4 - Generalized Advantage Estimation (GAE) [CartPole].ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "https://datascience.stackexchange.com/questions/32480/how-does-generalised-advantage-estimation-work\n",
    "https://lilianweng.github.io/lil-log/2018/02/19/a-long-peek-into-reinforcement-learning.html#combining-td-and-mc-learning\n",
    "https://arxiv.org/pdf/1506.02438.pdf\n",
    "https://github.com/higgsfield/RL-Adventure-2\n",
    "http://www.breloff.com/DeepRL-OnlineGAE/\n",
    "https://arxiv.org/pdf/1804.02717.pdf\n",
    "https://ewrl.files.wordpress.com/2015/02/ewrl12_2015_submission_18.pdf\n",
    "https://github.com/Kaixhin/Dist-A3C\n",
    "https://github.com/Kaixhin/Dist-A3C/blob/master/client.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as F\n",
    "import torch.distributions as distributions\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import gym"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_env = gym.make('CartPole-v1')\n",
    "test_env = gym.make('CartPole-v1')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "SEED = 1234\n",
    "\n",
    "train_env.seed(SEED);\n",
    "test_env.seed(SEED+1);\n",
    "np.random.seed(SEED);\n",
    "torch.manual_seed(SEED);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class MLP(nn.Module):\n",
    "    def __init__(self, input_dim, hidden_dim, output_dim, dropout = 0.5):\n",
    "        super().__init__()\n",
    "\n",
    "        self.fc_1 = nn.Linear(input_dim, hidden_dim)\n",
    "        self.fc_2 = nn.Linear(hidden_dim, output_dim)\n",
    "        self.dropout = nn.Dropout(dropout)\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = self.fc_1(x)\n",
    "        x = self.dropout(x)\n",
    "        x = F.relu(x)\n",
    "        x = self.fc_2(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ActorCritic(nn.Module):\n",
    "    def __init__(self, actor, critic):\n",
    "        super().__init__()\n",
    "        \n",
    "        self.actor = actor\n",
    "        self.critic = critic\n",
    "        \n",
    "    def forward(self, state):\n",
    "        \n",
    "        action_pred = self.actor(state)\n",
    "        value_pred = self.critic(state)\n",
    "        \n",
    "        return action_pred, value_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "INPUT_DIM = train_env.observation_space.shape[0]\n",
    "HIDDEN_DIM = 128\n",
    "OUTPUT_DIM = train_env.action_space.n\n",
    "\n",
    "actor = MLP(INPUT_DIM, HIDDEN_DIM, OUTPUT_DIM)\n",
    "critic = MLP(INPUT_DIM, HIDDEN_DIM, 1)\n",
    "\n",
    "policy = ActorCritic(actor, critic)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def init_weights(m):\n",
    "    if type(m) == nn.Linear:\n",
    "        torch.nn.init.xavier_normal_(m.weight)\n",
    "        m.bias.data.fill_(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ActorCritic(\n",
       "  (actor): MLP(\n",
       "    (fc_1): Linear(in_features=4, out_features=128, bias=True)\n",
       "    (fc_2): Linear(in_features=128, out_features=2, bias=True)\n",
       "    (dropout): Dropout(p=0.5, inplace=False)\n",
       "  )\n",
       "  (critic): MLP(\n",
       "    (fc_1): Linear(in_features=4, out_features=128, bias=True)\n",
       "    (fc_2): Linear(in_features=128, out_features=1, bias=True)\n",
       "    (dropout): Dropout(p=0.5, inplace=False)\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "policy.apply(init_weights)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "LEARNING_RATE = 0.01\n",
    "\n",
    "optimizer = optim.Adam(policy.parameters(), lr = LEARNING_RATE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(env, policy, optimizer, discount_factor, trace_decay):\n",
    "    \n",
    "    policy.train()\n",
    "    \n",
    "    log_prob_actions = []\n",
    "    values = []\n",
    "    rewards = []\n",
    "    done = False\n",
    "    episode_reward = 0\n",
    "\n",
    "    state = env.reset()\n",
    "\n",
    "    while not done:\n",
    "\n",
    "        state = torch.FloatTensor(state).unsqueeze(0)\n",
    "\n",
    "        action_pred, value_pred = policy(state)\n",
    "                \n",
    "        action_prob = F.softmax(action_pred, dim = -1)\n",
    "                \n",
    "        dist = distributions.Categorical(action_prob)\n",
    "\n",
    "        action = dist.sample()\n",
    "        \n",
    "        log_prob_action = dist.log_prob(action)\n",
    "        \n",
    "        state, reward, done, _ = env.step(action.item())\n",
    "\n",
    "        log_prob_actions.append(log_prob_action)\n",
    "        values.append(value_pred)\n",
    "        rewards.append(reward)\n",
    "\n",
    "        episode_reward += reward\n",
    "    \n",
    "    log_prob_actions = torch.cat(log_prob_actions)\n",
    "    values = torch.cat(values).squeeze(-1)\n",
    "    \n",
    "    returns = calculate_returns(rewards, discount_factor)\n",
    "    #note: calculate_advantages takes in rewards, not returns!\n",
    "    advantages = calculate_advantages(rewards, values, discount_factor, trace_decay)\n",
    "    \n",
    "    policy_loss, value_loss = update_policy(advantages, log_prob_actions, returns, values, optimizer)\n",
    "\n",
