WDQN_Learner.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "WDQN_Learner.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyMYtci3r2A3bnuA0scOoDAm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 460
        },
        "id": "NOdQXZN_FdmQ",
        "outputId": "5f295ed3-6b99-42b4-e083-661f81de6449"
      },
      "source": [
        "!git clone https://github.com/LukeHackett12/MODRL.git\n",
        "import shutil\n",
        "shutil.move(\"/content/MODRL/custom_envs\", \"/content/custom_envs\")\n",
        "shutil.move(\"/content/MODRL/dqn_utils\", \"/content/dqn_utils\") "
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "fatal: destination path 'MODRL' already exists and is not an empty directory.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "error",
          "ename": "FileNotFoundError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mFileNotFoundError\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m/usr/lib/python3.7/shutil.py\u001b[0m in \u001b[0;36mmove\u001b[0;34m(src, dst, copy_function)\u001b[0m\n\u001b[1;32m    565\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 566\u001b[0;31m         \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrename\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mreal_dst\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    567\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mOSError\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mFileNotFoundError\u001b[0m: [Errno 2] No such file or directory: '/content/MODRL/custom_envs' -> '/content/custom_envs/custom_envs'",
            "\nDuring handling of the above exception, another exception occurred:\n",
            "\u001b[0;31mFileNotFoundError\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-20-1b2e3e8a8ed2>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mget_ipython\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msystem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'git clone https://github.com/LukeHackett12/MODRL.git'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mshutil\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mshutil\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmove\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"/content/MODRL/custom_envs\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"/content/custom_envs\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0mshutil\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmove\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"/content/MODRL/dqn_utils\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"/content/dqn_utils\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/lib/python3.7/shutil.py\u001b[0m in \u001b[0;36mmove\u001b[0;34m(src, dst, copy_function)\u001b[0m\n\u001b[1;32m    578\u001b[0m             \u001b[0mrmtree\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    579\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 580\u001b[0;31m             \u001b[0mcopy_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mreal_dst\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    581\u001b[0m             \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munlink\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    582\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mreal_dst\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/lib/python3.7/shutil.py\u001b[0m in \u001b[0;36mcopy2\u001b[0;34m(src, dst, follow_symlinks)\u001b[0m\n\u001b[1;32m    264\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0misdir\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdst\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    265\u001b[0m         \u001b[0mdst\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mjoin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdst\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbasename\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 266\u001b[0;31m     \u001b[0mcopyfile\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdst\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfollow_symlinks\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mfollow_symlinks\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    267\u001b[0m     \u001b[0mcopystat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdst\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfollow_symlinks\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mfollow_symlinks\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    