EDL_12_6_Instinctual_DQN_GYMs.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "EDL_12_6_Instinctual_DQN_GYMs.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyN8pxjMgY8u4rfOMfMRgfpR",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/cxbxmxcx/EvolutionaryDeepLearning/blob/main/EDL_12_6_Instinctual_DQN_GYMs.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install livelossplot --quiet\n",
        "!pip install deap --quiet"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AP16ZeA2b5Gn",
        "outputId": "3c6daae7-b6c8-4427-ff10-d737cd309a29"
      },
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[K     |████████████████████████████████| 793 kB 3.1 MB/s \n",
            "\u001b[K     |████████████████████████████████| 381 kB 21.9 MB/s \n",
            "\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n",
            "jupyter-console 5.2.0 requires prompt-toolkit<2.0.0,>=1.0.0, but you have prompt-toolkit 3.0.30 which is incompatible.\n",
            "google-colab 1.0.0 requires ipython~=5.5.0, but you have ipython 7.34.0 which is incompatible.\u001b[0m\n",
            "\u001b[K     |████████████████████████████████| 160 kB 8.5 MB/s \n",
            "\u001b[?25h"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!apt-get install -y xvfb x11-utils\n",
        "# install required python dependencies\n",
        "!pip install gym[box2d]==0.17.* \\\n",
        "            pyvirtualdisplay==0.2.* \\\n",
        "            PyOpenGL==3.1.* \\\n",
        "            PyOpenGL-accelerate==3.1.* \\\n",
        "            mediapy \\\n",
        "            piglet -q"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GK0A_Lj6Qi5l",
        "outputId": "0ec126d9-4bb1-4d34-91ca-1f47a3143c20"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Reading package lists... Done\n",
            "Building dependency tree       \n",
            "Reading state information... Done\n",
            "The following package was automatically installed and is no longer required:\n",
            "  libnvidia-common-460\n",
            "Use 'apt autoremove' to remove it.\n",
            "The following additional packages will be installed:\n",
            "  libxxf86dga1\n",
            "Suggested packages:\n",
            "  mesa-utils\n",
            "The following NEW packages will be installed:\n",
            "  libxxf86dga1 x11-utils xvfb\n",
            "0 upgraded, 3 newly installed, 0 to remove and 49 not upgraded.\n",
            "Need to get 994 kB of archives.\n",
            "After this operation, 2,982 kB of additional disk space will be used.\n",
            "Get:1 http://archive.ubuntu.com/ubuntu bionic/main amd64 libxxf86dga1 amd64 2:1.1.4-1 [13.7 kB]\n",
            "Get:2 http://archive.ubuntu.com/ubuntu bionic/main amd64 x11-utils amd64 7.7+3build1 [196 kB]\n",
            "Get:3 http://archive.ubuntu.com/ubuntu bionic-updates/universe amd64 xvfb amd64 2:1.19.6-1ubuntu4.11 [785 kB]\n",
            "Fetched 994 kB in 1s (1,368 kB/s)\n",
            "Selecting previously unselected package libxxf86dga1:amd64.\n",
            "(Reading database ... 155653 files and directories currently installed.)\n",
            "Preparing to unpack .../libxxf86dga1_2%3a1.1.4-1_amd64.deb ...\n",
            "Unpacking libxxf86dga1:amd64 (2:1.1.4-1) ...\n",
            "Selecting previously unselected package x11-utils.\n",
            "Preparing to unpack .../x11-utils_7.7+3build1_amd64.deb ...\n",
            "Unpacking x11-utils (7.7+3build1) ...\n",
            "Selecting previously unselected package xvfb.\n",
            "Preparing to unpack .../xvfb_2%3a1.19.6-1ubuntu4.11_amd64.deb ...\n",
            "Unpacking xvfb (2:1.19.6-1ubuntu4.11) ...\n",
            "Setting up xvfb (2:1.19.6-1ubuntu4.11) ...\n",
            "Setting up libxxf86dga1:amd64 (2:1.1.4-1) ...\n",
            "Setting up x11-utils (7.7+3build1) ...\n",
            "Processing triggers for man-db (2.8.3-2ubuntu0.1) ...\n",
            "Processing triggers for libc-bin (2.27-3ubuntu1.3) ...\n",
