state, [done]))
+            self.memory.append((state.tolist(), [action], [reward], next_state.tolist(), [done]))
 
             state = next_state
 
             if (step != 0 and step % self.target_update_interval == 0):
-                with open('memory_samples.json', 'w+') as f:
+                with open('{}_samples.json'.format(self.output_path), 'w+') as f:
                     json.dump(self.memory, f)
+                print("")
+                print(time.time() - start)
                 print(step)
                 self.env.reset()
+                break
 
         self.env.close()
 
@@ -45,6 +53,24 @@ def sample(self):
     import os
     print("Restarting PostgreSQL...")
     os.system('sudo systemctl restart postgresql@12-main')
+
+    agent1 = Agent(env=Environment(allow_columns=False, flip=False, window_size=40), output_path='env1')
+    agent2 = Agent(env=Environment(allow_columns=False, flip=False, window_size=80), output_path='env2')
+
+    agent3 = Agent(env=Environment(allow_columns=True, flip=False, window_size=40), output_path='env3')
+    agent4 = Agent(env=Environment(allow_columns=True, flip=False, window_size=80), output_path='env4')
+
+    agent5 = Agent(env=Environment(allow_columns=False, flip=True, window_size=40), output_path='env5')
+    agent6 = Agent(env=Environment(allow_columns=False, flip=True, window_size=80), output_path='env6')
 
-    agent = Agent(env=Environment())
-    agent.sample()
+    agent7 = Agent(env=Environment(allow_columns=True, flip=True, window_size=40), output_path='env7')
+    agent8 = Agent(env=Environment(allow_columns=True, flip=True, window_size=80), output_path='env8')
+
+    agent1.sample()
+    agent2.sample()
+    agent3.sample()
+    agent4.sample()
+    agent5.sample()
+    agent6.sample()
+    agent7.sample()
+    agent8.sample()