268\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mdst\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/lib/python3.7/shutil.py\u001b[0m in \u001b[0;36mcopyfile\u001b[0;34m(src, dst, follow_symlinks)\u001b[0m\n\u001b[1;32m    118\u001b[0m         \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msymlink\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mreadlink\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdst\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    119\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 120\u001b[0;31m         \u001b[0;32mwith\u001b[0m \u001b[0mopen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'rb'\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mfsrc\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    121\u001b[0m             \u001b[0;32mwith\u001b[0m \u001b[0mopen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdst\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'wb'\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mfdst\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    122\u001b[0m                 \u001b[0mcopyfileobj\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfsrc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfdst\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mFileNotFoundError\u001b[0m: [Errno 2] No such file or directory: '/content/MODRL/custom_envs'"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eIdtkKVeGC0J",
        "outputId": "b347abb5-833a-4753-e282-1e8bda5b08fe"
      },
      "source": [
        "!pip install pyvirtualdisplay\n",
        "!pip install gym --upgrade\n",
        "!pip install piglet\n",
        "!pip install pygame \n",
        "\n",
        "!apt-get install python-opengl -y\n",
        "!apt install xvfb -y"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: pyvirtualdisplay in /usr/local/lib/python3.7/dist-packages (2.1)\n",
            "Requirement already satisfied: EasyProcess in /usr/local/lib/python3.7/dist-packages (from pyvirtualdisplay) (0.3)\n",
            "Requirement already up-to-date: gym in /usr/local/lib/python3.7/dist-packages (0.18.0)\n",
            "Requirement already satisfied, skipping upgrade: numpy>=1.10.4 in /usr/local/lib/python3.7/dist-packages (from gym) (1.19.5)\n",
            "Requirement already satisfied, skipping upgrade: scipy in /usr/local/lib/python3.7/dist-packages (from gym) (1.4.1)\n",
            "Requirement already satisfied, skipping upgrade: cloudpickle<1.7.0,>=1.2.0 in /usr/local/lib/python3.7/dist-packages (from gym) (1.3.0)\n",
            "Requirement already satisfied, skipping upgrade: pyglet<=1.5.0,>=1.4.0 in /usr/local/lib/python3.7/dist-packages (from gym) (1.5.0)\n",
            "Requirement already satisfied, skipping upgrade: Pillow<=7.2.0 in /usr/local/lib/python3.7/dist-packages (from gym) (7.0.0)\n",
            "Requirement already satisfied, skipping upgrade: future in /usr/local/lib/python3.7/dist-packages (from pyglet<=1.5.0,>=1.4.0->gym) (0.16.0)\n",
            "Requirement already satisfied: piglet in /usr/local/lib/python3.7/dist-packages (1.0.0)\n",
            "Requirement already satisfied: piglet-templates in /usr/local/lib/python3.7/dist-packages (from piglet) (1.1.0)\n",
            "Requirement already satisfied: astunparse in /usr/local/lib/python3.7/dist-packages (from piglet-templates->piglet) (1.6.3)\n",
            "Requirement already satisfied: Parsley in /usr/local/lib/python3.7/dist-packages (from piglet-templates->piglet) (1.3)\n",
            "Requirement already satisfied: markupsafe in /usr/local/lib/python3.7/dist-packages (from piglet-templates->piglet) (1.1.1)\n",
            "Requirement already satisfied: attrs in /usr/local/lib/python3.7/dist-packages (from piglet-templates->piglet) (20.3.0)\n",
            "Requirement already satisfied: six<2.0,>=1.6.1 in /usr/local/lib/python3.7/dist-packages (from astunparse->piglet-templates->piglet) (1.15.0)\n",
            "Requirement already satisfied: wheel<1.0,>=0.23.0 in /usr/local/lib/python3.7/dist-packages (from astunparse->piglet-templates->piglet) (0.36.2)\n",
            "Requirement already satisfied: pygame in /usr/local/lib/python3.7/dist-packages (2.0.1)\n",
            "Reading package lists... Done\n",
            "Building dependency tree       \n",
            "Reading state information... Done\n",
            "python-opengl is already the newest version (3.1.0+dfsg-1).\n",
            "0 upgraded, 0 newly installed, 0 to remove and 29 not upgraded.\n",
