使用强化学习AlphaZero算法训练五子棋AI

案例内容介绍

AlphaZero是一种强化学习算法，近期利用AlphaZero训练出的AI以绝对的优势战胜了多名围棋以及国际象棋冠军。AlphaZero创新点在于，它能够在不依赖于外部先验知识即专家知识、仅仅了解游戏规则的情况下，在棋盘类游戏中获得超越人类的表现。

本次案例将详细的介绍AlphaZero算法核心原理，包括神经网络构建、MCTS搜索、自博弈训练，以代码的形式加深算法理解，算法详情亦可见论文《Mastering the game of Go without human knowledge》。同时本案例提供五子棋强化学习环境，利用AlphaZero进行一次五子棋训练。最后可视化五子棋AI自博弈对局。

由于在标准棋盘下训练一个强力的五子棋AI需要大量的训练时间和资源，本案例将棋盘缩小到了6x6x4，且在运行过程中简化了训练过程，减少了自博弈次数和搜索次数。如果想要完整地训练一个五子棋AlphaZero AI,可在AI Gallery中订阅《Gomoku-训练五子棋小游戏》算法并在ModelArts中进行训练。

源码参考GitHub开源项目AlphaZero_Gomoku

!pip install gym

第

2.进行训练参数配置

为简化训练过程，涉及到影响训练时长的参数都设置的较小，且棋盘大小也减小为6x6,棋子连线降低为4。

步：导入相关的库

import os
import copy
import random
import time
from operator import itemgetter
from collections import defaultdictdeque

import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.autograd import Variable
import torch.optim as optim
import gym
from gym.spaces import BoxDiscrete
import matplotlib.pyplot as plt
from IPython import display

2.进行训练参数配置

为简化训练过程，涉及到影响训练时长的参数都设置的较小，且棋盘大小也减小为6x6,棋子连线降低为4。

board_width = 6        # 棋盘宽
board_height = 6       # 棋盘高
n_in_row = 4           # 胜利需要连成线棋子
c_puct = 5             # 决定探索程度
n_playout = 100        # 每步模拟次数
learn_rate = 0.002     # 学习率
lr_multiplier = 1.0    # 基于KL的自适应学习率调整
temperature = 1.0      # 温度参数
noise_eps = 0.75       # 噪声参数
dirichlet_alpha = 0.3  # dirichlet系数
buffer_size = 5000     # buffer大小
train_batch_size = 128 # batchsize大小
update_epochs = 5      # 多少个epoch更新一次
kl_coeff = 0.02        # kl系数
checkpoint_freq = 20   # 模型保存频率
mcts_infer = 200       # 纯mcts推理时间
restore_model = None   # 是否加载预训练模型
game_batch_num=40      # 训练步数
model_path="."         # 模型保存路径

3.构建环境

五子棋的环境是按照标准gym环境构建的，棋盘宽x高，先在横线、直线或斜对角线上形成n子连线的玩家获胜。
状态空间为[4,棋盘宽，棋盘高]，四个维度分别为当前视角下的位置，对手位置，上次位置以及轮次。

class GomokuEnv(gym.Env):
    def __init__(selfstart_player=0):
        self.start_player = start_player

        self.action_space = Discrete((board_width * board_height))
        self.observation_space = Box(01shape=(4board_widthboard_height))
        self.reward = 0
        self.info = {}
        self.players = [12]  # player1 and player2

    def step(selfaction):
        self.states[action] = self.current_player
        if action in self.availables:
            self.availables.remove(action)

        self.last_move = action

        donewinner = self.game_end()
        reward = 0
        if done:
            if winner == self.current_player:
                reward = 1
            else:
                reward = -1

        self.current_player = (
            self.players[0] if self.current_player == self.players[1]
            else self.players[1]
        )