            "/sbin/ldconfig.real: /usr/local/lib/python3.7/dist-packages/ideep4py/lib/libmkldnn.so.0 is not a symbolic link\n",
            "\n",
            "\u001b[K     |████████████████████████████████| 538 kB 8.8 MB/s \n",
            "\u001b[K     |████████████████████████████████| 448 kB 42.2 MB/s \n",
            "\u001b[K     |████████████████████████████████| 67 kB 5.1 MB/s \n",
            "\u001b[?25h  Building wheel for PyOpenGL-accelerate (setup.py) ... \u001b[?25l\u001b[?25hdone\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!echo $DISPLAY"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "n6uda62nTq-2",
        "outputId": "a42dbeb5-de4f-42d6-bc2e-6bcbd164f1d0"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from pyvirtualdisplay import Display\n",
        "display = Display(visible=0, size=(1400, 900))\n",
        "display.start()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jpWForNjXesx",
        "outputId": "e11f21a9-0b0d-44f4-9374-a442e8f25d88"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<Display cmd_param=['Xvfb', '-br', '-nolisten', 'tcp', '-screen', '0', '1400x900x24', ':1001'] cmd=['Xvfb', '-br', '-nolisten', 'tcp', '-screen', '0', '1400x900x24', ':1001'] oserror=None return_code=None stdout=\"None\" stderr=\"None\" timeout_happened=False>"
            ]
          },
          "metadata": {},
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import gym.wrappers\n",
        "import matplotlib.pyplot as plt\n",
        "import os\n",
        "import pickle\n",
        "import numpy as np\n",
        "import mediapy \n",
        "import random\n",
        "from collections import deque\n",
        "from livelossplot import PlotLosses\n",
        "\n",
        "import array\n",
        "\n",
        "#DEAP\n",
        "from deap import algorithms\n",
        "from deap import base\n",
        "from deap import benchmarks\n",
        "from deap import creator\n",
        "from deap import tools"
      ],
      "metadata": {
        "id": "aGu3z6Q5oDn4"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Setup Environment { run: \"auto\" }\n",
        "ENVIRONMENT = \"LunarLander-v2\" #@param [\"CartPole-v1\", \"Acrobot-v1\", \"CubeCrash-v0\", \"MountainCar-v0\", \"LunarLander-v2\"]\n",
        "\n",
        "env = gym.make(ENVIRONMENT)\n",
        "\n",
        "state = env.reset()\n",
        "plt.imshow(env.render(mode='rgb_array'))\n",
        "\n",
        "print(\"action space: {0!r}\".format(env.action_space))\n",
        "print(\"observation space: {0!r}\".format(env.observation_space))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 304
        },
        "id": "_Qzcx4MpWjtd",
        "outputId": "f98fbe1a-47ff-4d8b-8cb7-7bc915d2d219"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "action space: Discrete(4)\n",
            "observation space: Box(-inf, inf, (8,), float32)\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Siumlate Environment Run { run: \"auto\" }\n",
        "ENVIRONMENT = \"CartPole-v1\" #@param [\"CartPole-v1\", \"Acrobot-v1\", \"MountainCar-v0\", \"LunarLander-v2\"]\n",
        "SIMULATION_RUNS = 10 #@param {type:\"slider\", min:1, max:10, step:1}\n",
        "SIMULATION_ITERATIONS = 200 #@param {type:\"slider\", min:50, max:200, step:1}\n",
        "\n",
        "env = gym.make(ENVIRONMENT)\n",
        "\n",
        "fitnesses = []\n",
        "frames = []\n",
        "\n",
        "for run in range(SIMULATION_RUNS):  \n",
        "  state = env.reset()\n",
        "  fitness = 0\n",
        "  for i in range(SIMULATION_ITERATIONS):\n",
        "    action = env.action_space.sample()\n",
        "    state, reward, done, info = env.step(np.argmax(action))  \n",
        "    frames.append(env.render(mode='rgb_array'))  \n",
        "    fitness += reward       \n",
        "    if done: \n",
        "      fitnesses.append(fitness) \n",
        "      break \n",
        "\n",
        "mediapy.show_video(frames, fps=30)\n",
