ca7a5e9a39478054dbdfbfba9963b13c80b768f8,ch18/telegram-bot.py,Session,move_bot,#Session#,54

Before Change

        state_list = game.decode_binary(self.state)
        batch_v = model.state_lists_to_batch([state_list], [self.BOT_PLAYER])
        logits_v, values_v = self.model(batch_v)
        probs_v = F.softmax(logits_v)
        probs = probs_v[0].data.cpu().numpy()
        self.value = values_v.data.cpu().numpy()[0]
        while True:
            action = np.random.choice(game.GAME_COLS, p=probs)
            if action in game.possible_moves(self.state):

After Change

    def move_bot(self):
        self.mcts_store.search_batch(MCTS_SEARCHES, MCTS_BATCH_SIZE, self.state, self.BOT_PLAYER, self.model)
        probs, values = self.mcts_store.get_policy_value(self.state, tau=0)
        action = np.random.choice(game.GAME_COLS, p=probs)
        self.value = values[action]
        self.moves.append(action)

Instances