            "Reading package lists... Done\n",
            "Building dependency tree       \n",
            "Reading state information... Done\n",
            "xvfb is already the newest version (2:1.19.6-1ubuntu4.8).\n",
            "0 upgraded, 0 newly installed, 0 to remove and 29 not upgraded.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "O1SPmPLTGGbX"
      },
      "source": [
        "%matplotlib inline\n",
        "import matplotlib.pyplot as plt\n",
        "import matplotlib\n",
        "import matplotlib.pylab as pl\n",
        "plt.rcParams['figure.figsize'] = [10, 5]\n",
        "\n",
        "import numpy as np\n",
        "from typing import NamedTuple\n",
        "from collections import deque\n",
        "from copy import deepcopy\n",
        "import random\n",
        "import sys\n",
        "\n",
        "from numpy.core.numeric import NaN\n",
        "from custom_envs.mountain_car.engine import MountainCar\n",
        "\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras, Tensor\n",
        "from keras import backend as K\n",
        "\n",
        "# set up matplotlib\n",
        "is_ipython = 'inline' in matplotlib.get_backend()\n",
        "if is_ipython:\n",
        "    from IPython import display"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zUA1fsgsGWAP"
      },
      "source": [
        "class Transition(NamedTuple):\n",
        "    currStates: Tensor\n",
        "    actions: Tensor\n",
        "    policy: Tensor\n",
        "    rewards: Tensor\n",
        "    nextStates: Tensor\n",
        "    dones: Tensor\n",
        "\n",
        "\n",
        "\n",
        "class DQNAgent(object):\n",
        "    def __init__(self, stateShape, actionSpace, numPicks, memorySize, numRewards):\n",
        "        self.numPicks = numPicks\n",
        "        self.memorySize = memorySize\n",
        "        self.replayMemory = deque(maxlen=memorySize)\n",
        "        self.stateShape = stateShape\n",
        "        self.actionSpace = actionSpace\n",
        "        self.numRewards = numRewards\n",
        "\n",
        "        self.step = 0\n",
        "        self.sync = 200\n",
        "\n",
        "        self.alpha = 0.001\n",
        "        self.walpha = 0.001\n",
        "        self.delay = 1\n",
        "        self.epsilon = 1\n",
        "        self.epsilon_decay = 0.05\n",
        "        self.epsilon_min = 0.001\n",
        "        self.eps_threshold = 0\n",
        "\n",
        "        self.gamma = 0.9\n",
        "        self.tau = 0.01\n",
        "\n",
        "        self.train_network = self.createNetwork(stateShape, len(actionSpace), self.alpha)\n",
        "        self.target_network = self.createNetwork(stateShape, len(actionSpace), self.alpha)\n",
        "\n",
        "        # Store network weights for each policy\n",
        "        self.policy_train_weights = [deepcopy(self.train_network.get_weights())] * self.numRewards\n",
        "        self.policy_target_weights = [deepcopy(self.train_network.get_weights())] * self.numRewards\n",
        "\n",
        "        # Create and store network weights for W-values\n",
        "        self.w_network = self.createNetwork(stateShape, numRewards, self.alpha)\n",
        "        self.wnet_weights = [deepcopy(self.w_network.get_weights())] * self.numRewards\n",
        "\n",
        "    def createNetwork(self, n_input, n_output, learningRate):\n",
        "        model = keras.models.Sequential()\n",
        "\n",
        "        model.add(keras.layers.Dense(24, activation='relu', input_shape=n_input))\n",
        "        model.add(keras.layers.Dense(48, activation='relu'))\n",
        "        model.add(keras.layers.Dense(n_output, activation='linear'))\n",
        "        model.compile(loss='mse', optimizer=keras.optimizers.Adam(lr=learningRate))\n",
        "        return model\n",
        "\n",
        "    def trainDQN(self):\n",
        "        if len(self.replayMemory) <= self.numPicks:\n",
        "            return [(0, 0)] * self.numRewards\n",
        "\n",
        "        samples = random.sample(self.replayMemory, self.numPicks)\n",
        "        batch = Transition(*zip(*samples))\n",
        "        currStates, actions, policies, rewards, nextStates, dones = batch\n",
        "\n",
        "        currStates = np.squeeze(np.array(currStates), 1)\n",
        "        nextStates = np.squeeze(np.array(nextStates), 1)\n",
        "\n",
        "        rewards = np.array(rewards).transpose().astype(float)\n",
