predator_prey_NN_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "542bafaf",
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "8b428a26",
   "metadata": {},
   "outputs": [],
   "source": [
    "class Policy:\n",
    "    def __init__(self):\n",
    "        self.alpha = 0.01\n",
    "        self.policy = keras.models.Sequential([\n",
    "            keras.layers.Dense(25, activation=\"relu\", input_shape=[25]),\n",
    "            keras.layers.Dense(10, activation=\"relu\"),\n",
    "            keras.layers.Dense(5, activation=\"softmax\"),\n",
    "        ])\n",
    "        self.policy.compile(optimizer=Adam(learning_rate=self.alpha))\n",
    "        \n",
    "    def __call__(self, state):\n",
    "        pi = self.policy(state)\n",
    "        return pi\n",
    "    \n",
    "def get_obs(array, pos):\n",
    "    obs = []\n",
    "    i, j = pos\n",
    "    for ii in range(-2, 3, 1):\n",
    "        row = []\n",
    "        for jj in range(-2, 3, 1):\n",
    "            row.append(array[(i + ii) % len(array)][(j + jj) % len(array)])\n",
    "        obs.append(row)     \n",
    "    return obs\n",
    "\n",
    "def apply_action(array, pos, action):\n",
    "    i, j = pos\n",
    "    if action == 0:\n",
    "        array[i][j], array[i][(j + 1) % len(array)] = array[i][(j + 1) % len(array)], array[i][j]\n",
    "    elif action == 1:\n",
    "        array[i][j], array[i][(j - 1)] = array[i][(j - 1)], array[i][j]\n",
    "    elif action == 2:\n",
    "        array[i][j], array[i - 1][j] = array[i - 1][j], array[i][j]\n",
    "    elif action == 3:\n",
    "        array[i][j], array[i + 1][j] = array[(i + 1) % len(array)][j], array[i][j]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "dcb79bd7",
   "metadata": {},
   "outputs": [],
   "source": [
    "import gym\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "da001ff7",
   "metadata": {},
   "outputs": [],
   "source": [
    "types = [0, 1, 2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "771497b4",
   "metadata": {},
   "outputs": [],
   "source": [
    "obs = [[random.choice(types) for i in range(5)] for j in range(5)]\n",
    "obs[2][2] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "e1e68808",
   "metadata": {},
   "outputs": [],
   "source": [
    "array = np.array([[random.choice(types) for i in range(10)] for j in range(10)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "155fbcac",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 1, 1, 0, 1, 0, 2, 0, 2, 1],\n",
       "       [0, 1, 0, 0, 1, 0, 0, 0, 1, 0],\n",
       "       [0, 2, 2, 2, 1, 0, 0, 1, 2, 1],\n",
       "       [1, 0, 1, 1, 2, 0, 2, 2, 1, 1],\n",
       "       [1, 2, 2, 1, 0, 2, 1, 2, 0, 0],\n",
       "       [1, 0, 2, 1, 0, 0, 0, 2, 1, 1],\n",
       "       [2, 0, 2, 0, 2, 1, 1, 0, 1, 2],\n",
       "       [2, 2, 1, 1, 2, 0, 1, 0, 2, 2],\n",
       "       [0, 2, 1, 1, 0, 1, 1, 2, 0, 0],\n",
       "       [0, 1, 1, 1, 0, 2, 0, 2, 1, 2]])"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "array"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d4fda7bb",
   "metadata": {},
   "source": [
    "## Create Environment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "a99a6a19",
   "metadata": {},
   "outputs": [],
   "source": [
    "from gym import Env\n",
    "from gym.spaces import Discrete, Box, MultiDiscrete\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "fadabffa",
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "Lets create environment that we have just one predator in initial middle of table\n",
    "'''\n",
    "class PPEnv(Env):\n",
    "    def __init__(self, L=5, time_limit=40, pos=(0, 0)):\n",
    "        self.L = L\n",
    "        self.pos = (self.L // 2, self.L // 2)\n",
    "        self.action_space = Discrete(4)\n",
    "        self.obs_shape = (5, 5)\n",
    "        self.observation_space = MultiDiscrete(np.array([[3 for i in range(5)] for j in range(5)]))\n",
    "        self.types = Discrete(3)\n",
    "        self.array = np.array([[Discrete(2).sample() * 2 for i in range(self.L)] for j in range(self.L)])\n",
    "        # Put one predator\n",
    "        self.array[self.pos[0], self.pos[1]] = 1               \n",
    "        self.time = 0\n",
    "        self.time_limit = time_limit\n",
    "    \n",
    "    def get_obs(self):\n",
    "        obs = []\n",
    "        i, j = self.pos\n",
    "        for ii in range(-2, 3, 1):\n",
    "            row = []\n",
    "            for jj in range(-2, 3, 1):\n",
    "                row.append(self.array[(i + ii) % self.L][(j + jj) % self.L])\n",
    "            obs.append(row)     \n",
    "        return np.array(obs)\n",
    "    \n",
    "    # Jump\n",
    "    def apply_action(self, action):\n",
    "        i, j = self.pos\n",
    "        L = self.L\n",
    "        if action == 0:\n",
    "            self.array[i][j], self.array[i][(j + 1) % L] = self.array[i][(j + 1) % L], self.array[i][j]\n",
    "            self.pos = (i, (j + 1) % L)\n",
    "        elif action == 1:\n",
    "            self.array[i][j], self.array[i][(j - 1) % L] = self.array[i][(j - 1) % L], self.array[i][j]\n",
    "            self.pos = (i, (j - 1) % L)\n",
    "        elif action == 2:\n",
    "            self.array[i][j], self.array[(i - 1) % L][j] = self.array[(i - 1) % L][j], self.array[i][j]\n",
    "            self.pos = ((i - 1) % L, j)\n",
    "        elif action == 3:\n",
    "            self.array[i][j], self.array[(i + 1) % L][j] = self.array[(i + 1) % L][j], self.array[i][j]\n",
    "            self.pos = ((i + 1) % L, j)\n",
    "    \n",
    "    def get_neighboors(self):\n",
    "        i, j = self.pos\n",
    "        adress = [\n",
    "            (i, (j + 1) % self.L), \n",
    "            (i, (j - 1) % self.L), \n",