        "print(fitnesses)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 441
        },
        "id": "OijWAhygyaQi",
        "outputId": "6046f8e3-911d-4dbe-87e9-b175267da4f0"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "<table class=\"show_videos\" style=\"border-spacing:0px;\"><tr><td style=\"padding:1px;\"><video controls width=\"600\" height=\"400\" style=\"object-fit:cover;\" loop autoplay>\n",
              "      <source src=\"data:video/mp4;base64,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\" type=\"video/mp4\"/>\n",
              "      This browser does not support the video tag.\n",
              "      </video></td></tr></table>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[10.0, 10.0, 10.0, 9.0, 9.0, 10.0, 9.0, 9.0, 10.0, 9.0]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow.keras as k\n",
        "import tensorflow.keras.layers as kl\n",
        "\n",
        "class DQNAgent():\n",
        "    def __init__(self, instincts, state_size, action_size, memory):        \n",
        "        self.state_size         = state_size\n",
        "        self.action_size        = action_size\n",
        "        self.memory             = memory\n",
        "        self.learning_rate      = 0.001\n",
        "        self.gamma              = 0.95              \n",
        "        self.brain              = self._build_model(instincts)\n",
        "\n",
        "    def _build_model(self, instincts):\n",
        "        # Neural Net for Deep-Q learning Model \n",
        "        inputs = k.Input(shape=(self.state_size,))\n",
        "        dense_in = instincts[0]\n",
        "        dense1 = instincts[1]\n",
        "        dense2 = instincts[2]\n",
        "        dense3 = instincts[3]\n",
        "        dense_out = instincts[4] # kl.Dense(self.action_size, activation='linear')\n",
        "\n",
        "        x = dense_in(inputs)\n",
        "        x1 = dense1(x)\n",
        "        x2 = dense2(x)\n",
        "        x3 = dense3(x)\n",
        "        x = kl.concatenate([x1, x2, x3])\n",
        "        outputs = dense_out(x)\n",
        "        model = k.Model(inputs=inputs, outputs=outputs)\n",
        "        model.compile(loss='mse', optimizer=k.optimizers.Adam(learning_rate=self.learning_rate))\n",
        "        return model\n",
        "\n",
        "    def save_model(self):\n",
        "            self.brain.save(self.weight_backup)\n",
        "            \n",
        "    def mem_usage(self):\n",
        "            return len(self.memory)/2000\n",
        "\n",
        "    def act(self, state): \n",
        "        #no need for random search anymore       \n",
        "        #if np.random.rand() <= .3:\n",
        "            #return random.randrange(self.action_size)\n",
        "        act_values = self.brain.predict(state)        \n",
        "        return np.argmax(act_values[0])\n",
        "\n",
        "    def remember(self, state, action, reward, next_state, done):\n",
        "        self.memory.append((state, action, reward, next_state, done))    \n",
        "\n",
        "    def replay(self, batch_size):\n",
        "        if len(self.memory) < batch_size:\n",
        "            return        \n",
        "        sample_batch = random.sample(self.memory, batch_size)\n",
        "        for state, action, reward, next_state, done in sample_batch:\n",
        "            target = reward\n",
        "            if not done:\n",
        "              target = reward + self.gamma * np.amax(self.brain.predict(next_state)[0])\n",
        "            target_f = self.brain.predict(state)\n",
        "            target_f[0][action] = target\n",
        "            self.brain.fit(state, target_f, epochs=1, verbose=0)\n",
        "        "
      ],
      "metadata": {
        "id": "x1kV68sVPGjW"
      },
      "execution_count": 62,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "state_size = env.observation_space.shape[0]\n",
        "action_size = env.action_space.n\n",
