MountainCarContinious-vo_DDPG.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from gym import make\n",
    "import random\n",
    "import numpy as np\n",
    "from collections import namedtuple, deque\n",
    "import torch\n",
    "import numpy as np\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as F\n",
    "import matplotlib.pyplot as plt\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# создаем буффер с данными, чтобы семплировать не скореллированные состояния\n",
    "Transaction = namedtuple('Transaction',\n",
    "                         ('state', 'action', 'next_state', 'reward'))\n",
    "\n",
    "\n",
    "class Buffer(object):\n",
    "\n",
    "    def __init__(self, capacity):\n",
    "        self.capacity = capacity\n",
    "        self.memory = deque(maxlen=capacity)\n",
    "\n",
    "    def push(self, *args):\n",
    "        self.memory.appendleft(Transaction(*args))\n",
    "\n",
    "    def sample(self, batch_size):\n",
    "        batch = random.sample(self.memory, batch_size)\n",
    "        batch = Transaction(*zip(*batch))\n",
    "        self.state = torch.tensor(batch.state, dtype = torch.float)\n",
    "        self.action = torch.tensor(batch.action, dtype = torch.float)\n",
    "        self.next_state = torch.tensor(batch.next_state, dtype = torch.float)\n",
    "        self.reward = torch.tensor(batch.reward, dtype = torch.float)\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.memory)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# создаем скелеты двух сетей\n",
    "class Actor(nn.Module):\n",
    "    def __init__(self, state_dim, action_dim):\n",
    "        super(Actor, self).__init__()\n",
    "        self.fc1 = nn.Linear(state_dim, 64)\n",
    "        self.fc2 = nn.Linear(64, 64)\n",
    "        self.fc3 = nn.Linear(64, action_dim)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = F.relu(self.fc1(x))\n",
    "        x = F.relu(self.fc2(x))\n",
    "        x = self.fc3(x)\n",
    "        return torch.tanh(x)\n",
    "\n",
    "\n",
    "class Critic(nn.Module):\n",
    "    def __init__(self, state_dim, action_dim):\n",
    "        super(Critic, self).__init__()\n",
    "        self.fc1 = nn.Linear(state_dim, 64)\n",
    "        self.fc2 = nn.Linear(64 + action_dim, 64)\n",
    "        self.fc3 = nn.Linear(64, action_dim)\n",
    "\n",
    "    def forward(self, x, a):\n",
    "        x = F.relu(self.fc1(x))\n",
    "        x = F.relu(self.fc2(torch.cat([x, a], dim=1)))\n",
    "        x = self.fc3(x)\n",
    "        return x\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# создаем шум, который будем добавлять к действиям для более качественного експлора среды\n",
    "class Noise:\n",
    "    def __init__(self, mu=0.0, theta=0.1, sigma=0.1):\n",
    "        self.mu = mu\n",
    "        self.theta = theta\n",
    "        self.sigma = sigma\n",
    "        self.action_dim = 1\n",
    "        self.low = -1.0\n",
    "        self.high = 1.0\n",
    "        self.state = np.ones(self.action_dim) * self.mu\n",
    "\n",
    "    def noise(self):\n",
    "        x = self.state\n",
    "        dx = self.theta * (self.mu - x) + self.sigma * np.random.randn(self.action_dim)\n",
    "        self.state = x + dx\n",
    "        return self.state\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "class AQL:\n",
    "    def __init__(self, states, actions, eps, episodes, batch_size, mem_sz, alpha_actor,\n",
    "                 alpha_critic, gamma, tau):\n",
    "        self.gamma = gamma\n",
    "        self.alpha_actor = alpha_actor\n",
    "        self.alpha_critic = alpha_critic\n",
    "        self.eps = eps\n",
    "        self.episodes = episodes\n",