    "            ((i - 1) % self.L, j), \n",
    "            ((i + 1) % self.L, j)\n",
    "        ]\n",
    "        right, left, up, down = self.array[adress[0][0]][adress[0][1]], self.array[adress[1][0]][adress[1][1]], self.array[adress[2][0]][adress[2][1]], self.array[adress[3][0]][adress[3][1]]\n",
    "        return [right, left, up, down]\n",
    "    \n",
    "    def step(self, action):\n",
    "        # Get obs\n",
    "        obs = self.get_obs()\n",
    "        \n",
    "        # Colect reward\n",
    "        ngbs = [obs[2][2 + 1], obs[2][2 - 1], obs[2 - 1][2], obs[2 + 1][2]] #self.get_neighboors()\n",
    "        reward = 2 ** ngbs.count(2)\n",
    "        \n",
    "        # Get info\n",
    "        info = {}\n",
    "        \n",
    "        # Check if done\n",
    "        if self.time >= self.time_limit:\n",
    "            done = True\n",
    "            return obs, reward, done, info\n",
    "        else: \n",
    "            done = False\n",
    "            \n",
    "        # apply action\n",
    "        self.apply_action(action)\n",
    "        self.time += 1\n",
    "        return obs, reward, done, info\n",
    "    \n",
    "    def render(self):\n",
    "        print('-' * self.L)\n",
    "        print(self.array)\n",
    "        print('-' * self.L)\n",
    "    \n",
    "    def reset(self):\n",
    "        self.pos = (self.L // 2, self.L // 2)\n",
    "        self.action_space = Discrete(4)\n",
    "        self.observation_space = MultiDiscrete(np.array([[3 for i in range(5)] for j in range(5)]))\n",
    "        self.types = Discrete(3)\n",
    "        self.array = np.array([[Discrete(2).sample() * 2 for i in range(self.L)] for j in range(self.L)])\n",
    "        # Put one predator\n",
    "        self.array[self.pos[0], self.pos[1]] = 1               \n",
    "        self.time = 0\n",
    "        return np.array(self.get_obs())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "126bc534",
   "metadata": {},
   "outputs": [],
   "source": [
    "env = PPEnv()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "cc5d7a4d",
   "metadata": {},
   "outputs": [],
   "source": [
    "obs = env.reset()\n",
    "env.apply_action(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "f697032b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "env.action_space.sample()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ae36a41d",
   "metadata": {},
   "source": [
    "### Testing environment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "3efdb1d4",
   "metadata": {},
   "outputs": [],
   "source": [
    "render = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "3b587af9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ep: 0 || Score: 256\n",
      "Ep: 1 || Score: 292\n",
      "Ep: 2 || Score: 168\n",
      "Ep: 3 || Score: 214\n",
      "Ep: 4 || Score: 129\n",
      "Ep: 5 || Score: 184\n",
      "Ep: 6 || Score: 155\n",
      "Ep: 7 || Score: 215\n",
      "Ep: 8 || Score: 183\n",
      "Ep: 9 || Score: 172\n",
      "Ep: 10 || Score: 293\n",
      "Ep: 11 || Score: 177\n",
      "Ep: 12 || Score: 158\n",
      "Ep: 13 || Score: 144\n",
      "Ep: 14 || Score: 252\n",
      "Ep: 15 || Score: 211\n",
      "Ep: 16 || Score: 113\n",
      "Ep: 17 || Score: 211\n",
      "Ep: 18 || Score: 229\n",
      "Ep: 19 || Score: 195\n",
      "Ep: 20 || Score: 312\n",
      "Ep: 21 || Score: 210\n",
      "Ep: 22 || Score: 202\n",
      "Ep: 23 || Score: 258\n",
      "Ep: 24 || Score: 239\n",
      "Ep: 25 || Score: 176\n",
      "Ep: 26 || Score: 147\n",
      "Ep: 27 || Score: 200\n",
      "Ep: 28 || Score: 163\n",
      "Ep: 29 || Score: 316\n",
      "Ep: 30 || Score: 217\n",
      "Ep: 31 || Score: 152\n",
      "Ep: 32 || Score: 169\n",
      "Ep: 33 || Score: 164\n",
      "Ep: 34 || Score: 241\n",
      "Ep: 35 || Score: 239\n",
      "Ep: 36 || Score: 76\n",
      "Ep: 37 || Score: 257\n",
      "Ep: 38 || Score: 179\n",
      "Ep: 39 || Score: 219\n",
      "Ep: 40 || Score: 181\n",
      "Ep: 41 || Score: 308\n",
      "Ep: 42 || Score: 165\n",
      "Ep: 43 || Score: 214\n",
      "Ep: 44 || Score: 394\n",
      "Ep: 45 || Score: 215\n",
      "Ep: 46 || Score: 173\n",
      "Ep: 47 || Score: 127\n",
      "Ep: 48 || Score: 233\n",
      "Ep: 49 || Score: 122\n",
      "Ep: 50 || Score: 192\n",
      "Ep: 51 || Score: 436\n",
      "Ep: 52 || Score: 314\n",
      "Ep: 53 || Score: 243\n",
      "Ep: 54 || Score: 156\n",
      "Ep: 55 || Score: 186\n",
      "Ep: 56 || Score: 183\n",
      "Ep: 57 || Score: 274\n",
      "Ep: 58 || Score: 129\n",
      "Ep: 59 || Score: 184\n",
      "Ep: 60 || Score: 202\n",
      "Ep: 61 || Score: 198\n",
      "Ep: 62 || Score: 131\n",
      "Ep: 63 || Score: 173\n",
      "Ep: 64 || Score: 164\n",
      "Ep: 65 || Score: 290\n",
      "Ep: 66 || Score: 171\n",
      "Ep: 67 || Score: 180\n",
      "Ep: 68 || Score: 186\n",
      "Ep: 69 || Score: 211\n",
      "Ep: 70 || Score: 194\n",
      "Ep: 71 || Score: 137\n",
      "Ep: 72 || Score: 179\n",
      "Ep: 73 || Score: 192\n",
      "Ep: 74 || Score: 288\n",
      "Ep: 75 || Score: 183\n",
      "Ep: 76 || Score: 210\n",
      "Ep: 77 || Score: 141\n",
      "Ep: 78 || Score: 168\n",
      "Ep: 79 || Score: 220\n",
      "Ep: 80 || Score: 400\n",
      "Ep: 81 || Score: 177\n",
      "Ep: 82 || Score: 177\n",
      "Ep: 83 || Score: 258\n",
      "Ep: 84 || Score: 208\n",
      "Ep: 85 || Score: 209\n",
      "Ep: 86 || Score: 107\n",
      "Ep: 87 || Score: 228\n",
      "Ep: 88 || Score: 170\n",
      "Ep: 89 || Score: 195\n",
      "Ep: 90 || Score: 157\n",
      "Ep: 91 || Score: 201\n",
      "Ep: 92 || Score: 239\n",
      "Ep: 93 || Score: 164\n",
      "Ep: 94 || Score: 266\n",
      "Ep: 95 || Score: 372\n",
      "Ep: 96 || Score: 227\n",
      "Ep: 97 || Score: 240\n",
