iqn.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "iqn.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "metadata": {
        "id": "bhZ3n3TxXyot",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Part of https://github.com/htdt/cartpole-solved"
      ]
    },
    {
      "metadata": {
        "id": "6Dl1Vqz_uZ1Y",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "!pip install torch_nightly -f https://download.pytorch.org/whl/nightly/cpu/torch_nightly.html\n",
        "!pip install -q gym ipdb"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "fZg7tVEP6sGj",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from collections import deque\n",
        "import gym, random, math, ipdb\n",
        "from tqdm import trange\n",
        "import numpy as np\n",
        "\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.optim as optim\n",
        "\n",
        "gym.logger.set_level(40)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "04xWUEGXFoS1",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "hugvEbVZ16wR",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "class DQN(nn.Module):\n",
        "  input_dim = 4\n",
        "  output_dim = 2\n",
        "  hidden = 32\n",
        "  quantile_dim = 8 # embedding dim\n",
        "\n",
        "  def __init__(self):\n",
        "    super(DQN, self).__init__()\n",
        "    fc = lambda s_in, s_out: nn.Sequential(nn.Linear(s_in, s_out), nn.ReLU())\n",
        "    self.features = fc(self.input_dim, self.hidden)\n",
        "    self.quantile_fc = fc(self.quantile_dim, self.hidden)\n",
        "    self.head_fc = fc(self.hidden, self.hidden)\n",
        "    self.adv = nn.Linear(self.hidden, self.output_dim)\n",
        "    self.val = nn.Linear(self.hidden, 1)\n",
        "\n",
        "  def forward(self, x, num_quantiles):\n",
        "    batch_size = x.shape[0]\n",
        "    x = self.features(x) # batch x hidden\n",
        "    x = x.repeat(num_quantiles, 1) # batch * num_quantiles x hidden\n",
        "    qn, quantiles = self._get_quantile(batch_size * num_quantiles) # batch * num_quantiles x quantile_dim\n",
        "    qn = self.quantile_fc(qn) # batch * num_quantiles x hidden\n",
        "    x = x * qn\n",
        "    x = self.head_fc(x)\n",
        "    adv = self.adv(x)\n",
        "    val = self.val(x)\n",
        "    q = val + adv - adv.mean(1, keepdim=True)\n",
        "    return q, quantiles\n",
        "\n",
        "  def _get_quantile(self, size):\n",
        "    quantiles = torch.empty(size, 1, dtype=torch.float32).uniform_()\n",
        "    quantile_net = quantiles.repeat(1, self.quantile_dim)\n",
        "    embedding_range = torch.arange(1, self.quantile_dim + 1, dtype=torch.float32).unsqueeze(0)\n",
        "    return torch.cos(embedding_range * math.pi * quantile_net), quantiles\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "6EKVMKR__LpT",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "class Agent:\n",
        "  output_dim = 2\n",
        "  lr = 1e-3\n",
        "  batch_size = 32\n",
        "  gamma = 0.99\n",
        "\n",
        "  kappa = 1  # Huber loss cutoff\n",
        "  num_t = 8  # number of online quantile samples for loss estimation\n",
        "  num_tp = 8 # number of target quantile samples for loss estimation\n",
        "  num_q = 8  # number of quantile samples for computing Q-values\n",
        "\n",
        "  def __init__(self):\n",
        "    self.online_net = DQN()\n",
        "    self.online_net.train()\n",
        "\n",
        "    self.target_net = DQN()\n",
        "    self.update_target_net()\n",
        "    for param in self.target_net.parameters(): param.requires_grad = False\n",
        "\n",
        "    self.optimiser = optim.Adam(self.online_net.parameters(), lr=self.lr)\n",
        "\n",
        "  def act(self, state):\n",
        "    state = torch.FloatTensor(state).unsqueeze(0)\n",
        "    with torch.no_grad():\n",
        "      q, _ = self.online_net(state, self.num_q)\n",
        "      return q.mean(0).argmax(0).item()\n",
        "\n",
        "  def act_e_greedy(self, state, epsilon=0.01):\n",
        "    if random.random() < epsilon:\n",
        "      return random.randrange(self.output_dim)\n",
        "    else:\n",
        "      return self.act(state)\n",
        "\n",
        "  def _target_quantile_values(self, reward, next_state, terminal):\n",
        "    with torch.no_grad():\n",
        "      reward = reward.repeat(self.num_tp)\n",
        "      gamma_with_terminal = (self.gamma * (1 - terminal)).repeat(self.num_tp)\n",
        "\n",
        "      # for action selection for the next states\n",
