Update objects_9x9.py

objects_9x9.py

@@ -76,13 +76,13 @@ class State:
         # backpropagate reward
         if result == 1:
             self.p1.feedReward(1)
-            self.p2.feedReward(0)
+            self.p2.feedReward(-1)
         elif result == -1:
-            self.p1.feedReward(0)
+            self.p1.feedReward(-1)
             self.p2.feedReward(1)
         else:
             self.p1.feedReward(0.1)
-            self.p2.feedReward(0.5)
+            self.p2.feedReward(0.1)
 
     # board reset
     def reset(self):
@@ -189,7 +189,7 @@ class Player:
     def __init__(self, name, exp_rate=0.3):
         self.name = name
         self.states = []  # record all positions taken
-        self.lr = 0.2
+        self.lr = 0.1
         self.exp_rate = exp_rate
         self.decay_gamma = 0.9
         self.states_value = {}  # state -> value
@@ -270,7 +270,7 @@ if __name__ == "__main__":
 
     st = State(p1, p2)
     print("training...")
-    st.playwithbot(200000)
+    st.playwithbot(500000)
 
     p1.savePolicy()
     p2.savePolicy()