Added silent training

tic_tac_toe.py

@@ -122,7 +122,7 @@ class MachinePlayer(APlayer):
     def __init__(self, mark='X', with_debug=False, values=None):
         APlayer.__init__(self, mark)
         self.p_exp = 0.2
-        self.alpha = 0.2
+        self.alpha = 0.9
         self.with_debug = with_debug
         if values is None:
             self.values = {}
@@ -179,7 +179,7 @@ class MachinePlayer(APlayer):
             self.state = self.state_from_move(state, move)
 
             # Learn
-            if not is_exp and self.state_last is not None:
+            if not is_exp:
                 v0 = self.get_value(self.state_last)
                 v1 = self.get_value(self.state)
                 d = max(0, v1-v0)
@@ -246,16 +246,16 @@ def play(player_provider: PlayerProvider, k_max=10000, with_print=False):
                 move += 1
 
 
-px = MachinePlayer(mark='X', with_debug=True)
-po = MachinePlayer(mark='O', with_debug=True)
+px = MachinePlayer(mark='X', with_debug=False)
+po = MachinePlayer(mark='O', with_debug=False)
 
 do_training = 1
 if do_training:
     players = PlayerProvider(px, po)
-    play(players, 10000, True)
+    play(players, 20000, False)
 
-#players = PlayerProvider(MachinePlayer(mark='X', with_debug=True, values=px.values), MachinePlayer(mark='O', with_debug=True, values=po.values))
-#play(players, 100, True)
+players = PlayerProvider(MachinePlayer(mark='X', with_debug=True, values=px.values), MachinePlayer(mark='O', with_debug=True, values=po.values))
+play(players, 100, True)
 
 print("Testing")
 test_sample([0.7, 0.1, 0.1, 0.1])