go

agent_afterstate.py

@@ -115,6 +115,8 @@ class afterstateAgent:
                 comb += [int(np.log2(state[r][l]))]
             else:
                 comb += [0]
+        # if sum(comb) > sum(list(self._max_ind))
+            # self._max_ind = tuple(comb)
         return tuple(comb)
 
     # def _size(self, mat):
@@ -122,6 +124,7 @@ class afterstateAgent:
         
     def act(self):
         self.forget = self.TD_lambda
+        # self._max_ind = (0,0,0,0)
         action_index = np.argmax(np.array([self._afterstate(a,act=True) for a in range(len(Action))]))
         self._action_index = action_index
         return Action(action_index+1)