424 lines
65 KiB
Plaintext
424 lines
65 KiB
Plaintext
{
|
||
"cells": [
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 9,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"import sys\n",
|
||
"from pathlib import Path\n",
|
||
"curr_path = str(Path().absolute()) # 当前路径\n",
|
||
"parent_path = str(Path().absolute().parent) # 父路径\n",
|
||
"sys.path.append(parent_path) # 添加路径到系统路径\n",
|
||
"\n",
|
||
"import math,random\n",
|
||
"import gym\n",
|
||
"import torch\n",
|
||
"import torch.nn as nn\n",
|
||
"import torch.optim as optim\n",
|
||
"import torch.nn.functional as F\n",
|
||
"import numpy as np\n",
|
||
"import matplotlib.pyplot as plt\n",
|
||
"import seaborn as sns\n",
|
||
"from IPython.display import clear_output # 清空单元格输出区域"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 网络模型"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 10,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"class MLP(nn.Module):\n",
|
||
" def __init__(self, state_dim,action_dim,hidden_dim=128):\n",
|
||
" \"\"\" 初始化q网络，为全连接网络\n",
|
||
" state_dim: 输入的特征数即环境的状态数\n",
|
||
" action_dim: 输出的动作维度\n",
|
||
" \"\"\"\n",
|
||
" super(MLP, self).__init__()\n",
|
||
" self.fc1 = nn.Linear(state_dim, hidden_dim) # 输入层\n",
|
||
" self.fc2 = nn.Linear(hidden_dim,hidden_dim) # 隐藏层\n",
|
||
" self.fc3 = nn.Linear(hidden_dim, action_dim) # 输出层\n",
|
||
" \n",
|
||
" def forward(self, x):\n",
|
||
" # 各层对应的激活函数\n",
|
||
" x = F.relu(self.fc1(x)) \n",
|
||
" x = F.relu(self.fc2(x))\n",
|
||
" return self.fc3(x)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 经验回放"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 11,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"class ReplayBuffer:\n",
|
||
" def __init__(self, capacity):\n",
|
||
" self.capacity = capacity # 经验回放的容量\n",
|
||
" self.buffer = [] # 缓冲区\n",
|
||
" self.position = 0 \n",
|
||
" \n",
|
||
" def push(self, state, action, reward, next_state, done):\n",
|
||
" ''' 缓冲区是一个队列，容量超出时去掉开始存入的转移(transition)\n",
|
||
" '''\n",
|
||
" if len(self.buffer) < self.capacity:\n",
|
||
" self.buffer.append(None)\n",
|
||
" self.buffer[self.position] = (state, action, reward, next_state, done)\n",
|
||
" self.position = (self.position + 1) % self.capacity \n",
|
||
" \n",
|
||
" def sample(self, batch_size):\n",
|
||
" batch = random.sample(self.buffer, batch_size) # 随机采出小批量转移\n",
|
||
" state, action, reward, next_state, done = zip(*batch) # 解压成状态，动作等\n",
|
||
" return state, action, reward, next_state, done\n",
|
||
" \n",
|
||
" def __len__(self):\n",
|
||
" ''' 返回当前存储的量\n",
|
||
" '''\n",
|
||
" return len(self.buffer)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## DQN"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 12,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"class DQN:\n",
|
||
" def __init__(self, state_dim, action_dim, cfg):\n",
|
||
"\n",
|
||
" self.action_dim = action_dim # 总的动作个数\n",
|
||
" self.device = cfg.device # 设备，cpu或gpu等\n",
|
||
" self.gamma = cfg.gamma # 奖励的折扣因子\n",
|
||
" # e-greedy策略相关参数\n",
|
||
" self.frame_idx = 0 # 用于epsilon的衰减计数\n",
|
||
" self.epsilon = lambda frame_idx: cfg.epsilon_end + \\\n",
|
||
" (cfg.epsilon_start - cfg.epsilon_end) * \\\n",
|
||
" math.exp(-1. * frame_idx / cfg.epsilon_decay)\n",
|
||
" self.batch_size = cfg.batch_size\n",
|
||