      "Ep: 98 || Score: 324\n",
      "Ep: 99 || Score: 195\n"
     ]
    }
   ],
   "source": [
    "episodes = 100\n",
    "for ep in range(episodes):\n",
    "    obs = env.reset()\n",
    "    done = False\n",
    "    score = 0\n",
    "    while not done:\n",
    "        obs, rew, done, info = env.step(env.action_space.sample())\n",
    "        if render:\n",
    "            env.render()\n",
    "        score += rew\n",
    "    print(f\"Ep: {ep} || Score: {score}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "2aa64ba6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25,)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array([Discrete(3)] * 25).shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "c872f175",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "MultiDiscrete(2).sample()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "114cfe87",
   "metadata": {},
   "source": [
    "## Genetic Algorithm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "573f15ba",
   "metadata": {},
   "outputs": [],
   "source": [
    "env = PPEnv()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "15f90502",
   "metadata": {},
   "outputs": [],
   "source": [
    "from tqdm import tqdm\n",
    "import random\n",
    "\n",
    "num_models = 100\n",
    "num_generations = 100\n",
    "nb_of_simulations = 10\n",
    "fraction = 0.2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "a2d745a9",
   "metadata": {},
   "outputs": [],
   "source": [
    "class NeuralModel:\n",
    "\n",
    "    @staticmethod\n",
    "    def get_random_params(obs_size, act_size):\n",
    "        \n",
    "#         return [\n",
    "#             np.random.randn(obs_size, obs_size + 1),\n",
    "#             np.zeros(obs_size + 1),\n",
    "#             np.random.randn(obs_size + 1, act_size),\n",
    "#             np.zeros(act_size),\n",
    "#         ]\n",
    "        return [\n",
    "            np.random.randn(25, 25),\n",
    "            np.zeros(25),\n",
    "            \n",
    "            np.random.randn(25, 25),\n",
    "            np.zeros(25),\n",
    "            \n",
    "            np.random.randn(25, 4),\n",
    "            np.zeros(4),\n",
    "        ]\n",
    "        \n",
    "\n",
    "    def __init__(self, obs_act_sizes, params=None):\n",
    "        self.obs_act_sizes = obs_act_sizes\n",
    "        obs_size, act_size = self.obs_act_sizes\n",
    "#         obs_size = len(env.observation_space.sample())\n",
    "#         act_size = env.action_space.n\n",
    "        if not params:\n",
    "            params = self.get_random_params(obs_size, act_size)\n",
    "\n",
    "        n_inputs = obs_size  # env.observation_space.shape[0]\n",
    "        self.model = keras.models.Sequential([\n",
    "            \n",
    "#             keras.layers.Dense(obs_size + 1, activation=\"elu\", input_shape=[n_inputs]),\n",
    "#             keras.layers.Dense(act_size, activation=\"softmax\"),\n",
    "            \n",
    "            keras.layers.Dense(25, activation=\"relu\", input_shape=[25]),\n",
    "            keras.layers.Dense(25, activation=\"relu\"),\n",
    "            keras.layers.Dense(4, activation=\"softmax\"),\n",
    "            \n",
    "        ])\n",
    "        \n",
    "        self.model.compile()\n",
    "\n",
    "        self.model.set_weights(params)\n",
    "\n",
    "    def __call__(self, obs):\n",
    "        return self.model(obs)\n",
    "\n",
    "    def make_offspring(self):\n",
    "        weights = self.model.get_weights()\n",
    "        \n",
    "#         new_weights = [w + np.random.randn(w.shape) for w in weights]\n",
    "        new_weights = []\n",
    "        for w in weights:\n",
    "            if len(w.shape) == 2:\n",
    "                new_weights.append(w + np.random.randn(w.shape[0], w.shape[1])/10)\n",
    "            else:\n",
    "                new_weights.append(w)\n",
    "        #print(new_weights)\n",
    "        \n",
    "        return NeuralModel(self.obs_act_sizes, params=new_weights)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "b1c46bbb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(25, 4)\n"
     ]
    }
   ],
   "source": [
    "def run_simulation(model, nb_of_simulations=10):\n",
    "\n",
    "    total_reward = 0\n",
    "    steps = 0\n",
    "    for _ in range(nb_of_simulations):\n",
    "        \n",
    "        observation = env.reset().reshape(1,-1) #seed=random.randrange(1000))\n",
    "        done = False\n",
    "\n",
    "        while not done:\n",
    "            action = model.model(observation).numpy().argmax()\n",
    "            observation, reward, done, info = env.step(action)\n",
    "            observation = observation.reshape(1,-1)\n",
    "            total_reward += reward\n",
    "            steps += 1\n",
    "#             env.render()\n",
    "    #env.close()\n",
    "    return total_reward / nb_of_simulations, steps\n",
    "\n",
    "obs_act_sizes = (len(env.observation_space.sample()) ** 2, env.action_space.n)\n",
    "models = [NeuralModel(obs_act_sizes) for _ in range(num_models)]\n",
    "print(obs_act_sizes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "9c199e56",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 0, 1, 0, 2],\n",
       "       [1, 2, 1, 0, 2],\n",
       "       [1, 0, 0, 1, 1],\n",
       "       [1, 0, 1, 1, 2],\n",