    "        self.batch_size = batch_size\n",
    "        self.memory = Buffer(mem_sz)\n",
    "        self.tau = tau\n",
    "        self.noise = Noise()\n",
    "        self.device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "        self.actor = Actor(states, actions).to(self.device).apply(self.weights)\n",
    "        self.critic = Critic(states, actions).to(self.device).apply(self.weights)\n",
    "\n",
    "        self.actor_target = Actor(states, actions).to(self.device)\n",
    "        self.actor_target.load_state_dict(self.actor.state_dict())\n",
    "\n",
    "        self.critic_target = Critic(states, actions).to(self.device)\n",
    "        self.critic_target.load_state_dict(self.critic.state_dict())\n",
    "\n",
    "        self.actor_optim = optim.Adam(self.actor.parameters(), lr=self.alpha_actor)\n",
    "        self.critic_optim = optim.Adam(self.critic.parameters(), lr=self.alpha_critic)\n",
    "        self.critic_criterion = nn.MSELoss()\n",
    "\n",
    "    @staticmethod\n",
    "    def weights(layer):\n",
    "        if isinstance(layer, nn.Linear):\n",
    "            nn.init.xavier_normal_(layer.weight)\n",
    "\n",
    "    def update(self):\n",
    "        if len(self.memory) <= self.batch_size:\n",
    "            return\n",
    "        self.memory.sample(self.batch_size)\n",
    "        Qvals = self.critic.forward(self.memory.state, self.memory.action.unsqueeze(1))\n",
    "        next_actions = self.actor_target.forward(self.memory.next_state)\n",
    "        next_Q = self.critic_target.forward(self.memory.next_state, next_actions.detach())\n",
    "        Qprime = self.memory.reward.unsqueeze(1) + self.gamma * next_Q\n",
    "        critic_loss = self.critic_criterion(Qvals, Qprime)\n",
    "        policy_loss = -self.critic.forward(self.memory.state, self.actor.forward(self.memory.state)).mean()\n",
    "\n",
    "        self.actor_optim.zero_grad()\n",
    "        policy_loss.backward()\n",
    "        self.critic_optim.zero_grad()\n",
    "        critic_loss.backward()\n",
    "\n",
    "        for param_actor, param_critic in zip(self.actor.parameters(), self.critic.parameters()):\n",
    "            param_actor.grad.data.clamp_(-1, 1)\n",
    "            param_critic.grad.data.clamp_(-1, 1)\n",
    "        self.actor_optim.step()\n",
    "        self.critic_optim.step()\n",
    "\n",
    "        for target_param, param in zip(self.actor_target.parameters(), self.actor.parameters()):\n",
    "            target_param.data.copy_(param.data * self.tau + target_param.data * (1.0 - self.tau))\n",
    "\n",
    "        for target_param, param in zip(self.critic_target.parameters(), self.critic.parameters()):\n",
    "            target_param.data.copy_(param.data * self.tau + target_param.data * (1.0 - self.tau))\n",
    "\n",
    "    def act(self, state):\n",
    "        return np.clip(self.actor(state).detach() + self.eps * self.noise.noise(), -1.0, 1.0)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Test:\n",
    "    def __init__(self, path, rewards, test_episodes):\n",
    "        self.rewards = rewards\n",
    "        self.test_episodes = test_episodes\n",
    "        self.device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "        self.model = torch.load(path).to(self.device)\n",
    "        self.run()\n",
    "\n",
    "    def act(self, state):\n",
    "        state = torch.tensor(state, dtype = torch.float).unsqueeze(0)\n",
    "        with torch.no_grad():\n",
    "            return self.model(state)\n",
    "\n",
    "    def run(self):\n",
    "        env = make('MountainCarContinuous-v0')\n",
    "        rewards = []\n",
    "        for i in range(self.test_episodes):\n",
