- bugfix not rewarding properly

- bugfix not set value properly after learn

machine_player.py

@@ -60,7 +60,7 @@ class MachinePlayer(APlayer):
 		rev_hist = list(reversed(self.episode_history))
 		for hist in rev_hist:
 				d = self.calc_value(hist['state'], hist['next_state'])
-				if hist['is_exp']:
+				if not hist['is_exp']:
 					self.set_value(hist['state'], d)
 
 	def calc_value(self, state: np.array, next_state: np.array):

tic_tac_toe.py

@@ -41,17 +41,17 @@ def play(player_provider: PlayerProvider, k_max=10000, with_debug=False):
 				if with_debug:
 					print(to_state_string(last_state, state))
 				if not has_moved:
+					player.end_game(state, 0.0)
+					other_player.end_game(state, 0.0)
+					run = False
 					if with_debug:
 						print(f"{player.mark}: No more moves")
-						player.end_game(state, 0.0)
-						other_player.end_game(state, 0.0)
-					run = False
 				if player.has_won(state):
+					player.end_game(state, 1.0)
+					other_player.end_game(state, 0.0)
+					run = False
 					if with_debug:
 						print(f"{player.mark}: Has won the game")
-						player.end_game(state, 1.0)
-						other_player.end_game(state, 0.0)
-					run = False
 
 				other_player = player
 				if not run:
@@ -83,7 +83,7 @@ p_mx = MachinePlayer(mark='X', params=MachinePlayer.Params(p_explore=0.1, alpha=
 p_mo = MachinePlayer(mark='O', params=MachinePlayer.Params(p_explore=0.1, alpha=0.1), values=o_values)
 
 if do_training:
-	play(PlayerProvider(p_mx, p_mo), 50000, False)
+	play(PlayerProvider(p_mx, p_mo), 5000, False)
 
 	# Values after training
 	# Convert and write JSON object to file