        "        actions = np.array(actions).reshape(self.numPicks,).astype(int)\n",
        "        policies = np.array(policies).reshape(self.numPicks,).astype(int)\n",
        "\n",
        "        dones = np.array(dones).astype(bool)\n",
        "        notDones = (~dones).astype(float)\n",
        "        dones = dones.astype(float)\n",
        "\n",
        "        agentsLoss = []\n",
        "\n",
        "        for i, reward in enumerate(rewards):\n",
        "            self.train_network.set_weights(self.policy_train_weights[i])\n",
        "            self.target_network.set_weights(self.policy_target_weights[i])\n",
        "\n",
        "            Q_currents_all = self.train_network(currStates, training=False).numpy()\n",
        "            Q_futures_all = self.target_network(nextStates, training=False).numpy().max(axis=1)\n",
        "\n",
        "            Q_currents = np.copy(Q_currents_all)\n",
        "            Q_futures = np.copy(Q_futures_all)\n",
        "\n",
        "            reward = np.array(reward).reshape(self.numPicks,).astype(float)\n",
        "            # Q-Learning\n",
        "            Q_currents[np.arange(self.numPicks), actions] = reward * dones + (reward + Q_futures * self.gamma)*notDones\n",
        "            lossQ = self.train_network.train_on_batch(currStates,Q_currents)\n",
        "            self.policy_train_weights[i] = deepcopy(self.train_network.get_weights())\n",
        "\n",
        "            lossW = 0\n",
        "            #Leave in exploration actions for now, can remove with \"policy[p] != -1\"\n",
        "            inverted_policy_mask = np.array([p for p in range(self.numPicks) if policies[p] != i])\n",
        "            if len(inverted_policy_mask) > 0:\n",
        "                # W-Learning\n",
        "                self.w_network.set_weights(self.wnet_weights[i])\n",
        "                currStatesNP = currStates[inverted_policy_mask]\n",
        "                policiesNP = policies[inverted_policy_mask]\n",
        "                rewardNP = reward[inverted_policy_mask]\n",
        "                donesNP = dones[inverted_policy_mask]\n",
        "                notDonesNP = notDones[inverted_policy_mask]\n",
        "\n",
        "                Q_currents_np = Q_currents_all[inverted_policy_mask].max(axis=1)\n",
        "                Q_futures_np = Q_futures_all[inverted_policy_mask]\n",
        "\n",
        "                w_targets = self.w_network(currStatesNP, training=False).numpy()\n",
        "\n",
        "                # maybe (Q_currents_not_policy - ((rewardNP * donesNP) + (self.gamma * Q_futures_not_policy) * notDonesNP)) * walpha^delay ?\n",
        "                w_targets[np.arange(len(inverted_policy_mask)), policiesNP] = ((1-self.walpha) * w_targets[np.arange(len(inverted_policy_mask)), policiesNP]) + ((self.walpha**self.delay) * (Q_currents_np - ((rewardNP * donesNP) + (self.gamma * Q_futures_np) * notDonesNP)))\n",
        "                lossW = self.w_network.train_on_batch(currStatesNP, w_targets)\n",
        "                self.wnet_weights[i] = deepcopy(self.w_network.get_weights())\n",
        "            \n",
        "            agentsLoss.append((lossQ, lossW))\n",
        "\n",
        "        return agentsLoss\n",
        "\n",
        "    def selectAction(self, state):\n",
        "        self.step += 1\n",
        "\n",
        "        if self.step % self.sync == 0:\n",
        "            self.policy_target_weights = deepcopy(self.policy_train_weights)\n",
        "            self.epsilon = max(self.epsilon-self.epsilon_decay, self.epsilon_min)\n",
        "\n",
        "        emptyPolicies = [0] * self.numRewards\n",
        "        policy, qs, ws = (-1, emptyPolicies, emptyPolicies)\n",
        "        random = True\n",
        "        if np.random.rand(1) < self.epsilon:\n",
        "            action = np.random.randint(0, 3)\n",
        "        else:\n",
        "            ws = []\n",
        "            actions = []\n",
        "            preds = []\n",
        "            for i in range(self.numRewards):\n",
        "                self.train_network.set_weights(self.policy_train_weights[i])\n",
        "\n",
        "                pred = np.squeeze(self.train_network(state, training=False).numpy(), axis=0)\n",
        "                preds.append(pred)\n",
        "\n",
        "                action = np.argmax(pred)\n",
        "                actions.append(action)\n",