    "            state = env.reset()\n",
    "            total_reward = 0\n",
    "            done = False\n",
    "            while not done:\n",
    "                action = self.act(state)\n",
    "                next_state, reward, done, _ = env.step(action)\n",
    "                total_reward += reward\n",
    "                state = next_state\n",
    "            rewards.append(total_reward)\n",
    "        self.test_rewards = rewards\n",
    "        self.m = np.mean(rewards)\n",
    "\n",
    "    def picture(self):\n",
    "        plt.figure(figsize=(16, 16))\n",
    "        ax1 = plt.subplot(421)\n",
    "        ax2 = plt.subplot(422)\n",
    "        ax1.plot(np.arange(0, len(self.rewards)), self.rewards, label=f'rewards')\n",
    "        ax1.set_title(f'Train')\n",
    "        ax1.set_xlabel('Episode')\n",
    "        ax1.set_ylabel('Reward')\n",
    "        ax1.legend()\n",
    "\n",
    "        ax2.plot(np.arange(0, len(self.test_rewards)), self.test_rewards, label=f'rewards')\n",
    "        ax2.axhline(y=self.m, color='r', linestyle='-', label=f'mean reward: {self.m}')\n",
    "        ax2.set_title(f'Test')\n",
    "        ax2.set_xlabel('Episode')\n",
    "        ax2.set_ylabel('Reward')\n",
    "        ax2.legend()\n",
    "        plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "env = make('MountainCarContinuous-v0')\n",
    "variable = 12061997\n",
    "env.seed(variable)\n",
    "np.random.seed(variable)\n",
    "random.seed(variable)\n",
    "torch.manual_seed(variable)\n",
    "\n",
    "aql = AQL(states=env.observation_space.shape[0], actions=env.action_space.shape[0],\n",
    "          eps=1.0, episodes=50, batch_size=64, mem_sz=60000,\n",
    "          alpha_actor=1e-3, alpha_critic=5e-4, gamma=0.85, tau=0.45)\n",
    "\n",
    "rewards = []\n",
    "score = deque(maxlen=5)\n",
    "best_score = -10000\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Episode 0, total_reward: -80.8496206904777, mean five: -80.8496206904777, total mean nan, EPS: 0.9\n",
      "Episode 1, total_reward: -68.84009220495113, mean five: -74.84485644771442, total mean -80.8496206904777, EPS: 0.81\n",
      "Episode 2, total_reward: -46.1423921697793, mean five: -65.27736835506938, total mean -74.84485644771442, EPS: 0.7290000000000001\n",
      "Episode 3, total_reward: 92.45502383545076, mean five: -25.844270307439345, total mean -65.27736835506938, EPS: 0.6561000000000001\n",
      "Episode 4, total_reward: -88.72918451835159, mean five: -38.42125314962179, total mean -25.844270307439345, EPS: 0.5904900000000002\n",
      "Episode 5, total_reward: 70.94897182621901, mean five: -8.06153464628245, total mean -38.42125314962179, EPS: 0.5314410000000002\n",
      "updated!!!! -8.06153464628245 ---------------------\n",
      "Episode 6, total_reward: 89.64092184337572, mean five: 23.634668163382923, total mean -20.192882320314993, EPS: 0.47829690000000014\n",
      "updated!!!! 23.634668163382923 ---------------------\n",
      "Episode 7, total_reward: 88.39194561907017, mean five: 50.54153572115281, total mean -4.5023388683591765, EPS: 0.43046721000000016\n",
      "updated!!!! 50.54153572115281 ---------------------\n",
      "Episode 8, total_reward: 81.30795339241087, mean five: 48.31212163254484, total mean 7.109446692569492, EPS: 0.38742048900000015\n",
      "Episode 9, total_reward: 88.73394426991158, mean five: 83.80474739019748, total mean 15.353725214774089, EPS: 0.34867844010000015\n",
      "updated!!!! 83.80474739019748 ---------------------\n",
      "Episode 10, total_reward: -7.833258986286725, mean five: 68.04830122769633, total mean 22.691747120287836, EPS: 0.31381059609000017\n",