        "print(state_size, action_size)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "y8ZqeJQoEvsJ",
        "outputId": "2679b9f2-0276-4b56-d5e3-55d735ed496c"
      },
      "execution_count": 63,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "8 4\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "input_layers = [\n",
        "          kl.Dense(24, activation=\"relu\"),\n",
        "          kl.Dense(24, activation=\"relu\"),\n",
        "          kl.Dense(24, activation=\"relu\"),          \n",
        "]\n",
        "\n",
        "middle_layers = [\n",
        "          kl.Dense(8, activation=\"relu\"),\n",
        "          kl.Dense(8, activation=\"relu\"),\n",
        "          kl.Dense(8, activation=\"relu\"),               \n",
        "]\n",
        "\n",
        "output_layers = [\n",
        "          kl.Dense(action_size, activation=\"linear\"),\n",
        "          kl.Dense(action_size, activation=\"linear\"),\n",
        "          kl.Dense(action_size, activation=\"linear\"),          \n",
        "]\n",
        "\n",
        "state_size = env.observation_space.shape[0]\n",
        "action_size = env.action_space.n\n",
        "memory = deque(maxlen=2000)\n",
        "\n",
        "instincts = [\n",
        "             random.choice(input_layers),\n",
        "             random.choice(middle_layers), \n",
        "             random.choice(middle_layers),\n",
        "             random.choice(middle_layers),\n",
        "             random.choice(output_layers)\n",
        "             ]\n",
        "agent = DQNAgent(instincts, state_size, action_size, memory)"
      ],
      "metadata": {
        "id": "b7btGPXd6KVi"
      },
      "execution_count": 64,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "done = False\n",
        "\n",
        "for i in range(5):\n",
        "  rewards = 0\n",
        "  instincts = [random.choice(input_layers),\n",
        "               random.choice(middle_layers),\n",
        "               random.choice(middle_layers),\n",
        "               random.choice(middle_layers),\n",
        "               random.choice(output_layers)\n",
        "               ]\n",
        "  agent = DQNAgent(instincts, state_size, action_size, memory)  \n",
        "  state = env.reset()\n",
        "  state = np.reshape(state, [1, state_size])\n",
        "  done=False\n",
        "  while not done:    \n",
        "    action = agent.act(state)   \n",
        "    next_state, reward, done, _ = env.step(action) \n",
        "    rewards += reward   \n",
        "    next_state = np.reshape(next_state, [1, state_size])\n",
        "    agent.remember(state, action, reward, next_state, done)\n",
        "    state = next_state \n",
        "  agent.replay(32)    \n",
        "  print(rewards)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dDIY9C8I63im",
        "outputId": "1b0780b6-d9a9-4c37-f0c4-607fcfaacc73"
      },
      "execution_count": 65,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "-114.60003355548169\n",
            "-523.1273890915277\n",
            "-129.35058013052787\n",
            "-2556.198132513828\n",
            "-703.3459464809407\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "frames = []\n",
        "state = env.reset()\n",
        "done = False\n",
        "while not done:  \n",
        "  state = np.reshape(state, [1, state_size])\n",
        "  action = agent.act(state)\n",
        "  next_state, reward, done, _ = env.step(action)  \n",
        "  state = next_state\n",
        "  fitness += reward\n",
        "  #frames.append(env.render(mode='rgb_array'))\n",
        "env.close()\n",
        "\n",
        "print(fitness)\n",
        "#mediapy.show_video(frames, fps=30)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_A4EfUci_juk",
        "outputId": "9e55ce1c-ee60-4d4c-e9a6-e57da3d4ab65"
      },