        "\n",
        "                self.w_network.set_weights(self.wnet_weights[i])\n",
        "                w_val = self.w_network(state, training=False).numpy()[0]\n",
        "                ws.append(w_val[np.argmax(w_val)])\n",
        "\n",
        "            policy = np.argmax(ws)\n",
        "            action = actions[policy]\n",
        "            qs = preds[policy]\n",
        "            random = False\n",
        "\n",
        "        return action, policy, qs, ws, random\n",
        "\n",
        "    def addMemory(self, memory):\n",
        "        self.replayMemory.append(memory)\n",
        "\n",
        "    def save(self):\n",
        "        save_path = (\n",
        "            f\"./mountain_car_wnet_{int(self.step)}.chkpt\"\n",
        "        )\n",
        "        #TODO\n",
        "        print(f\"MountainNet saved to {save_path} done!\")\n"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lxEr82uBGiRi"
      },
      "source": [
        "class MultiObjectiveWMountainCar(object):\n",
        "    def __init__(self, episodes):\n",
        "        self.current_episode = 0\n",
        "        self.episodes = episodes\n",
        "\n",
        "        self.episode_score = []\n",
        "        self.episode_qs = []\n",
        "        self.episode_height = []\n",
        "        self.episode_loss = []\n",
        "        self.episode_ws = []\n",
        "        self.episode_policies = []\n",
        "\n",
        "        self.fig, self.ax = plt.subplots(2, 3)\n",
        "        self.fig.tight_layout()\n",
        "        self.fig.canvas.draw()\n",
        "        plt.show(block=False)\n",
        "\n",
        "        self.numRewards = 3\n",
        "\n",
        "        self.env = MountainCar(speed=1e8, graphical_state=False,\n",
        "                               render=False, is_debug=False, random_starts=True)\n",
        "        self.agent = DQNAgent(stateShape=(\n",
        "            2,), actionSpace=self.env.get_action_space(), numPicks=32, memorySize=10000, numRewards=self.numRewards)\n",
        "\n",
        "    def train(self):\n",
        "        for _ in range(self.episodes):\n",
        "            self.episode()\n",
        "            self.current_episode += 1\n",
        "\n",
        "        self.env.close()\n",
        "\n",
        "    def episode(self):\n",
        "        done = False\n",
        "        rewardsSum = 0\n",
        "        \n",
        "        lossSums = [0] * (self.numRewards)\n",
        "        policies = [0] * (self.numRewards)\n",
        "        qSums = [0] * (self.numRewards)\n",
        "        wSums = [0] * (self.numRewards)\n",
        "        actions = 1\n",
        "\n",
        "        state = self.env.reset().reshape(1, 2)\n",
        "        maxHeight = -1\n",
        "\n",
        "        while not done:\n",
        "            action, policy, qs, ws, random = self.agent.selectAction(state)\n",
        "            if not random:\n",
        "                policies[policy] += 1\n",
        "                qSums = [qSums[i] + qs[i] for i in range(len(policies))]\n",
        "                wSums = [wSums[i] + ws[i] for i in range(len(policies))]\n",
        "                actions += 1\n",
        "\n",
        "            #print(\"{} action taken by {} policy, q-val: {}, w-val: {}\".format(Action(action).name, Policy(policy).name,q,w))\n",
        "\n",
        "            obs, reward, done, total_score = self.env.step_all(action)\n",
        "\n",
        "            maxHeight = max(obs[0], maxHeight)\n",
        "            if obs[0] >= 0.5:\n",
        "                reward = [r+10 for r in reward]\n",
        "\n",
        "            nextState = obs.reshape(1, 2)\n",
        "            rewardsSum = np.add(rewardsSum, sum(reward))\n",
        "\n",
        "            loss = self.agent.trainDQN()\n",
        "            self.agent.addMemory((state, action, policy, reward, nextState, done))\n",
        "            state = nextState\n",
        "            lossSums = [lossSums[i] + loss[i][0] for i in range(len(policies))]\n",
        "\n",
        "        if (total_score < 200) and self.current_episode > 150:\n",
        "            self.agent.save()\n",
        "\n",
        "        self.episode_score.append(rewardsSum)\n",
        "        self.episode_height.append(maxHeight)\n",
        "        self.episode_loss.append(lossSums)\n",
        "        self.episode_policies.append(policies)\n",