       "       [0, 0, 1, 0, 0]], dtype=int64)"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "env.observation_space.sample()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "ceeee314",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  0%|                                                                                          | 0/100 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "41000 207.3029999999999 269.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  1%|▊                                                                               | 1/100 [00:56<1:32:48, 56.25s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "82000 221.85899999999995 317.3 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  2%|█▌                                                                              | 2/100 [02:04<1:43:50, 63.58s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "123000 227.51100000000002 346.2 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  3%|██▍                                                                             | 3/100 [03:06<1:41:06, 62.54s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "164000 240.15099999999995 333.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  4%|███▏                                                                            | 4/100 [04:07<1:39:21, 62.10s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "205000 256.76400000000007 369.8 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  5%|████                                                                            | 5/100 [05:04<1:35:23, 60.25s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "246000 253.71799999999996 351.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  6%|████▊                                                                           | 6/100 [06:04<1:33:57, 59.97s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "287000 266.324 360.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  7%|█████▌                                                                          | 7/100 [07:04<1:33:01, 60.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "328000 261.013 357.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  8%|██████▍                                                                         | 8/100 [07:59<1:29:52, 58.61s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "369000 262.283 397.9 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      "  9%|███████▏                                                                        | 9/100 [09:00<1:29:53, 59.27s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "410000 265.96400000000006 360.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 10%|███████▉                                                                       | 10/100 [09:56<1:27:34, 58.38s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "451000 259.0819999999999 359.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 11%|████████▋                                                                      | 11/100 [10:52<1:25:28, 57.62s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "492000 269.13400000000007 370.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 12%|█████████▍                                                                     | 12/100 [11:49<1:24:15, 57.45s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "533000 267.197 356.8 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 13%|██████████▎                                                                    | 13/100 [12:52<1:25:27, 58.94s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "574000 272.00300000000004 402.8 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 14%|███████████                                                                    | 14/100 [13:56<1:26:42, 60.50s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "615000 265.33700000000016 334.7 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 15%|███████████▊                                                                   | 15/100 [14:52<1:23:56, 59.25s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "656000 263.8930000000001 361.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 16%|████████████▋                                                                  | 16/100 [15:48<1:21:27, 58.18s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "697000 269.06500000000005 378.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 17%|█████████████▍                                                                 | 17/100 [17:03<1:27:43, 63.42s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "738000 268.96299999999997 411.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 18%|██████████████▏                                                                | 18/100 [18:24<1:33:42, 68.57s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "779000 275.443 373.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 19%|███████████████                                                                | 19/100 [19:44<1:37:14, 72.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "820000 267.79400000000004 373.8 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 20%|███████████████▊                                                               | 20/100 [21:02<1:38:34, 73.93s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "861000 277.964 388.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 21%|████████████████▌                                                              | 21/100 [22:14<1:36:27, 73.26s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "902000 277.966 372.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 22%|█████████████████▍                                                             | 22/100 [23:12<1:29:20, 68.73s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "943000 274.668 378.2 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 23%|██████████████████▏                                                            | 23/100 [24:13<1:24:54, 66.17s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "984000 279.915 381.