      "execution_count": 66,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "-1867.653082091376\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Create the Environments \"CartPole-v1\", \"Acrobot-v1\", \"MountainCar-v0\", \"LunarLander-v2\"\n",
        "environments = [\n",
        "                #gym.make(\"MountainCar-v0\"),\n",
        "                gym.make(\"CartPole-v1\"),\n",
        "                #gym.make(\"Acrobot-v1\"),\n",
        "                gym.make(\"LunarLander-v2\"),\n",
        "                ]\n",
        "print(str(environments[0]))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "j8Y8jRIy0W2S",
        "outputId": "b120213d-4e6b-4692-c133-13c27417ec25"
      },
      "execution_count": 68,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "<TimeLimit<CartPoleEnv<CartPole-v1>>>\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Simulate over all environments\n",
        "middle_layers = [\n",
        "          kl.Dense(8, activation=\"relu\"),\n",
        "          kl.Dense(8, activation=\"relu\"),\n",
        "          kl.Dense(8, activation=\"relu\"), \n",
        "          kl.Dense(8, activation=\"relu\"),\n",
        "          kl.Dense(8, activation=\"relu\"),\n",
        "          kl.Dense(8, activation=\"relu\"),                \n",
        "]\n",
        "\n",
        "input_layers = [kl.Dense(24, activation=\"relu\") for e in environments]\n",
        "output_layers = [kl.Dense(e.action_space.n, activation=\"linear\") for e in environments]\n",
        "memories = [ deque(maxlen=2000) for e in environments]            \n",
        "\n",
        "for i in range(3):\n",
        "  total = 0\n",
        "  for i, env in enumerate(environments):\n",
        "    rewards = 0\n",
        "    instincts = [input_layers[i], \n",
        "                 random.choice(middle_layers),\n",
        "                 random.choice(middle_layers),\n",
        "                 random.choice(middle_layers),\n",
        "                 output_layers[i],\n",
        "                 ]\n",
        "    state_size = env.observation_space.shape[0]\n",
        "    action_size = env.action_space.n\n",
        "    agent = DQNAgent(instincts, state_size, action_size, memories[i])  \n",
        "    state = env.reset()\n",
        "    state = np.reshape(state, [1, state_size])\n",
        "    done=False\n",
        "    while not done:    \n",
        "      action = agent.act(state)   \n",
        "      next_state, reward, done, _ = env.step(action) \n",
        "      rewards += reward   \n",
        "      next_state = np.reshape(next_state, [1, state_size])\n",
        "      agent.remember(state, action, reward, next_state, done)\n",
        "      state = next_state \n",
        "    agent.replay(32) \n",
        "    total += rewards   \n",
        "  print(total/len(environments))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8DDXDgEtJHiZ",
        "outputId": "3e6a176e-d5db-4239-ba91-0c70ab799cfa"
      },
      "execution_count": 69,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "-372.00228252323103\n",
            "-166.98226534651968\n",
            "-1657.0538473574677\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Define the base gene sequence\n",
        "ENVIRONMENTS = len(environments)\n",
        "\n",
        "LAYERS = [\n",
        "          kl.Dense(8, activation=\"relu\"),\n",
        "          kl.Dense(8, activation=\"relu\"),\n",
        "          kl.Dense(8, activation=\"relu\"),\n",
        "          kl.Dense(8, activation=\"relu\"),                           \n",
        "]\n",
        "\n",
        "input_layers = [kl.Dense(24, activation=\"relu\") for e in environments]\n",
        "output_layers = [kl.Dense(e.action_space.n, activation=\"linear\") for e in environments]\n",
        "memories = [ deque(maxlen=2000) for e in environments] \n",
        "\n",
        "def create_offspring():\n",
        "  individual = []\n",
        "  for e in environments:\n",
        "    individual.append(random.uniform(0., 1.))\n",
        "    individual.append(random.uniform(0., 1.))\n",