        # update state
        obs = self.current_state()

        return obsrewarddoneself.info

    def reset(self):
        if board_width < n_in_row or board_height < n_in_row:
            raise Exception('board width and height can not be '
                            'less than {}'.format(n_in_row))
        self.current_player = self.players[self.start_player]  # start player
        # keep available moves in a list
        self.availables = list(range(board_width * board_height))
        self.states = {}
        self.last_move = -1

        return self.current_state()

    def render(selfmode='human'start_player=0):
        width = board_width
        height = board_height

        p1p2 = self.players

        print()
        for x in range(width):
            print("{0:8}".format(x)end='')
        print('\r\n')
        for i in range(height - 1-1-1):
            print("{0:4d}".format(i)end='')
            for j in range(width):
                loc = i * width + j
                p = self.states.get(loc-1)
                if p == p1:
                    print('B'.center(8)end='')
                elif p == p2:
                    print('W'.center(8)end='')
                else:
                    print('_'.center(8)end='')
            print('\r\n\r\n')

    def has_a_winner(self):
        states = self.states
        moved = list(set(range(board_width * board_height)) - set(self.availables))
        if len(moved) < n_in_row * 2 - 1:
            return False-1

        for m in moved:
            h = m // board_width
            w = m % board_width
            player = states[m]

            if (w in range(board_width - n_in_row + 1) and
                    len(set(states.get(i-1) for i in range(mm + n_in_row))) == 1):
                return Trueplayer

            if (h in range(board_height - n_in_row + 1) and
                    len(set(states.get(i-1) for i in range(mm + n_in_row * board_widthboard_width))) == 1):
                return Trueplayer

            if (w in range(board_width - n_in_row + 1) and h in range(board_height - n_in_row + 1) and
                    len(set(
                        states.get(i-1) for i in range(mm + n_in_row * (board_width + 1)board_width + 1))) == 1):
                return Trueplayer

            if (w in range(n_in_row - 1board_width) and h in range(board_height - n_in_row + 1) and
                    len(set(
                        states.get(i-1) for i in range(mm + n_in_row * (board_width - 1)board_width - 1))) == 1):
                return Trueplayer

        return False-1

    def game_end(self):
        """Check whether the game is ended or not"""
        winwinner = self.has_a_winner()
        if win:
            # print("winner is player{}".format(winner))
            return Truewinner
        elif not len(self.availables):
            return True-1
        return False-1

    def current_state(self):
        """return the board state from the perspective of the current player.
        state shape: 4*width*height
        """
        square_state = np.zeros((4board_widthboard_height))
        if self.states:
            movesplayers = np.array(list(zip(*self.states.items())))
            move_curr = moves[players == self.current_player]
            move_oppo = moves[players != self.current_player]
            square_state[0][move_curr // board_width,
                            move_curr % board_height] = 1.0
            square_state[1][move_oppo // board_width,
                            move_oppo % board_height] = 1.0
            # indicate the last move location
            square_state[2][self.last_move // board_width,
                            self.last_move % board_height] = 1.0
        if len(self.states) % 2 == 0:
            square_state[3][::] = 1.0  # indicate the colour to play
        return square_state[:::-1:]

    def start_play(selfplayer1player2start_player=0):
        """start a game between two players"""
        if start_player not in (01):
            raise Exception('start_player should be either 0 (player1 first) '
                            'or 1 (player2 first)')
        self.reset()
        p1p2 = self.players
        player1.set_player_ind(p1)
        player2.set_player_ind(p2)
        players = {p1: player1p2: player2}
        while True:
            player_in_turn = players[self.current_player]
            move = player_in_turn.get_action(self)
            self.step(move)
            endwinner = self.game_end()
            if end:
                return winner

    def start_self_play(selfplayer):
        """ start a self-play game using a MCTS playerreuse the search tree,
        and store the self-play data: (statemcts_probsz) for training
        """
        self.reset()
        statesmcts_probscurrent_players = [][][]
        while True:
            movemove_probs = player.get_action(selfreturn_prob=1)
            # store the data
            states.append(self.current_state())
            mcts_probs.append(move_probs)
            current_players.append(self.current_player)
            # perform a move
            self.step(move)
            endwinner = self.game_end()
            if end:
                # winner from the perspective of the current player of each state
                winners_z = np.zeros(len(current_players))
                if winner != -1:
                    winners_z[np.array(current_players) == winner] = 1.0
                    winners_z[np.array(current_players) != winner] = -1.0
                # reset MCTS root node
                player.reset_player()
                return winnerzip(statesmcts_probswinners_z)
            
    def location_to_move(selflocation):
        if (len(location) != 2):
            return -1
        h = location[0]
        w = location[1]
        move = h * board_width + w
        if (move not in range(board_width * board_width)):
            return -1
        return move

    def move_to_location(selfmove):
        """
        3*3 board's moves like:
        6 7 8
        3 4 5
        0 1 2
        and move 5's location is (1,2)
        """
        h = move // board_width
        w = move % board_width
        return [hw]