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 24%|██████████████████▉                                                            | 24/100 [25:10<1:20:21, 63.44s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1025000 267.2129999999999 375.7 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 25%|███████████████████▊                                                           | 25/100 [26:05<1:16:19, 61.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1066000 287.6430000000001 384.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 26%|████████████████████▌                                                          | 26/100 [27:09<1:16:24, 61.95s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1107000 269.1819999999999 347.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 27%|█████████████████████▎                                                         | 27/100 [28:06<1:13:38, 60.53s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1148000 284.1309999999999 407.2 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 28%|██████████████████████                                                         | 28/100 [29:29<1:20:32, 67.12s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1189000 274.74800000000016 376.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 29%|██████████████████████▉                                                        | 29/100 [30:50<1:24:13, 71.17s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1230000 277.842 379.3 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 30%|███████████████████████▋                                                       | 30/100 [32:10<1:26:17, 73.97s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1271000 272.8990000000001 371.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 31%|████████████████████████▍                                                      | 31/100 [33:27<1:26:16, 75.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1312000 273.217 370.7 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 32%|█████████████████████████▎                                                     | 32/100 [34:46<1:26:12, 76.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1353000 280.07499999999993 386.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 33%|██████████████████████████                                                     | 33/100 [36:10<1:27:31, 78.37s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1394000 278.75899999999996 416.7 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 34%|██████████████████████████▊                                                    | 34/100 [37:29<1:26:24, 78.55s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1435000 274.31600000000003 378.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 35%|███████████████████████████▋                                                   | 35/100 [38:57<1:28:11, 81.41s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1476000 276.899 367.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 36%|████████████████████████████▍                                                  | 36/100 [40:24<1:28:47, 83.24s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1517000 268.75099999999986 342.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 37%|█████████████████████████████▏                                                 | 37/100 [41:44<1:26:18, 82.20s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1558000 278.26300000000015 370.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 38%|██████████████████████████████                                                 | 38/100 [43:02<1:23:40, 80.98s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1599000 276.60799999999995 404.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 39%|██████████████████████████████▊                                                | 39/100 [44:21<1:21:45, 80.41s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1640000 278.5439999999999 405.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 40%|███████████████████████████████▌                                               | 40/100 [45:40<1:20:01, 80.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1681000 285.99600000000015 414.7 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 41%|████████████████████████████████▍                                              | 41/100 [47:07<1:20:43, 82.10s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1722000 281.1530000000001 388.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 42%|█████████████████████████████████▏                                             | 42/100 [48:13<1:14:43, 77.30s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1763000 278.70000000000016 359.7 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 43%|█████████████████████████████████▉                                             | 43/100 [49:17<1:09:28, 73.13s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1804000 282.76300000000003 397.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 44%|██████████████████████████████████▊                                            | 44/100 [50:22<1:06:03, 70.78s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1845000 284.45099999999996 378.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 45%|███████████████████████████████████▌                                           | 45/100 [51:25<1:02:43, 68.43s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1886000 275.58899999999994 377.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 46%|████████████████████████████████████▎                                          | 46/100 [52:32<1:01:16, 68.08s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1927000 277.2279999999999 381.