        "    individual.append(random.uniform(0., 1.))  \n",
        "  return individual\n",
        "\n",
        "individual = create_offspring()\n",
        "print(individual)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XMZx_-6x0CEh",
        "outputId": "fe2bb26b-6103-40cf-f0a9-56f226f58793"
      },
      "execution_count": 70,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0.03845566253792376, 0.028797214902520074, 0.09473298188859736, 0.8399233710532628, 0.21627573356577445, 0.4192286655351187]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Setting up the Creator\n",
        "creator.create(\"FitnessMax\", base.Fitness, weights=(1.0,))\n",
        "creator.create(\"Individual\", list, fitness=creator.FitnessMax)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LLvhFNHQyL9O",
        "outputId": "cbd7c3a8-512e-4330-a2a3-bc4b5b8c2015"
      },
      "execution_count": 71,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/deap/creator.py:141: RuntimeWarning: A class named 'FitnessMax' has already been created and it will be overwritten. Consider deleting previous creation of that class or rename it.\n",
            "  RuntimeWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/deap/creator.py:141: RuntimeWarning: A class named 'Individual' has already been created and it will be overwritten. Consider deleting previous creation of that class or rename it.\n",
            "  RuntimeWarning)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Create Individual and Population\n",
        "NDIM = len(LAYERS)\n",
        "\n",
        "toolbox = base.Toolbox()\n",
        "toolbox.register(\"network\", create_offspring)\n",
        "toolbox.register(\"individual\", tools.initIterate, creator.Individual, toolbox.network)\n",
        "toolbox.register(\"population\", tools.initRepeat, list, toolbox.individual)\n",
        "\n",
        "toolbox.register(\"mate\", tools.cxTwoPoint)\n",
        "toolbox.register(\"mutate\", tools.mutGaussian, mu=0.0, sigma=1, indpb=1.0/NDIM)\n",
        "toolbox.register(\"select\", tools.selTournament, tournsize=3)"
      ],
      "metadata": {
        "id": "AQ5E1q8ly2hU"
      },
      "execution_count": 72,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def clamp(minv, maxv, v):\n",
        "  return min(max(v, minv), maxv)\n",
        "\n",
        "def convert_layer(layer):\n",
        "  return clamp(0, len(LAYERS)-1, int(layer * len(LAYERS)))\n",
        "\n",
        "print(convert_layer(.1))\n",
        "print(convert_layer(.3))\n",
        "print(convert_layer(.5))\n",
        "print(convert_layer(.7))\n",
        "print(convert_layer(.9))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SgD3J_f-UG6n",
        "outputId": "14743bf6-8416-4550-e360-f8f3a17149cb"
      },
      "execution_count": 73,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0\n",
            "1\n",
            "2\n",
            "2\n",
            "3\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Create and Register Evaluate Function\n",
        "#environment_max_rewards = [-1000 for e in environments]\n",
        "input_layers = [kl.Dense(24, activation=\"relu\") for e in environments]\n",
        "output_layers = [kl.Dense(e.action_space.n, activation=\"linear\") for e in environments]\n",
        "memories = [ deque(maxlen=20000) for e in environments]  \n",
        "\n",
        "def evaluate(individual, train=False):  \n",
        "  total = 0\n",
        "  for i, env in enumerate(environments):   \n",
        "    rewards = 0    \n",
        "    layer1 = convert_layer(individual[i*3])\n",
        "    layer2 = convert_layer(individual[i*3+1])    \n",
        "    layer3 = convert_layer(individual[i*3+2])  \n",
        "\n",
        "    instincts = [input_layers[i],\n",
        "                 LAYERS[layer1],\n",
        "                 LAYERS[layer2],\n",
        "                 LAYERS[layer3],\n",
        "                 output_layers[i],\n",
        "                 ]\n",