4.构建神经网络

网络结构较为简单，backbone部分是三层卷积神经网络，提取特征后分为两个分支。一个是价值分支，输出当前棋面价值。另一个是决策分支，输出神经网络计算得到的动作对应概率。

class Net(nn.Module):
    """policy-value network module"""

    def __init__(self):
        super(Netself).__init__()

        # common layers
        self.conv1 = nn.Conv2d(432kernel_size=3padding=1)
        self.conv2 = nn.Conv2d(3264kernel_size=3padding=1)
        self.conv3 = nn.Conv2d(64128kernel_size=3padding=1)
        # action policy layers
        self.act_conv1 = nn.Conv2d(1284kernel_size=1)
        self.act_fc1 = nn.Linear(4 * board_width * board_height,
                                 board_width * board_height)
        # state value layers
        self.val_conv1 = nn.Conv2d(1282kernel_size=1)
        self.val_fc1 = nn.Linear(2 * board_width * board_height64)
        self.val_fc2 = nn.Linear(641)

    def forward(selfstate_input):
        # common layers
        x = F.relu(self.conv1(state_input))
        x = F.relu(self.conv2(x))
        x = F.relu(self.conv3(x))
        # action policy layers
        x_act = F.relu(self.act_conv1(x))
        x_act = x_act.view(-14 * board_width * board_height)
        x_act = F.log_softmax(self.act_fc1(x_act))
        # state value layers
        x_val = F.relu(self.val_conv1(x))
        x_val = x_val.view(-12 * board_width * board_height)
        x_val = F.relu(self.val_fc1(x_val))
        x_val = F.tanh(self.val_fc2(x_val))

        return x_actx_val


class PolicyValueNet:
    """policy-value network """

    def __init__(selfmodel_file=None):

        if torch.cuda.is_available():
            self.device = torch.device("cuda")
        else:
            self.device = torch.device("cpu")

        self.l2_const = 1e-4  # coef of l2 penalty
        # the policy value net module

        self.policy_value_net = Net().to(self.device)
        self.optimizer = optim.Adam(self.policy_value_net.parameters(),
                                    weight_decay=self.l2_const)

        if model_file:
            net_params = torch.load(model_file)
            self.policy_value_net.load_state_dict(net_params)

    def policy_value(selfstate_batch):
        """
        input: a batch of states
        output: a batch of action probabilities and state values
        """
        state_batch = Variable(torch.FloatTensor(state_batch).to(self.device))
        log_act_probsvalue = self.policy_value_net(state_batch)
        act_probs = np.exp(log_act_probs.data.cpu().numpy())
        return act_probsvalue.data.cpu().numpy()

    def policy_value_fn(selfboard):
        """
        input: board
        output: a list of (actionprobability) tuples for each available
        action and the score of the board state
        """
        legal_positions = board.availables
        current_state = np.ascontiguousarray(board.current_state().reshape(
            -14board_widthboard_height))
        log_act_probsvalue = self.policy_value_net(
            Variable(torch.from_numpy(current_state)).to(self.device).float())
        act_probs = np.exp(log_act_probs.data.cpu().numpy().flatten())
        act_probs = zip(legal_positionsact_probs[legal_positions])
        value = value.data[0][0]
        return act_probsvalue

    def train_step(selfstate_batchmcts_probswinner_batchlr):
        """perform a training step"""
        # wrap in Variable
        state_batch = Variable(torch.FloatTensor(state_batch).to(self.device))
        mcts_probs = Variable(torch.FloatTensor(mcts_probs).to(self.device))
        winner_batch = Variable(torch.FloatTensor(winner_batch).to(self.device))

        # zero the parameter gradients
        self.optimizer.zero_grad()
        # set learning rate
        for param_group in self.optimizer.param_groups:
            param_group['lr'] = lr