      "Episode 11, total_reward: 92.51380822363122, mean five: 68.62287850374742, total mean 19.916746565144695, EPS: 0.28242953648100017\n",
      "Episode 12, total_reward: 85.88361637342501, mean five: 68.12121265461839, total mean 25.966501703351906, EPS: 0.25418658283290013\n",
      "Episode 13, total_reward: -10.424981165196838, mean five: 49.77462574309685, total mean 30.57551052412676, EPS: 0.22876792454961012\n",
      "Episode 14, total_reward: 95.18545260397231, mean five: 51.064927409909, total mean 27.646903974889362, EPS: 0.2058911320946491\n",
      "Episode 15, total_reward: 94.96339832930398, mean five: 71.62425887302713, total mean 32.14947388349489, EPS: 0.1853020188851842\n",
      "Episode 16, total_reward: 94.16769325888397, mean five: 71.9550358800777, total mean 36.07534416135796, EPS: 0.16677181699666577\n",
      "Episode 17, total_reward: 96.2208715122288, mean five: 74.02248690783844, total mean 39.492541167094785, EPS: 0.1500946352969992\n",
      "Episode 18, total_reward: 93.8186729757242, mean five: 94.87121773602264, total mean 42.644115075157785, EPS: 0.13508517176729928\n",
      "updated!!!! 94.87121773602264 ---------------------\n",
      "Episode 19, total_reward: 95.76037205868776, mean five: 94.98620162696574, total mean 45.337512859398124, EPS: 0.12157665459056936\n",
      "updated!!!! 94.98620162696574 ---------------------\n",
      "Episode 20, total_reward: 96.02764373840081, mean five: 95.19905070878511, total mean 47.8586558193626, EPS: 0.10941898913151243\n",
      "updated!!!! 95.19905070878511 ---------------------\n",
      "Episode 21, total_reward: 96.82896969458274, mean five: 95.73130599592487, total mean 50.15241714884062, EPS: 0.0984770902183612\n",
      "updated!!!! 95.73130599592487 ---------------------\n",
      "Episode 22, total_reward: 96.06610824656278, mean five: 95.70035334279166, total mean 52.27407862819254, EPS: 0.08862938119652508\n",
      "Episode 23, total_reward: 95.39990433984691, mean five: 96.0165996156162, total mean 54.17807991594777, EPS: 0.07976644307687257\n",
      "updated!!!! 96.0165996156162 ---------------------\n",
      "Episode 24, total_reward: 96.44580574674553, mean five: 96.15368635322776, total mean 55.89565593361022, EPS: 0.07178979876918531\n",
      "updated!!!! 96.15368635322776 ---------------------\n",
      "Episode 25, total_reward: 94.77184268404206, mean five: 95.90252614235601, total mean 57.51766192613563, EPS: 0.06461081889226679\n",
      "Episode 26, total_reward: 96.6358076346898, mean five: 95.86389373037741, total mean 58.950515032208955, EPS: 0.05814973700304011\n",
      "Episode 27, total_reward: 96.71305124805352, mean five: 95.99328233067556, total mean 60.346266610078615, EPS: 0.0523347633027361\n",
      "Episode 28, total_reward: 95.86324690885047, mean five: 96.08595084447627, total mean 61.64508034714915, EPS: 0.04710128697246249\n",
      "Episode 29, total_reward: 96.92698845663067, mean five: 96.18218738645331, total mean 62.82501712513885, EPS: 0.042391158275216244\n",
      "updated!!!! 96.18218738645331 ---------------------\n",
      "Episode 30, total_reward: 97.2590220222991, mean five: 96.67962325410471, total mean 63.96174950285524, EPS: 0.03815204244769462\n",
      "updated!!!! 96.67962325410471 ---------------------\n",
      "Episode 31, total_reward: 96.81737710202881, mean five: 96.7159371475725, total mean 65.0358550679986, EPS: 0.03433683820292516\n",
      "updated!!!! 96.7159371475725 ---------------------\n",
      "Episode 32, total_reward: 96.14557674990024, mean five: 96.60244224794187, total mean 66.02902763156204, EPS: 0.030903154382632643\n",
      "Episode 33, total_reward: 97.50856589029053, mean five: 96.93150604422988, total mean 66.94165033211775, EPS: 0.02781283894436938\n",