        "        self.episode_qs.append([qSum/actions for qSum in qSums])\n",
        "        self.episode_ws.append([wSum/actions for wSum in wSums])\n",
        "        self.plot()\n",
        "\n",
        "        print(\"now epsilon is {}, the reward is {} with loss {} in episode {}\".format(\n",
        "            self.agent.epsilon, rewardsSum, lossSums, self.current_episode))\n",
        "        print(\"Report: \\nrewardSum:{}\\nheight:{}\\nloss:{}\\npolicies:{}\\nqAverage:{}\\nws:{}\".format(self.episode_score[-1],\n",
        "                                                                                                  self.episode_height[-1],\n",
        "                                                                                                  self.episode_loss[-1],\n",
        "                                                                                                  self.episode_policies[-1],\n",
        "                                                                                                  self.episode_qs[-1],\n",
        "                                                                                                  self.episode_ws[-1]))\n",
        "\n",
        "\n",
        "    def plot(self):\n",
        "        plt.figure(2)\n",
        "        self.fig, self.ax = plt.subplots(2, 3)\n",
        "        self.fig.tight_layout()\n",
        "        self.fig.canvas.draw()\n",
        "\n",
        "        self.ax[0][0].title.set_text('Score')\n",
        "        self.ax[0][0].plot(self.episode_score, 'b')\n",
        "\n",
        "        self.ax[0][1].title.set_text('Height')\n",
        "        self.ax[0][1].plot(self.episode_height, 'g')\n",
        "\n",
        "        policies = np.transpose(self.episode_policies)\n",
        "        colors = pl.cm.jet(np.linspace(0, 1, len(policies)))\n",
        "\n",
        "        self.ax[0][2].clear()\n",
        "        self.ax[0][2].title.set_text('Loss')\n",
        "        losses = np.transpose(self.episode_loss)\n",
        "        for i, loss in enumerate(losses):\n",
        "            self.ax[0][2].plot(loss, color=colors[i], label=i)\n",
        "        self.ax[0][2].legend()\n",
        "\n",
        "        self.ax[1][0].clear()\n",
        "        self.ax[1][0].title.set_text('Policy Choices')\n",
        "        for i, policy in enumerate(policies):\n",
        "            self.ax[1][0].plot(policy, color=colors[i], label=i)\n",
        "        self.ax[1][0].legend()\n",
        "\n",
        "        self.ax[1][1].clear()\n",
        "        self.ax[1][1].title.set_text('Q-Val')\n",
        "        qs = np.transpose(self.episode_qs)\n",
        "        for i, q in enumerate(qs):\n",
        "            self.ax[1][1].plot(q, color=colors[i], label=i)\n",
        "        self.ax[1][1].legend()\n",
        "\n",
        "        self.ax[1][2].clear()\n",
        "        self.ax[1][2].title.set_text('W-Val')\n",
        "        ws = np.transpose(self.episode_ws)\n",
        "        for i, w in enumerate(ws):\n",
        "            self.ax[1][2].plot(w, color=colors[i], label=i)\n",
        "        self.ax[1][2].legend()\n",
        "\n",
        "        display.display(plt.gcf())\n",
        "        display.clear_output(wait=True)\n",
        "        plt.pause(0.001)  # pause a bit so that plots are updated\n"
      ],
      "execution_count": 24,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 517
        },
        "id": "88D-Jp8nGoTW",
        "outputId": "be43d2ec-88ea-4bd1-81f5-c1ec32f6ceeb"
      },
      "source": [
        "agent = MultiObjectiveWMountainCar(2000)\n",
        "agent.train()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x360 with 0 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 6 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Report: \n",
            "rewardSum:-337\n",
            "qAverage:0.1287893347081638\n",
            "height:-0.44639075691781427\n",
            "loss:35.41157548909541\n",
            "policies:[0, 12, 111, 79]\n",
            "dones:0.004508393179182143\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NpwG0oA7PH_n"
      },
      "source": [
        "from google.colab import files\n",
        "!zip -r /content/mountain_car_tfngmo_99905.chkpt.zip /content/mountain_car_tfngmo_99905.chkpt\n",
        "files.download('mountain_car_tfngmo_99905.chkpt.zip') "
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}