        # forward
        log_act_probsvalue = self.policy_value_net(state_batch)
        # define the loss = (z - v)^2 - pi^T * log(p) + c||theta||^2
        # Note: the L2 penalty is incorporated in optimizer
        value_loss = F.mse_loss(value.view(-1)winner_batch)
        policy_loss = -torch.mean(torch.sum(mcts_probs * log_act_probs1))
        loss = value_loss + policy_loss
        # backward and optimize
        loss.backward()
        self.optimizer.step()
        # calc policy entropyfor monitoring only
        entropy = -torch.mean(
            torch.sum(torch.exp(log_act_probs) * log_act_probs1)
        )
        # return loss.dataentropy.data
        # for pytorch version >= 0.5 please use the following line instead.
        return loss.item()entropy.item()

    def get_policy_param(self):
        net_params = self.policy_value_net.state_dict()
        return net_params

    def save_model(selfmodel_file):
        """ save model params to file """
        net_params = self.get_policy_param()  # get model params
        torch.save(net_paramsmodel_file)

5.实现MCTS¶

AlphaZero利用MCTS来自博弈生成棋局，MCTS搜索原理简述如下：

1. 每次模拟通过选择具有最大行动价值Q的边加上取决于所存储的先验概率P和该边的访问计数N（每次访问都被增加一次）的上限置信区间U来遍历树。

2. 展开叶子节点，通过神经网络来评估局面s；向量P的值存储在叶子结点扩展的边上。

3. 更新行动价值Q等于在该行动下的子树中的所有评估值V的均值。

4. 一旦MCTS搜索完成，返回局面s下的落子概率π。

def softmax(x):
    probs = np.exp(x - np.max(x))
    probs /= np.sum(probs)
    return probs


def rollout_policy_fn(board):
    """a coarsefast version of policy_fn used in the rollout phase."""
    # rollout randomly
    action_probs = np.random.rand(len(board.availables))
    return zip(board.availablesaction_probs)


def policy_value_fn(board):
    """a function that takes in a state and outputs a list of (actionprobability)
    tuples and a score for the state"""
    # return uniform probabilities and 0 score for pure MCTS
    action_probs = np.ones(len(board.availables)) / len(board.availables)
    return zip(board.availablesaction_probs)0


class TreeNode:
    """A node in the MCTS tree.

    Each node keeps track of its own value Qprior probability Pand
    its visit-count-adjusted prior score u.
    """

    def __init__(selfparentprior_p):
        self._parent = parent
        self._children = {}  # a map from action to TreeNode
        self._n_visits = 0
        self._Q = 0
        self._u = 0
        self._P = prior_p

    def expand(selfaction_priors):
        """Expand tree by creating new children.
        action_priors: a list of tuples of actions and their prior probability
            according to the policy function.
        """
        for actionprob in action_priors:
            if action not in self._children:
                self._children[action] = TreeNode(selfprob)

    def select(selfc_puct):
        """Select action among children that gives maximum action value Q
        plus bonus u(P).
        Return: A tuple of (actionnext_node)
        """
        return max(self._children.items(),
                   key=lambda act_node: act_node[1].get_value(c_puct))

    def update(selfleaf_value):
        """Update node values from leaf evaluation.
        leaf_value: the value of subtree evaluation from the current player's
            perspective.
        """
        # Count visit.
        self._n_visits += 1
        # Update Qa running average of values for all visits.
        self._Q += 1.0 * (leaf_value - self._Q) / self._n_visits

    def update_recursive(selfleaf_value):
        """Like a call to update()but applied recursively for all ancestors.
        """
        # If it is not rootthis node's parent should be updated first.
        if self._parent:
            self._parent.update_recursive(-leaf_value)
        self.update(leaf_value)

    def get_value(selfc_puct):
        """Calculate and return the value for this node.
        It is a combination of leaf evaluations Qand this node's prior
        adjusted for its visit countu.
        c_puct: a number in (0inf) controlling the relative impact of
            value Qand prior probability Pon this node's score.
        """
        self._u = (c_puct * self._P *
                   np.sqrt(self._parent._n_visits) / (1 + self._n_visits))
        return self._Q + self._u

    def is_leaf(self):
        """Check if leaf node (i.e. no nodes below this have been expanded)."""
        return self._children == {}

    def is_root(self):
        return self._parent is None


class MCTS:
    """An implementation of Monte Carlo Tree Search."""