        "      action_q, _ = self.online_net(next_state, self.num_q) # online_net to get action for next_state\n",
        "      action_q = action_q.view(self.num_q, self.batch_size, self.output_dim).mean(0)\n",
        "      actions = action_q.argmax(1) # batch_size\n",
        "      actions = actions.repeat(self.num_tp)\n",
        "\n",
        "      target_q, _ = self.target_net(next_state, self.num_tp) # target_net to get Q\n",
        "      target_q = target_q[range(self.batch_size * self.num_tp), actions]\n",
        "\n",
        "      return reward + gamma_with_terminal * target_q\n",
        "\n",
        "  def _sample_batch(self, buffer):\n",
        "    s, a, r, ns, t = zip(*random.sample(buffer, self.batch_size))\n",
        "    f, l = torch.FloatTensor, torch.LongTensor\n",
        "    return f(s), l(a), f(r), f(ns), f(t)\n",
        "  \n",
        "  def _huber_loss(self, bellman_errors):\n",
        "    be_abs = bellman_errors.abs()\n",
        "    huber_loss_1 = (be_abs <= self.kappa).float() * 0.5 * bellman_errors ** 2\n",
        "    huber_loss_2 = (be_abs > self.kappa).float() * self.kappa * (be_abs - 0.5 * self.kappa)\n",
        "    return huber_loss_1 + huber_loss_2\n",
        "  \n",
        "  def _same_shape(self, q_value, target_qv, quantiles): # batch x num_tp x num_t\n",
        "    return q_value.view(self.num_t, self.batch_size).t().unsqueeze(1),\\\n",
        "      target_qv.view(self.num_tp, self.batch_size).t().unsqueeze(2),\\\n",
        "      quantiles.view(self.num_t, self.batch_size).t().unsqueeze(1).repeat(1, self.num_tp, 1)\n",
        "\n",
        "  def train_iter(self, buffer):\n",
        "    state, action, reward, next_state, terminal = self._sample_batch(buffer)\n",
        "    \n",
        "    q_values, quantiles = self.online_net(state, self.num_t)\n",
        "    action = action.repeat(self.num_t)\n",
        "    q_value = q_values[range(self.batch_size * self.num_t), action]\n",
        "\n",
        "    target_qv = self._target_quantile_values(reward, next_state, terminal)\n",
        "\n",
        "    q_value, target_qv, quantiles = self._same_shape(q_value, target_qv, quantiles)\n",
        "\n",
        "    bellman_errors = target_qv - q_value\n",
        "    huber_loss = self._huber_loss(bellman_errors)\n",
        "    neg_errors = (bellman_errors.detach() < 0).float()\n",
        "    quantile_huber_loss = (quantiles - neg_errors).abs() * huber_loss / self.kappa\n",
        "    loss = quantile_huber_loss.sum(2).mean()\n",
        "\n",
        "    self.optimiser.zero_grad()\n",
        "    loss.backward()\n",
        "    self.optimiser.step()\n",
        "    return loss.item()\n",
        "\n",
        "  def update_target_net(self):\n",
        "    self.target_net.load_state_dict(self.online_net.state_dict())"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "2Urg4wcU-pof",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 354
        },
        "outputId": "b4574443-a1c8-4435-de5e-c1483da6a1b4"
      },
      "cell_type": "code",
      "source": [
        "buffer = deque(maxlen=5000)\n",
        "env = gym.make(\"CartPole-v0\")\n",
        "agent = Agent()\n",
        "\n",
        "rewards, loss = [], []\n",
        "episode_reward = 0\n",
        "state = env.reset()\n",
        "\n",
        "for i in trange(10000):\n",
        "  eps = .01 + .99 * math.exp(-i / 500) # close to .01 at 3500\n",
        "  action = agent.act_e_greedy(state, eps)\n",
        "  next_state, reward, done, _ = env.step(action)\n",
        "  buffer.append((state, action, reward, next_state, done))\n",
        "  \n",
        "  state = next_state\n",
        "  episode_reward += reward\n",
        "  \n",
        "  if done:\n",
        "    state = env.reset()\n",
        "    rewards.append(episode_reward)\n",
        "    episode_reward = 0\n",
        "    \n",
        "  if i > 100:\n",
        "    loss.append(agent.train_iter(buffer))\n",
        "   \n",
        "  if i > 100 and i % 100 == 0:\n",
        "    agent.update_target_net()\n",
        "\n",
        "plt.figure(figsize=(20,5))\n",
        "plt.subplot(131)\n",
        "plt.plot(rewards)\n",
        "plt.subplot(132)\n",
        "plt.plot(loss)"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 10000/10000 [01:37<00:00, 102.29it/s]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7f9b3a6ff828>]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAvoAAAEvCAYAAAA0BGXUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f9b479694a8>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "sgDb0LJMvyf1",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}