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 47%|██████████████████████████████████████                                           | 47/100 [53:38<59:28, 67.32s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1968000 282.5589999999999 390.2 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 48%|█████████████████████████████████████▉                                         | 48/100 [55:13<1:05:38, 75.75s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2009000 283.87599999999986 394.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 49%|██████████████████████████████████████▋                                        | 49/100 [56:37<1:06:21, 78.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2050000 285.09400000000005 376.9 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 50%|███████████████████████████████████████▌                                       | 50/100 [57:44<1:02:13, 74.67s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2091000 284.057 423.8 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 51%|████████████████████████████████████████▎                                      | 51/100 [58:58<1:00:55, 74.59s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2132000 289.0869999999999 398.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 52%|█████████████████████████████████████████                                      | 52/100 [1:00:02<57:08, 71.43s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2173000 283.941 389.3 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 53%|█████████████████████████████████████████▊                                     | 53/100 [1:01:10<55:10, 70.44s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2214000 279.89199999999994 376.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 54%|██████████████████████████████████████████▋                                    | 54/100 [1:02:18<53:27, 69.73s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2255000 277.09900000000005 390.3 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 55%|███████████████████████████████████████████▍                                   | 55/100 [1:03:24<51:28, 68.63s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2296000 289.2919999999998 393.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 56%|████████████████████████████████████████████▏                                  | 56/100 [1:04:33<50:26, 68.79s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2337000 282.07099999999997 355.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 57%|█████████████████████████████████████████████                                  | 57/100 [1:05:38<48:21, 67.48s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2378000 287.0469999999999 377.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 58%|█████████████████████████████████████████████▊                                 | 58/100 [1:06:46<47:28, 67.82s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2419000 290.152 397.3 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 59%|██████████████████████████████████████████████▌                                | 59/100 [1:07:55<46:35, 68.17s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2460000 285.60800000000006 391.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 60%|███████████████████████████████████████████████▍                               | 60/100 [1:08:58<44:16, 66.40s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2501000 285.95099999999996 357.2 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 61%|████████████████████████████████████████████████▏                              | 61/100 [1:10:10<44:16, 68.10s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2542000 286.2060000000001 391.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 62%|████████████████████████████████████████████████▉                              | 62/100 [1:11:14<42:26, 67.01s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2583000 288.1359999999999 374.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 63%|█████████████████████████████████████████████████▊                             | 63/100 [1:12:22<41:28, 67.24s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2624000 283.24399999999986 389.2 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 64%|██████████████████████████████████████████████████▌                            | 64/100 [1:13:34<41:10, 68.63s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2665000 286.5740000000001 376.7 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 65%|███████████████████████████████████████████████████▎                           | 65/100 [1:14:37<39:07, 67.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2706000 284.43600000000004 377.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 66%|████████████████████████████████████████████████████▏                          | 66/100 [1:15:48<38:34, 68.08s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2747000 287.7059999999999 367.3 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 67%|████████████████████████████████████████████████████▉                          | 67/100 [1:16:56<37:28, 68.14s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2788000 290.427 414.2 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 68%|█████████████████████████████████████████████████████▋                         | 68/100 [1:18:02<36:02, 67.59s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2829000 290.54 379.9 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 69%|██████████████████████████████████████████████████████▌                        | 69/100 [1:19:14<35:32, 68.78s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2870000 290.44399999999996 413.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 70%|███████████████████████████████████████████████████████▎                       | 70/100 [1:20:18<33:43, 67.47s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2911000 288.536 386.