    def __init__(selfpolicy_value_fnc_puct=5):
        """
        policy_value_fn: a function that takes in a board state and outputs
            a list of (actionprobability) tuples and also a score in [-11]
            (i.e. the expected value of the end game score from the current
            player's perspective) for the current player.
        c_puct: a number in (0inf) that controls how quickly exploration
            converges to the maximum-value policy. A higher value means
            relying on the prior more.
        """
        self._root = TreeNode(None1.0)
        self._policy = policy_value_fn
        self._c_puct = c_puct

    def _playout(selfstate):
        """Run a single playout from the root to the leafgetting a value at
        the leaf and propagating it back through its parents.
        State is modified in-placeso a copy must be provided.
        """
        node = self._root
        while (1):
            if node.is_leaf():
                break
            # Greedily select next move.
            actionnode = node.select(self._c_puct)
            state.step(action)

        # Evaluate the leaf using a network which outputs a list of
        # (actionprobability) tuples p and also a score v in [-11]
        # for the current player.
        action_probsleaf_value = self._policy(state)
        # Check for end of game.
        endwinner = state.game_end()
        if not end:
            node.expand(action_probs)
        else:
            # for end statereturn the true leaf_value
            if winner == -1:  # tie
                leaf_value = 0.0
            else:
                leaf_value = (
                    1.0 if winner == state.current_player else -1.0
                )

        # Update value and visit count of nodes in this traversal.
        node.update_recursive(-leaf_value)

    def _playout_p(selfstate):
        """Run a single playout from the root to the leafgetting a value at
        the leaf and propagating it back through its parents.
        State is modified in-placeso a copy must be provided.
        """
        node = self._root
        while (1):
            if node.is_leaf():
                break
            # Greedily select next move.
            actionnode = node.select(self._c_puct)
            state.step(action)

        action_probs_ = self._policy(state)
        # Check for end of game
        endwinner = state.game_end()
        if not end:
            node.expand(action_probs)
        # Evaluate the leaf node by random rollout
        leaf_value = self._evaluate_rollout(state)
        # Update value and visit count of nodes in this traversal.
        node.update_recursive(-leaf_value)

    def _evaluate_rollout(selfenvlimit=1000):
        """Use the rollout policy to play until the end of the game,
        returning +1 if the current player wins-1 if the opponent wins,
        and 0 if it is a tie.
        """
        player = env.current_player
        for i in range(limit):
            endwinner = env.game_end()
            if end:
                break
            action_probs = rollout_policy_fn(env)
            max_action = max(action_probskey=itemgetter(1))[0]
            env.step(max_action)
        else:
            # If no break from the loopissue a warning.
            print("WARNING: rollout reached move limit")
        if winner == -1:  # tie
            return 0
        else:
            return 1 if winner == player else -1

    def get_move_probs(selfstatetemp=1e-3):
        """Run all playouts sequentially and return the available actions and
        their corresponding probabilities.
        state: the current game state
        temp: temperature parameter in (01] controls the level of exploration
        """
        for n in range(n_playout):
            state_copy = copy.deepcopy(state)
            self._playout(state_copy)

        # calc the move probabilities based on visit counts at the root node
        act_visits = [(actnode._n_visits)
                      for actnode in self._root._children.items()]
        actsvisits = zip(*act_visits)
        act_probs = softmax(1.0 / temp * np.log(np.array(visits) + 1e-10))

        return actsact_probs

    def get_move(selfstate):
        """Runs all playouts sequentially and returns the most visited action.
        state: the current game state

        Return: the selected action
        """
        for n in range(n_playout):
            state_copy = copy.deepcopy(state)
            self._playout_p(state_copy)
        return max(self._root._children.items(),
                   key=lambda act_node: act_node[1]._n_visits)[0]

    def update_with_move(selflast_move):
        """Step forward in the treekeeping everything we already know
        about the subtree.
        """
        if last_move in self._root._children:
            self._root = self._root._children[last_move]
            self._root._parent = None
        else:
            self._root = TreeNode(None1.0)

    def __str__(self):
        return "MCTS"