        "    state_size = env.observation_space.shape[0]\n",
        "    action_size = env.action_space.n\n",
        "    agent = DQNAgent(instincts, state_size, action_size, memories[i])  \n",
        "    state = env.reset()\n",
        "    state = np.reshape(state, [1, state_size])\n",
        "\n",
        "    done=False\n",
        "    while not done:    \n",
        "      action = agent.act(state)   \n",
        "      next_state, reward, done, _ = env.step(action) \n",
        "      rewards += reward   \n",
        "      next_state = np.reshape(next_state, [1, state_size])\n",
        "      agent.remember(state, action, reward, next_state, done)\n",
        "      state = next_state    \n",
        "    total += rewards \n",
        "    if train:\n",
        "      #only train if the agent is the best      \n",
        "      agent.replay(32)\n",
        "    \n",
        "  print(total/len(environments))\n",
        "  return total/len(environments),\n",
        "\n",
        "toolbox.register(\"evaluate\", evaluate)   "
      ],
      "metadata": {
        "id": "92nfzNTTQlWu"
      },
      "execution_count": 85,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Setup Evolution\n",
        "MU = 5 #@param {type:\"slider\", min:5, max:1000, step:5}\n",
        "NGEN = 1000 #@param {type:\"slider\", min:5, max:1000, step:1}\n",
        "RGEN = 1 #@param {type:\"slider\", min:1, max:100, step:1}\n",
        "CXPB = .6\n",
        "MUTPB = .3\n",
        "\n",
        "random.seed(64)\n",
        "\n",
        "pop = toolbox.population(n=MU)\n",
        "hof = tools.HallOfFame(1)\n",
        "stats = tools.Statistics(lambda ind: ind.fitness.values)\n",
        "stats.register(\"avg\", np.mean)\n",
        "stats.register(\"std\", np.std)\n",
        "stats.register(\"min\", np.min)\n",
        "stats.register(\"max\", np.max)"
      ],
      "metadata": {
        "id": "TkP_2nWWPwev"
      },
      "execution_count": 87,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "best = None\n",
        "groups = { \"fitness\" : {\"min\", \"max\"}}\n",
        "plotlosses = PlotLosses(groups=groups)\n",
        "\n",
        "for g in range(NGEN):\n",
        "  pop, logbook = algorithms.eaSimple(pop, toolbox, \n",
        "            cxpb=CXPB, mutpb=MUTPB, ngen=RGEN, stats=stats, halloffame=hof, verbose=False)\n",
        "  best = hof[0] \n",
        "  best_fit = evaluate(best, train=True) \n",
        "        \n",
        "  for l in logbook:\n",
        "    plotlosses.update({'max': l[\"max\"], 'min': l[\"min\"]})    \n",
        "  plotlosses.send()  # draw, update logs, etc\n",
        "  print(f\"Gen ({(g+1)*RGEN}) Best Fitness ({best_fit})\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 689
        },
        "id": "B6vJQ6sDRfhE",
        "outputId": "a5c440e5-b6aa-48d7-f30e-2d604978bc2c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 864x576 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "fitness\n",
            "\tmax              \t (min:  -42.364, max:  124.727, cur:   72.857)\n",
            "\tmin              \t (min: -1155.257, max:   -1.862, cur:  -64.080)\n",
            "Gen (102) Best Fitness ((-57.133287809951526,))\n",
            "-50.358696611826126\n",
            "108.80536639192476\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "fitness = 0\n",
        "frames = []\n",
        "\n",
        "state = env.reset()\n",
        "done = False\n",
        "while not done:  \n",
        "  state = np.reshape(state, [1, state_size])\n",
        "  action = agent.act(state)\n",
        "  next_state, reward, done, _ = env.step(action)  \n",
        "  state = next_state\n",
        "  fitness += reward\n",
        "  frames.append(env.render(mode='rgb_array'))\n",
        "env.close()\n",
        "\n",
        "print(fitness)\n",
        "mediapy.show_video(frames, fps=30)"
      ],
      "metadata": {
        "id": "L5Mm0dwAfmlT"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}