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 71%|████████████████████████████████████████████████████████                       | 71/100 [1:21:08<30:01, 62.11s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2952000 295.1349999999999 423.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 72%|████████████████████████████████████████████████████████▉                      | 72/100 [1:21:51<26:18, 56.37s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2993000 290.599 369.4 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 73%|█████████████████████████████████████████████████████████▋                     | 73/100 [1:22:33<23:22, 51.96s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3034000 286.9310000000001 371.2 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 74%|██████████████████████████████████████████████████████████▍                    | 74/100 [1:23:15<21:17, 49.13s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3075000 288.344 433.3 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 75%|███████████████████████████████████████████████████████████▎                   | 75/100 [1:23:57<19:35, 47.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3116000 283.018 369.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 76%|████████████████████████████████████████████████████████████                   | 76/100 [1:24:47<19:06, 47.77s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3157000 292.7989999999999 401.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 77%|████████████████████████████████████████████████████████████▊                  | 77/100 [1:25:41<19:05, 49.81s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3198000 287.61799999999994 384.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 78%|█████████████████████████████████████████████████████████████▌                 | 78/100 [1:26:27<17:47, 48.52s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3239000 286.4180000000001 405.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 79%|██████████████████████████████████████████████████████████████▍                | 79/100 [1:27:15<16:56, 48.43s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3280000 288.81200000000007 411.2 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 80%|███████████████████████████████████████████████████████████████▏               | 80/100 [1:28:13<17:03, 51.18s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3321000 297.75600000000003 424.7 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 81%|███████████████████████████████████████████████████████████████▉               | 81/100 [1:28:57<15:36, 49.29s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3362000 296.03400000000005 390.3 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 82%|████████████████████████████████████████████████████████████████▊              | 82/100 [1:29:49<14:59, 49.96s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3403000 288.994 400.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 83%|█████████████████████████████████████████████████████████████████▌             | 83/100 [1:31:03<16:11, 57.12s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3444000 282.0310000000001 390.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 84%|██████████████████████████████████████████████████████████████████▎            | 84/100 [1:32:07<15:47, 59.19s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3485000 293.71100000000007 367.9 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 85%|███████████████████████████████████████████████████████████████████▏           | 85/100 [1:33:16<15:32, 62.16s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3526000 284.612 408.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 86%|███████████████████████████████████████████████████████████████████▉           | 86/100 [1:34:22<14:44, 63.20s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3567000 294.389 361.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 87%|████████████████████████████████████████████████████████████████████▋          | 87/100 [1:35:27<13:50, 63.89s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3608000 286.78499999999997 389.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 88%|█████████████████████████████████████████████████████████████████████▌         | 88/100 [1:36:34<12:58, 64.86s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3649000 294.29400000000004 412.6 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 89%|██████████████████████████████████████████████████████████████████████▎        | 89/100 [1:37:37<11:46, 64.19s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3690000 282.43399999999997 359.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 90%|███████████████████████████████████████████████████████████████████████        | 90/100 [1:38:42<10:44, 64.49s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3731000 291.70700000000005 392.0 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 91%|███████████████████████████████████████████████████████████████████████▉       | 91/100 [1:39:44<09:34, 63.82s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3772000 285.37300000000005 380.9 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 92%|████████████████████████████████████████████████████████████████████████▋      | 92/100 [1:40:51<08:37, 64.74s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3813000 286.1299999999998 447.8 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 93%|█████████████████████████████████████████████████████████████████████████▍     | 93/100 [1:41:52<07:25, 63.62s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3854000 292.7749999999999 417.