6.实现自博弈过程

实现自博弈训练，此处博弈双方分别为基于MCTS的神经网络和纯MCTS，对弈过程中，前者基于神经网络和MCTS获取最优下子策略，而后者则仅根据MCTS搜索下子策略。保存对局数据

class MCTS_Pure:
    """AI player based on MCTS"""

    def __init__(self):
        self.mcts = MCTS(policy_value_fnc_puct)

    def set_player_ind(selfp):
        self.player = p

    def reset_player(self):
        self.mcts.update_with_move(-1)

    def get_action(selfboard):
        sensible_moves = board.availables
        if len(sensible_moves) > 0:
            move = self.mcts.get_move(board)
            self.mcts.update_with_move(-1)
            return move
        else:
            print("WARNING: the board is full")

    def __str__(self):
        return "MCTS {}".format(self.player)


class MCTSPlayer(MCTS_Pure):
    """AI player based on MCTS"""

    def __init__(selfpolicy_value_functionis_selfplay=0):
        super(MCTS_Pureself).__init__()
        self.mcts = MCTS(policy_value_functionc_puct)
        self._is_selfplay = is_selfplay

    def get_action(selfenvreturn_prob=0):
        sensible_moves = env.availables
        # the pi vector returned by MCTS as in the alphaGo Zero paper
        move_probs = np.zeros(board_width * board_width)
        if len(sensible_moves) > 0:
            actsprobs = self.mcts.get_move_probs(envtemperature)
            move_probs[list(acts)] = probs
            if self._is_selfplay:
                # add Dirichlet Noise for exploration (needed for
                # self-play training)
                move = np.random.choice(
                    acts,
                    p=noise_eps * probs + (1 - noise_eps) * np.random.dirichlet(
                        dirichlet_alpha * np.ones(len(probs))))
                # update the root node and reuse the search tree
                self.mcts.update_with_move(move)
            else:
                # with the default temp=1e-3it is almost equivalent
                # to choosing the move with the highest prob
                move = np.random.choice(actsp=probs)
                # reset the root node
                self.mcts.update_with_move(-1)

            if return_prob:
                return movemove_probs
            else:
                return move
        else:
            print("WARNING: the board is full")

7.训练主函数

训练过程包括自我对局，数据生成，模型更新和保存

class TrainPipeline:
    def __init__(self):
        # params of the board and the game
        self.env = GomokuEnv()

        # training params
        self.data_buffer = deque(maxlen=buffer_size)
        self.play_batch_size = 1
        self.best_win_ratio = 0.0

        # start training from an initial policy-value net
        self.policy_value_net = PolicyValueNet(model_file=restore_model)

        self.mcts_player = MCTSPlayer(self.policy_value_net.policy_value_fn,
                                      is_selfplay=1)
        self.mcts_infer = mcts_infer
        self.lr_multiplier = lr_multiplier

    def get_equi_data(selfplay_data):
        """augment the data set by rotation and flipping
        play_data: [(statemcts_probwinner_z)......]
        """
        extend_data = []
        for statemcts_porbwinner in play_data:
            for i in [1234]:
                # rotate counterclockwise
                equi_state = np.array([np.rot90(si) for s in state])
                equi_mcts_prob = np.rot90(np.flipud(
                    mcts_porb.reshape(board_heightboard_width))i)
                extend_data.append((equi_state,
                                    np.flipud(equi_mcts_prob).flatten(),
                                    winner))
                # flip horizontally
                equi_state = np.array([np.fliplr(s) for s in equi_state])
                equi_mcts_prob = np.fliplr(equi_mcts_prob)
                extend_data.append((equi_state,
                                    np.flipud(equi_mcts_prob).flatten(),
                                    winner))
        return extend_data

    def collect_selfplay_data(selfn_games=1):
        """collect self-play data for training"""
        for i in range(n_games):
            winnerplay_data = self.env.start_self_play(self.mcts_player)
            play_data = list(play_data)[:]
            self.episode_len = len(play_data)
            # augment the data
            play_data = self.get_equi_data(play_data)
            self.data_buffer.extend(play_data)

