Refactored

tic_tac_toe.py

@@ -1,5 +1,5 @@
 import numpy as np
-from helper import create_empty_state, to_state_string
+from state import create_empty_state, to_state_string
 from machine_player import MachinePlayer
 from human_player import HumanPlayer
 from player_provider import PlayerProvider
@@ -66,21 +66,28 @@ with open("values_x.json", "r") as fp:
 with open("values_o.json", "r") as fp:
 	o_values = json.load(fp)
 
-#px = MachinePlayer(mark='X', params=MachinePlayer.Params(p_exp=0.2, alpha=0.1), values=x_values)
-px = HumanPlayer(mark='X', name="Jens")
-po = MachinePlayer(mark='O', params=MachinePlayer.Params(p_exp=0.2, alpha=0.1), values=o_values)
-
-player_provider = PlayerProvider(px, po)
-
 do_training = 1
+do_human_player = 1
+
+p_hx = HumanPlayer(mark='X', name="Jens")
+p_mx = MachinePlayer(mark='X', params=MachinePlayer.Params(p_explore=0.2, alpha=0.1), values=x_values)
+p_mo = MachinePlayer(mark='O', params=MachinePlayer.Params(p_explore=0.2, alpha=0.1), values=o_values)
+
 if do_training:
-	play(player_provider, 10000, True)
+	play(PlayerProvider(p_mx, p_mo), 10000, False)
 
-play(player_provider, 10, True)
+	# Values after training
+	# Convert and write JSON object to file
+	with open("values_x.json", "w") as fp:
+		json.dump(p_mx.values, fp, indent=0)
 
-# Convert and write JSON object to file
-with open("values_x.json", "w") as fp:
-	json.dump(x_values, fp, indent=0)
+	with open("values_o.json", "w") as fp:
+		json.dump(p_mo.values, fp, indent=0)
+
+if do_human_player:
+	pp = PlayerProvider(p_hx, p_mo)
+else:
+	pp = PlayerProvider(p_mx, p_mo)
+
+play(pp, 10, True)
 
-with open("values_o.json", "w") as fp:
-	json.dump(o_values, fp, indent=0)