" self.policy_net = MLP(state_dim, action_dim,hidden_dim=cfg.hidden_dim).to(self.device)\n",
|
||
" self.target_net = MLP(state_dim, action_dim,hidden_dim=cfg.hidden_dim).to(self.device)\n",
|
||
" for target_param, param in zip(self.target_net.parameters(),self.policy_net.parameters()): # 复制参数到目标网路targe_net\n",
|
||
" target_param.data.copy_(param.data)\n",
|
||
" self.optimizer = optim.Adam(self.policy_net.parameters(), lr=cfg.lr) # 优化器\n",
|
||
" self.memory = ReplayBuffer(cfg.memory_capacity) # 经验回放\n",
|
||
"\n",
|
||
" def choose_action(self, state):\n",
|
||
" ''' 选择动作\n",
|
||
" '''\n",
|
||
" self.frame_idx += 1\n",
|
||
" if random.random() > self.epsilon(self.frame_idx):\n",
|
||
" with torch.no_grad():\n",
|
||
" state = torch.tensor([state], device=self.device, dtype=torch.float32)\n",
|
||
" q_values = self.policy_net(state)\n",
|
||
" action = q_values.max(1)[1].item() # 选择Q值最大的动作\n",
|
||
" else:\n",
|
||
" action = random.randrange(self.action_dim)\n",
|
||
" return action\n",
|
||
" def update(self):\n",
|
||
" if len(self.memory) < self.batch_size: # 当memory中不满足一个批量时，不更新策略\n",
|
||
" return\n",
|
||
" # 从经验回放中(replay memory)中随机采样一个批量的转移(transition)\n",
|
||
" state_batch, action_batch, reward_batch, next_state_batch, done_batch = self.memory.sample(\n",
|
||
" self.batch_size)\n",
|
||
" # 转为张量\n",
|
||
" state_batch = torch.tensor(state_batch, device=self.device, dtype=torch.float)\n",
|
||
" action_batch = torch.tensor(action_batch, device=self.device).unsqueeze(1) \n",
|
||
" reward_batch = torch.tensor(reward_batch, device=self.device, dtype=torch.float) \n",
|
||
" next_state_batch = torch.tensor(next_state_batch, device=self.device, dtype=torch.float)\n",
|
||
" done_batch = torch.tensor(np.float32(done_batch), device=self.device)\n",
|
||
" q_values = self.policy_net(state_batch).gather(dim=1, index=action_batch) # 计算当前状态(s_t,a)对应的Q(s_t, a)\n",
|
||
" next_q_values = self.target_net(next_state_batch).max(1)[0].detach() # 计算下一时刻的状态(s_t_,a)对应的Q值\n",
|
||
" # 计算期望的Q值，对于终止状态，此时done_batch[0]=1, 对应的expected_q_value等于reward\n",
|
||
" expected_q_values = reward_batch + self.gamma * next_q_values * (1-done_batch)\n",
|
||
" loss = nn.MSELoss()(q_values, expected_q_values.unsqueeze(1)) # 计算均方根损失\n",
|
||
" # 优化更新模型\n",
|
||
" self.optimizer.zero_grad() \n",
|
||
" loss.backward()\n",
|
||
" for param in self.policy_net.parameters(): # clip防止梯度爆炸\n",
|
||
" param.grad.data.clamp_(-1, 1)\n",
|
||
" self.optimizer.step()"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"### DQN参数"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 13,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"class DQNConfig:\n",
|
||
" def __init__(self):\n",
|
||
" self.algo = \"DQN\" # 算法名称\n",
|
||
" self.env = 'CartPole-v0' # 环境名称\n",
|
||
" self.train_eps = 200 # 训练的回合数\n",
|
||
" self.eval_eps = 20 # 测试的回合数\n",
|
||
" self.gamma = 0.95 # 强化学习中的折扣因子\n",
|
||
" self.epsilon_start = 0.90 # e-greedy策略中初始epsilon\n",
|
||
" self.epsilon_end = 0.01 # e-greedy策略中的终止epsilon\n",
|
||
" self.epsilon_decay = 500 # e-greedy策略中epsilon的衰减率\n",
|
||
" self.lr = 0.0001 # 学习率\n",
|
||
" self.memory_capacity = 100000 # 经验回放的容量\n",
|
||
" self.batch_size = 64 # mini-batch SGD中的批量大小\n",
|
||
" self.target_update = 4 # 目标网络的更新频率\n",
|
||
" self.device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\") # 检测GPU\n",