    def policy_update(self):
        """update the policy-value net"""
        mini_batch = random.sample(self.data_buffertrain_batch_size)
        state_batch = [data[0] for data in mini_batch]
        mcts_probs_batch = [data[1] for data in mini_batch]
        winner_batch = [data[2] for data in mini_batch]
        old_probsold_v = self.policy_value_net.policy_value(state_batch)
        for i in range(update_epochs):
            lossentropy = self.policy_value_net.train_step(
                state_batch,
                mcts_probs_batch,
                winner_batch,
                learn_rate * self.lr_multiplier)
            new_probsnew_v = self.policy_value_net.policy_value(state_batch)
            kl = np.mean(np.sum(old_probs * (
                    np.log(old_probs + 1e-10) - np.log(new_probs + 1e-10)),
                                axis=1)
                         )
            if kl > kl_coeff * 4:  # early stopping if D_KL diverges badly
                break
        # adaptively adjust the learning rate
        if kl > kl_coeff * 2 and self.lr_multiplier > 0.1:
            self.lr_multiplier /= 1.5
        elif kl < kl_coeff / 2 and self.lr_multiplier < 10:
            self.lr_multiplier *= 1.5
        return lossentropy

    def policy_evaluate(selfn_games=10):
        """
        Evaluate the trained policy by playing against the pure MCTS player
        Note: this is only for monitoring the progress of training
        """
        current_mcts_player = MCTSPlayer(self.policy_value_net.policy_value_fn)
        pure_mcts_player = MCTS_Pure()
        win_cnt = defaultdict(int)
        for i in range(n_games):
            winner = self.env.start_play(current_mcts_player,
                                         pure_mcts_player,
                                         start_player=i % 2)
            win_cnt[winner] += 1
        win_ratio = 1.0 * (win_cnt[1] + 0.5 * win_cnt[-1]) / n_games
        print("num_playouts:{}win: {}lose: {}tie:{}".format(self.mcts_infer,
                                                                  win_cnt[1]win_cnt[2]win_cnt[-1]))
        return win_ratio

    def run(self):
        """run the training pipeline"""
        win_num = 0
        try:
            for i_step in range(game_batch_num):
                self.collect_selfplay_data(self.play_batch_size)
                print("batch i:{}episode_len:{}".format(
                    i_step + 1self.episode_len))
                if len(self.data_buffer) > train_batch_size:
                    lossentropy = self.policy_update()
                # check the performance of the current model,
                # and save the model params
                if (i_step + 1) % checkpoint_freq == 0:
                    print("current self-play batch: {}".format(i_step + 1))
                    win_ratio = self.policy_evaluate()
                    self.policy_value_net.save_model(os.path.join(model_path"newest_model.pt"))
                    if win_ratio > self.best_win_ratio:
                        win_num += 1
                        # print("New best policy!!!!!!!!")
                        self.best_win_ratio = win_ratio
                        # update the best_policy
                        self.policy_value_net.save_model(os.path.join(model_path"best_model.pt"))
                        if self.best_win_ratio == 1.0 and self.mcts_infer < 5000:
                            self.mcts_infer += 1000
                            self.best_win_ratio = 0.0
        except KeyboardInterrupt:
            print('\n\rquit')

        return win_num

8.开始自博弈训练，并保存模型

GPU训练耗时约4分钟

start_t = time.time()
training_pipeline = TrainPipeline()
training_pipeline.run()
print("time cost is {}".format(time.time()-start_t))