      "updated!!!! 96.93150604422988 ---------------------\n",
      "Episode 34, total_reward: 96.09894961865608, mean five: 96.76589827663496, total mean 67.84067726029929, EPS: 0.025031555049932444\n",
      "Episode 35, total_reward: 95.728518241153, mean five: 96.45979752040573, total mean 68.64805647053807, EPS: 0.0225283995449392\n",
      "Episode 36, total_reward: 96.39388442120139, mean five: 96.37509898424025, total mean 69.4002915197218, EPS: 0.020275559590445278\n",
      "Episode 37, total_reward: 95.88290675655526, mean five: 96.32256498557125, total mean 70.12984808462666, EPS: 0.01824800363140075\n",
      "Episode 38, total_reward: 96.74456829852681, mean five: 96.16976546721851, total mean 70.80756015494057, EPS: 0.016423203268260675\n",
      "Episode 39, total_reward: 94.02659031837183, mean five: 95.75529360716166, total mean 71.47261164580176, EPS: 0.014780882941434608\n",
      "Episode 40, total_reward: 95.92037707001654, mean five: 95.79366537293437, total mean 72.03646111261601, EPS: 0.013302794647291147\n",
      "Episode 41, total_reward: 96.18586432042994, mean five: 95.75206135278009, total mean 72.61899564816235, EPS: 0.011972515182562033\n",
      "Episode 42, total_reward: 95.77570264709979, mean five: 95.730620530889, total mean 73.18011156893064, EPS: 0.01077526366430583\n",
      "Episode 43, total_reward: 95.57724154509006, mean five: 95.49715518020164, total mean 73.70559043121365, EPS: 0.009697737297875247\n",
      "Episode 44, total_reward: 94.35998124425824, mean five: 95.56383336537893, total mean 74.20267341107447, EPS: 0.008727963568087723\n",
      "Episode 45, total_reward: 94.3104504329441, mean five: 95.24184803796443, total mean 74.65061358514522, EPS: 0.00785516721127895\n",
      "Episode 46, total_reward: 95.60092701636626, mean five: 95.12486057715168, total mean 75.07800134270606, EPS: 0.007069650490151055\n",
      "Episode 47, total_reward: 95.57283326155004, mean five: 95.08428670004173, total mean 75.51465933576266, EPS: 0.00636268544113595\n",
      "Episode 48, total_reward: 95.94039674686448, mean five: 95.15691774039661, total mean 75.93253795921656, EPS: 0.005726416897022355\n",
      "Episode 49, total_reward: 94.43701364874019, mean five: 95.17232422129301, total mean 76.34086160794406, EPS: 0.00515377520732012\n"
     ]
    }
   ],
   "source": [
    "for i in range(aql.episodes):\n",
    "    state = env.reset()\n",
    "    total_reward = 0\n",
    "    done = False\n",
    "    while not done:\n",
    "        action = aql.act(torch.tensor(state, dtype = torch.float).unsqueeze(0))\n",
    "        next_state, reward, done, _ = env.step(action)\n",
    "        total_reward += reward\n",
    "        reward += 13 * abs(next_state[1])\n",
    "        aql.memory.push(state, action, next_state, reward)\n",
    "        state = next_state\n",
    "        aql.update()\n",
    "    aql.eps *= 0.9\n",
    "    score.appendleft(total_reward)\n",
    "    # save best five episodes\n",
    "    print(f'Episode {i},'\n",
    "          f'total_reward: {total_reward},'\n",
    "          f' mean five: {np.mean(score)},'\n",
    "          f' total mean {np.mean(rewards)},'\n",
    "          f' EPS: {aql.eps}')\n",
    "    if np.mean(score) > best_score and i >= 5:\n",
    "        torch.save(aql.actor, \"model.pickle\")\n",
    "        best_score = np.mean(score)\n",
    "        print('updated!!!!', best_score, '---------------------')\n",
    "    rewards.append(total_reward)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x1152 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Test('model.pickle', rewards, 30).picture()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}