|
||
" self.hidden_dim = 256 # 网络隐藏层"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 创建环境"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 14,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"def env_agent_config(cfg,seed=1):\n",
|
||
" ''' 创建环境和智能体\n",
|
||
" '''\n",
|
||
" env = gym.make(cfg.env) # 创建环境\n",
|
||
" env.seed(seed) # 设置随机种子\n",
|
||
" state_dim = env.observation_space.shape[0] # 状态数\n",
|
||
" action_dim = env.action_space.n # 动作数\n",
|
||
" agent = DQN(state_dim,action_dim,cfg) # 创建智能体\n",
|
||
" return env,agent"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 训练"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 15,
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"开始训练!\n",
|
||
"环境：CartPole-v0, 算法：DQN, 设备：cuda\n",
|
||
"回合：10/200, 奖励：12.0\n",
|
||
"回合：20/200, 奖励：16.0\n",
|
||
"回合：30/200, 奖励：15.0\n",
|
||
"回合：40/200, 奖励：14.0\n",
|
||
"回合：50/200, 奖励：13.0\n",
|
||
"回合：60/200, 奖励：27.0\n",
|
||
"回合：70/200, 奖励：36.0\n",
|
||
"回合：80/200, 奖励：33.0\n",
|
||
"回合：90/200, 奖励：200.0\n",
|
||
"回合：100/200, 奖励：200.0\n",
|
||
"回合：110/200, 奖励：200.0\n",
|
||
"回合：120/200, 奖励：200.0\n",
|
||
"回合：130/200, 奖励：200.0\n",
|
||
"回合：140/200, 奖励：200.0\n",
|
||
"回合：150/200, 奖励：200.0\n",
|
||
"回合：160/200, 奖励：200.0\n",
|
||
"回合：170/200, 奖励：200.0\n",
|
||
"回合：180/200, 奖励：200.0\n",
|
||
"回合：190/200, 奖励：200.0\n",
|
||
"回合：200/200, 奖励：200.0\n",
|
||
"完成训练！\n"
|
||
]
|
||
},
|
||
{
|
||
"data": {
|
||
"image/png": "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",
|
||
"text/plain": [
|
||
"<Figure size 432x288 with 1 Axes>"
|
||
]
|
||
},
|
||
"metadata": {},
|
||
"output_type": "display_data"
|
||
}
|
||
],
|
||
"source": [
|
||
"def train(cfg, env, agent):\n",
|
||
" ''' 训练\n",
|
||
" '''\n",
|
||
" print('开始训练!')\n",
|
||
" print(f'环境：{cfg.env}, 算法：{cfg.algo}, 设备：{cfg.device}')\n",
|
||
" rewards = [] # 记录所有回合的奖励\n",
|
||
" ma_rewards = [] # 记录所有回合的滑动平均奖励\n",
|
||
" for i_ep in range(cfg.train_eps):\n",
|
||
" ep_reward = 0 # 记录一回合内的奖励\n",
|
||
" state = env.reset() # 重置环境，返回初始状态\n",
|
||
" while True:\n",
|
||
" action = agent.choose_action(state) # 选择动作\n",
|
||
" next_state, reward, done, _ = env.step(action) # 更新环境，返回transition\n",
|
||
" agent.memory.push(state, action, reward, next_state, done) # 保存transition\n",
|
||
" state = next_state # 更新下一个状态\n",
|
||
" agent.update() # 更新智能体\n",
|
||
" ep_reward += reward # 累加奖励\n",
|
||
" if done:\n",
|
||
" break\n",
|
||
" if (i_ep+1) % cfg.target_update == 0: # 智能体目标网络更新\n",
|
||
" agent.target_net.load_state_dict(agent.policy_net.state_dict())\n",
|
||
" if (i_ep+1)%10 == 0: \n",
|
||
" print('回合：{}/{}, 奖励：{}'.format(i_ep+1, cfg.train_eps, ep_reward))\n",
|
||
" rewards.append(ep_reward)\n",
|
||
" if ma_rewards:\n",
|
||
" ma_rewards.append(0.9*ma_rewards[-1]+0.1*ep_reward)\n",
|
||
" else:\n",
|
||
" ma_rewards.append(ep_reward)\n",
|
||
" print('完成训练！')\n",
|
||
" return rewards, ma_rewards\n",
|
||
"\n",
|
||
"def plot_rewards(rewards,ma_rewards,plot_cfg):\n",
|
||
" # clear_output(True) # 清空单元格输出区域，因为多次打印，每次需要清楚前面打印的图片\n",
|
||
" sns.set() \n",
|
||
" plt.figure() # 创建一个图形实例，方便同时多画几个图\n",
|
||
" plt.title(\"learning curve on {} of {} for {}\".format(plot_cfg.device, plot_cfg.algo, plot_cfg.env))\n",
|
||
" plt.xlabel('epsiodes')\n",
|
||
" plt.plot(rewards,label='rewards')\n",