batch i:1episode_len:13

batch i:2episode_len:16

batch i:3episode_len:13

batch i:4episode_len:11

batch i:5episode_len:11

batch i:6episode_len:15

batch i:7episode_len:13

batch i:8episode_len:13

batch i:9episode_len:19

batch i:10episode_len:14

batch i:11episode_len:11

batch i:12episode_len:19

batch i:13episode_len:15

batch i:14episode_len:22

batch i:15episode_len:8

batch i:16episode_len:16

batch i:17episode_len:15

batch i:18episode_len:13

batch i:19episode_len:15

batch i:20episode_len:17

current self-play batch: 20

num_playouts:200win: 2lose: 8tie:0

batch i:21episode_len:12

batch i:22episode_len:14

batch i:23episode_len:12

batch i:24episode_len:11

batch i:25episode_len:15

batch i:26episode_len:7

batch i:27episode_len:13

batch i:28episode_len:10

batch i:29episode_len:10

batch i:30episode_len:13

batch i:31episode_len:17

batch i:32episode_len:12

batch i:33episode_len:11

batch i:34episode_len:11

batch i:35episode_len:9

batch i:36episode_len:14

batch i:37episode_len:17

batch i:38episode_len:12

batch i:39episode_len:13

batch i:40episode_len:18

current self-play batch: 40

num_playouts:200win: 3lose: 7tie:0

time cost is 250.64277577400208

9.AI对战

（等待第8步运行结束后再运行此步）
加载模型，进行人机对战

# 定义当前玩家
class CurPlayer:
    player_id = 0


# 可视化部分
class Game(object):
    def __init__(selfboard):
        self.board = board
        self.cell_size = board_width - 1
        self.chess_size = 50 * self.cell_size

        self.whitex = []
        self.whitey = []
        self.blackx = []
        self.blacky = []

        # 棋盘背景色
        self.color = "#e4ce9f"
        self.colors = [[self.color] * self.cell_size for _ in range(self.cell_size)]

    def graphic(selfboardplayer1player2):
        """Draw the board and show game info"""
        plt_figax = plt.subplots(facecolor=self.color)
        ax.set_facecolor(self.color)

        # 制作棋盘
        # mytable = ax.table(cellColours=self.colorsloc='center')
        mytable = plt.table(cellColours=self.colors,
                            colWidths=[1 / board_width] * self.cell_size,
                            loc='center'
                            )

        ax.set_aspect('equal')

        # 网格大小
        cell_height = 1 / board_width
        for poscell in mytable.get_celld().items():
            cell.set_height(cell_height)

        mytable.auto_set_font_size(False)
        mytable.set_fontsize(self.cell_size)
        ax.set_xlim([1board_width * 2 + 1])
        ax.set_ylim([board_height * 2 + 11])
        plt.title("Gomoku")

        plt.axis('off')
        cur_player = CurPlayer()

        while True:
            # left down of mouse
            try:
                if cur_player.player_id == 1:
                    move = player1.get_action(self.board)
                    self.board.step(move)
                    xy = self.board.move_to_location(move)
                    plt.scatter((y + 1) * 2(x + 1) * 2s=self.chess_sizec='white')
                    cur_player.player_id = 0
                elif cur_player.player_id == 0:
                    move = player2.get_action(self.board)
                    self.board.step(move)
                    xy = self.board.move_to_location(move)
                    plt.scatter((y + 1) * 2(x + 1) * 2s=self.chess_sizec='black')
                    cur_player.player_id = 1

                endwinner = self.board.game_end()
                if end:
                    if winner != -1:
                        ax.text(x=board_widthy=(board_height + 1) * 2 + 0.1,
                                s="Game end. Winner is player {}".format(cur_player.player_id)fontsize=10,
                                color='red'weight='bold',
                                horizontalalignment='center')
                    else:
                        ax.text(x=board_widthy=(board_height + 1) * 2 + 0.1,
                                s="Game end. Tie Round".format(cur_player.player_id)fontsize=10color='red',
                                weight='bold',
                                horizontalalignment='center')                   
                    return winner
                display.display(plt.gcf())
                display.clear_output(wait=True)
            except:
                pass

    def start_play(selfplayer1player2start_player=0):
        """start a game between two players"""
        if start_player not in (01):
            raise Exception('start_player should be either 0 (player1 first) '
                            'or 1 (player2 first)')
        self.board.reset()
        p1p2 = self.board.players
        player1.set_player_ind(p1)
        player2.set_player_ind(p2)

        self.graphic(self.boardplayer1player2)

# 初始化棋盘
board = GomokuEnv()
game = Game(board)
# 加载模型
best_policy = PolicyValueNet(model_file="best_model.pt")
# 两个AI对打
mcts_player = MCTSPlayer(best_policy.policy_value_fn)
#开始对打
game.start_play(mcts_playermcts_playerstart_player=0

至此，本案例结束，如果想要完整地训练一个五子棋AlphaZero AI,可在AI Gallery中订阅《Gomoku-训练五子棋小游戏》算法并在ModelArts中进行训练。