Merge pull request #111 from SainsburyWellcomeCentre/history_agent_fix

fixing agent history to pop first element when exceding 1000 items

neuralplayground/agents/agent_core.py

@@ -80,9 +80,7 @@ def act(self, obs, policy_func=None):
 
         self.obs_history.append(obs)
         if len(self.obs_history) >= 1000:  # reset every 1000
-            self.obs_history = [
-                obs,
-            ]
+            self.obs_history.pop(0)
         if policy_func is not None:
             return policy_func(obs)