කුන්පෝකර් මතප්රතිවිරුද්ධ කනගාටුව අවම කිරීම (සීඑෆ්ආර්)

38from typing import List, cast, Dict
39
40import numpy as np
41
42from labml import experiment
43from labml.configs import option
44from labml_nn.cfr import History as _History, InfoSet as _InfoSet, Action, Player, CFRConfigs
45from labml_nn.cfr.infoset_saver import InfoSetSaver

48ACTIONS = cast(List[Action], ['p', 'b'])

50CHANCES = cast(List[Action], ['A', 'K', 'Q'])

52PLAYERS = cast(List[Player], [0, 1])

55class InfoSet(_InfoSet):

60    @staticmethod
61    def from_dict(data: Dict[str, any]) -> 'InfoSet':

63        pass

65    def actions(self) -> List[Action]:

69        return ACTIONS

71    def __repr__(self):

75        total = sum(self.cumulative_strategy.values())
76        total = max(total, 1e-6)
77        bet = self.cumulative_strategy[cast(Action, 'b')] / total
78        return f'{bet * 100: .1f}%'

81class History(_History):

95    history: str

97    def __init__(self, history: str = ''):

101        self.history = history

103    def is_terminal(self):

108        if len(self.history) <= 2:
109            return False

111        elif self.history[-1] == 'p':
112            return True

114        elif self.history[-2:] == 'bb':
115            return True

117        else:
118            return False

120    def _terminal_utility_p1(self) -> float:

125        winner = -1 + 2 * (self.history[0] < self.history[1])

128        if self.history[-2:] == 'bp':
129            return 1

131        elif self.history[-2:] == 'bb':
132            return winner * 2

134        elif self.history[-1] == 'p':
135            return winner

137        else:
138            raise RuntimeError()

140    def terminal_utility(self, i: Player) -> float:

145        if i == PLAYERS[0]:
146            return self._terminal_utility_p1()

148        else:
149            return -1 * self._terminal_utility_p1()

151    def is_chance(self) -> bool:

155        return len(self.history) < 2

157    def __add__(self, other: Action):

161        return History(self.history + other)

163    def player(self) -> Player:

167        return cast(Player, len(self.history) % 2)

169    def sample_chance(self) -> Action:

173        while True:

175            r = np.random.randint(len(CHANCES))
176            chance = CHANCES[r]

178            for c in self.history:
179                if c == chance:
180                    chance = None
181                    break

184            if chance is not None:
185                return cast(Action, chance)

187    def __repr__(self):

191        return repr(self.history)

193    def info_set_key(self) -> str:

199        i = self.player()

201        return self.history[i] + self.history[2:]

203    def new_info_set(self) -> InfoSet:

205        return InfoSet(self.info_set_key())

208def create_new_history():

210    return History()

213class Configs(CFRConfigs):

217    pass

220@option(Configs.create_new_history)
221def _cnh():

225    return create_new_history

228def main():

237    experiment.create(name='kuhn_poker', writers={'sqlite'})

239    conf = Configs()

241    experiment.configs(conf)

243    experiment.add_model_savers({'info_sets': InfoSetSaver(conf.cfr.info_sets)})

245    with experiment.start():

247        conf.cfr.iterate()

251if __name__ == '__main__':
252    main()