    "    return policy_loss, value_loss, episode_reward"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calculate_returns(rewards, discount_factor, normalize = True):\n",
    "    \n",
    "    returns = []\n",
    "    R = 0\n",
    "    \n",
    "    for r in reversed(rewards):\n",
    "        R = r + R * discount_factor\n",
    "        returns.insert(0, R)\n",
    "        \n",
    "    returns = torch.tensor(returns)\n",
    "    \n",
    "    if normalize:\n",
    "        \n",
    "        returns = (returns - returns.mean()) / returns.std()\n",
    "        \n",
    "    return returns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calculate_advantages(rewards, values, discount_factor, trace_decay, normalize = True):\n",
    "    \n",
    "    advantages = []\n",
    "    advantage = 0\n",
    "    next_value = 0\n",
    "    \n",
    "    for r, v in zip(reversed(rewards), reversed(values)):\n",
    "        td_error = r + next_value * discount_factor - v\n",
    "        advantage = td_error + advantage * discount_factor * trace_decay\n",
    "        next_value = v\n",
    "        advantages.insert(0, advantage)\n",
    "        \n",
    "    advantages = torch.tensor(advantages)\n",
    "    \n",
    "    if normalize:\n",
    "        advantages = (advantages - advantages.mean()) / advantages.std()\n",
    "        \n",
    "    return advantages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def update_policy(advantages, log_prob_actions, returns, values, optimizer):\n",
    "        \n",
    "    advantages = advantages.detach()\n",
    "    returns = returns.detach()\n",
    "        \n",
    "    policy_loss = - (advantages * log_prob_actions).sum()\n",
    "    \n",
    "    value_loss = F.smooth_l1_loss(returns, values).sum()\n",
    "        \n",
    "    optimizer.zero_grad()\n",
    "    \n",
    "    policy_loss.backward()\n",
    "    value_loss.backward()\n",
    "    \n",
    "    optimizer.step()\n",
    "    \n",
    "    return policy_loss.item(), value_loss.item()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate(env, policy):\n",
    "    \n",
    "    policy.eval()\n",
    "    \n",
    "    rewards = []\n",
    "    done = False\n",
    "    episode_reward = 0\n",
    "\n",
    "    state = env.reset()\n",
    "\n",
    "    while not done:\n",
    "\n",
    "        state = torch.FloatTensor(state).unsqueeze(0)\n",
    "\n",
    "        with torch.no_grad():\n",
    "        \n",
    "            action_pred, _ = policy(state)\n",
    "\n",
    "            action_prob = F.softmax(action_pred, dim = -1)\n",
    "                \n",
    "        action = torch.argmax(action_prob, dim = -1)\n",
    "                \n",
    "        state, reward, done, _ = env.step(action.item())\n",
    "\n",
    "        episode_reward += reward\n",
    "        \n",
    "    return episode_reward"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| Episode:  10 | Mean Train Rewards:  38.9 | Mean Test Rewards:  79.0 |\n",
      "| Episode:  20 | Mean Train Rewards:  42.5 | Mean Test Rewards:  96.7 |\n",
      "| Episode:  30 | Mean Train Rewards:  59.5 | Mean Test Rewards: 180.9 |\n",
      "| Episode:  40 | Mean Train Rewards: 107.4 | Mean Test Rewards: 290.1 |\n",
      "| Episode:  50 | Mean Train Rewards: 128.7 | Mean Test Rewards: 300.9 |\n",
      "| Episode:  60 | Mean Train Rewards: 143.0 | Mean Test Rewards: 185.8 |\n",
      "| Episode:  70 | Mean Train Rewards: 199.1 | Mean Test Rewards: 263.6 |\n",
      "| Episode:  80 | Mean Train Rewards: 318.2 | Mean Test Rewards: 397.4 |\n",
      "Reached reward threshold in 86 episodes\n"
     ]
    }
   ],
   "source": [
    "MAX_EPISODES = 500\n",
    "DISCOUNT_FACTOR = 0.99\n",
    "TRACE_DECAY = 0.99\n",
    "N_TRIALS = 25\n",
    "REWARD_THRESHOLD = 475\n",
    "PRINT_EVERY = 10\n",
    "\n",
    "train_rewards = []\n",
    "test_rewards = []\n",
    "\n",
    "for episode in range(1, MAX_EPISODES+1):\n",
    "    \n",
    "    policy_loss, value_loss, train_reward = train(train_env, policy, optimizer, DISCOUNT_FACTOR, TRACE_DECAY)\n",
    "    \n",
    "    test_reward = evaluate(test_env, policy)\n",
    "    \n",
    "    train_rewards.append(train_reward)\n",
    "    test_rewards.append(test_reward)\n",
    "    \n",
    "    mean_train_rewards = np.mean(train_rewards[-N_TRIALS:])\n",
    "    mean_test_rewards = np.mean(test_rewards[-N_TRIALS:])\n",
    "    \n",
    "    if episode % PRINT_EVERY == 0:\n",
    "    \n",
    "        print(f'| Episode: {episode:3} | Mean Train Rewards: {mean_train_rewards:5.1f} | Mean Test Rewards: {mean_test_rewards:5.1f} |')\n",
    "    \n",
    "    if mean_test_rewards >= REWARD_THRESHOLD:\n",
    "        \n",
    "        print(f'Reached reward threshold in {episode} episodes')\n",
    "        \n",
    "        break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8))\n",
    "plt.plot(test_rewards, label='Test Reward')\n",
    "plt.plot(train_rewards, label='Train Reward')\n",
    "plt.xlabel('Episode', fontsize=20)\n",
    "plt.ylabel('Reward', fontsize=20)\n",
    "plt.hlines(REWARD_THRESHOLD, 0, len(test_rewards), color='r')\n",
    "plt.legend(loc='lower right')\n",
    "plt.grid()"
   ]
  }
 ],
 "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": 2
}