|
||
" plt.plot(ma_rewards,label='ma rewards')\n",
|
||
" plt.legend()\n",
|
||
" plt.show()\n",
|
||
"\n",
|
||
"class PlotConfig:\n",
|
||
" def __init__(self) -> None:\n",
|
||
" self.algo = \"DQN\" # 算法名称\n",
|
||
" self.env = 'CartPole-v0' # 环境名称\n",
|
||
" self.device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\") # 检测GPU\n",
|
||
"\n",
|
||
"cfg = DQNConfig()\n",
|
||
"plot_cfg = PlotConfig()\n",
|
||
"env,agent = env_agent_config(cfg,seed=1)\n",
|
||
"rewards, ma_rewards = train(cfg, env, agent)\n",
|
||
"plot_rewards(rewards, ma_rewards, plot_cfg) # 画出结果"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": 16,
|
||
"metadata": {},
|
||
"outputs": [
|
||
{
|
||
"name": "stdout",
|
||
"output_type": "stream",
|
||
"text": [
|
||
"开始测试!\n",
|
||
"环境：CartPole-v0, 算法：DQN, 设备：cuda\n",
|
||
"回合：3/20, 奖励：200.0\n",
|
||
"回合：6/20, 奖励：200.0\n",
|
||
"回合：9/20, 奖励：200.0\n",
|
||
"回合：12/20, 奖励：200.0\n",
|
||
"回合：15/20, 奖励：200.0\n",
|
||
"回合：18/20, 奖励：200.0\n",
|
||
"完成测试！\n"
|
||
]
|
||
},
|
||
{
|
||
"data": {
|
||
"image/png": "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",
|
||
"text/plain": [
|
||
"<Figure size 432x288 with 1 Axes>"
|
||
]
|
||
},
|
||
"metadata": {},
|
||
"output_type": "display_data"
|
||
}
|
||
],
|
||
"source": [
|
||
"def eval(cfg,env,agent):\n",
|
||
" print('开始测试!')\n",
|
||
" print(f'环境：{cfg.env}, 算法：{cfg.algo}, 设备：{cfg.device}')\n",
|
||
" # 由于测试不需要使用epsilon-greedy策略，所以相应的值设置为0\n",
|
||
" cfg.epsilon_start = 0.0 # e-greedy策略中初始epsilon\n",
|
||
" cfg.epsilon_end = 0.0 # e-greedy策略中的终止epsilon\n",
|
||
" rewards = [] # 记录所有回合的奖励\n",
|
||
" ma_rewards = [] # 记录所有回合的滑动平均奖励\n",
|
||
" for i_ep in range(cfg.eval_eps):\n",
|
||
" ep_reward = 0 # 记录一回合内的奖励\n",
|
||
" state = env.reset() # 重置环境，返回初始状态\n",
|
||
" while True:\n",
|
||
" action = agent.choose_action(state) # 选择动作\n",
|
||
" next_state, reward, done, _ = env.step(action) # 更新环境，返回transition\n",
|
||
" state = next_state # 更新下一个状态\n",
|
||
" ep_reward += reward # 累加奖励\n",
|
||
" if done:\n",
|
||
" break\n",
|
||
" rewards.append(ep_reward)\n",
|
||
" if ma_rewards:\n",
|
||
" ma_rewards.append(ma_rewards[-1]*0.9+ep_reward*0.1)\n",
|
||
" else:\n",
|
||
" ma_rewards.append(ep_reward)\n",
|
||
" if (i_ep+1)%3 == 0: \n",
|
||
" print(f\"回合：{i_ep+1}/{cfg.eval_eps}, 奖励：{ep_reward:.1f}\")\n",
|
||
" print('完成测试！')\n",
|
||
" return rewards,ma_rewards\n",
|
||
"\n",
|
||
"rewards,ma_rewards = eval(cfg,env,agent)\n",
|
||
"plot_rewards(rewards,ma_rewards, plot_cfg) # 画出结果\n"
|
||
]
|
||
}
|
||
],
|
||
"metadata": {
|
||
"interpreter": {
|
||
"hash": "fe38df673a99c62a9fea33a7aceda74c9b65b12ee9d076c5851d98b692a4989a"
|
||
},
|
||
"kernelspec": {
|
||
"display_name": "Python 3.7.10 64-bit ('py37': conda)",
|
||
"language": "python",
|
||
"name": "python3"
|
||
},
|
||
"language_info": {
|
||
"codemirror_mode": {
|
||
"name": "ipython",
|
||
"version": 3
|
||
},
|
||
"file_extension": ".py",
|
||
"mimetype": "text/x-python",
|
||
"name": "python",
|
||
"nbconvert_exporter": "python",
|
||
"pygments_lexer": "ipython3",
|
||
"version": "3.7.9"
|
||
},
|
||
"metadata": {
|
||
"interpreter": {
|
||
"hash": "366e1054dee9d4501b0eb8f87335afd3c67fc62db6ee611bbc7f8f5a1fefe232"
|
||
}
|
||
},
|
||
"orig_nbformat": 2
|
||
},
|
||
"nbformat": 4,
|
||
"nbformat_minor": 2
|
||
}
|