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 94%|██████████████████████████████████████████████████████████████████████████▎    | 94/100 [1:42:58<06:26, 64.36s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3895000 288.18600000000004 418.1 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 95%|███████████████████████████████████████████████████████████████████████████    | 95/100 [1:44:00<05:17, 63.46s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3936000 293.0220000000001 381.3 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 96%|███████████████████████████████████████████████████████████████████████████▊   | 96/100 [1:45:06<04:16, 64.22s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3977000 290.4959999999999 400.9 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 97%|████████████████████████████████████████████████████████████████████████████▋  | 97/100 [1:46:07<03:10, 63.36s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4018000 286.03299999999984 353.9 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 98%|█████████████████████████████████████████████████████████████████████████████▍ | 98/100 [1:47:08<02:05, 62.55s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4059000 283.31800000000004 372.5 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r\n",
      " 99%|██████████████████████████████████████████████████████████████████████████████▏| 99/100 [1:48:10<01:02, 62.51s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4100000 276.5959999999999 393.9 410.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████████████████████████████████████████| 100/100 [1:49:14<00:00, 65.54s/it]\n"
     ]
    }
   ],
   "source": [
    "fitness_y = []\n",
    "len_ep_mean_y = []\n",
    "steps_x = []\n",
    "\n",
    "for generation in tqdm(range(num_generations)):\n",
    "    # Roda cada modelo e obtem seu fitness.\n",
    "    # fitness = [run_simulation(model, 10) for model in models]\n",
    "    fitness = []\n",
    "    steps = []\n",
    "    for model in models:\n",
    "        fit_steps = run_simulation(model, nb_of_simulations)\n",
    "        fitness.append(fit_steps[0])\n",
    "        steps.append(fit_steps[1])\n",
    "        \n",
    "    # Acumula fitness médio pra plotar gráfico\n",
    "    fitness_y.append(sum(fitness) / len(fitness))\n",
    "    len_ep_mean_y.append(sum(steps) / len(steps))\n",
    "    if steps_x == []:\n",
    "        steps_x.append(sum(steps))\n",
    "    else:\n",
    "        steps_x.append(steps_x[-1] + (sum(steps)))\n",
    "    \n",
    "    # Mostra último ponto de fitness\n",
    "    print(steps_x[-1], fitness_y[-1], max(fitness), len_ep_mean_y[-1])\n",
    "    \n",
    "    # Seleciona os melhores.\n",
    "    sorted_indexes = list(item[1] for item in sorted(\n",
    "        ((fitness, k) for k, fitness in enumerate(fitness)), reverse=True))\n",
    "    \n",
    "    number_suvivors = int(num_models * fraction)\n",
    "    models_to_keep = [models[k] for k in sorted_indexes[:number_suvivors]]\n",
    "\n",
    "    # Cria as cria.\n",
    "    #new_models = [m.make_offspring() for m in models_to_keep]\n",
    "    new_models = [models_to_keep[i%number_suvivors].make_offspring() for i in range(num_models - number_suvivors)]\n",
    "\n",
    "    models = models_to_keep + new_models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 640,
   "id": "537bce4c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-----\n",
      "[[2 2 2 0 2]\n",
      " [0 0 0 2 2]\n",
      " [0 2 0 0 0]\n",
      " [0 0 2 2 2]\n",
      " [0 2 1 2 0]]\n",
      "-----\n"
     ]
    }
   ],
   "source": [
    "env.render()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "ec7f940a",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[207.3029999999999,\n",
       " 221.85899999999995,\n",
       " 227.51100000000002,\n",
       " 240.15099999999995,\n",
       " 256.76400000000007,\n",
       " 253.71799999999996,\n",
       " 266.324,\n",
       " 261.013,\n",
       " 262.283,\n",
       " 265.96400000000006,\n",
       " 259.0819999999999,\n",
       " 269.13400000000007,\n",
       " 267.197,\n",
       " 272.00300000000004,\n",
       " 265.33700000000016,\n",
       " 263.8930000000001,\n",
       " 269.06500000000005,\n",
       " 268.96299999999997,\n",
       " 275.443,\n",
       " 267.79400000000004,\n",
       " 277.964,\n",
       " 277.966,\n",
       " 274.668,\n",
       " 279.915,\n",
       " 267.2129999999999,\n",
       " 287.6430000000001,\n",
       " 269.1819999999999,\n",
       " 284.1309999999999,\n",
       " 274.74800000000016,\n",
       " 277.842,\n",
       " 272.8990000000001,\n",
       " 273.217,\n",
       " 280.07499999999993,\n",
       " 278.75899999999996,\n",
       " 274.31600000000003,\n",
       " 276.899,\n",
       " 268.75099999999986,\n",
       " 278.26300000000015,\n",
       " 276.60799999999995,\n",
       " 278.5439999999999,\n",
       " 285.99600000000015,\n",
       " 281.1530000000001,\n",
       " 278.70000000000016,\n",
       " 282.76300000000003,\n",
       " 284.45099999999996,\n",
       " 275.58899999999994,\n",
       " 277.2279999999999,\n",
       " 282.5589999999999,\n",
       " 283.87599999999986,\n",
       " 285.09400000000005,\n",
       " 284.057,\n",
       " 289.0869999999999,\n",
       " 283.941,\n",
       " 279.89199999999994,\n",
       " 277.09900000000005,\n",
       " 289.2919999999998,\n",
       " 282.07099999999997,\n",
       " 287.0469999999999,\n",
       " 290.152,\n",
       " 285.60800000000006,\n",
       " 285.95099999999996,\n",
       " 286.2060000000001,\n",
       " 288.1359999999999,\n",
       " 283.24399999999986,\n",
       " 286.5740000000001,\n",
       " 284.43600000000004,\n",
       " 287.7059999999999,\n",
       " 290.427,\n",
       " 290.54,\n",
       " 290.44399999999996,\n",
       " 288.536,\n",
       " 295.1349999999999,\n",
       " 290.599,\n",
       " 286.9310000000001,\n",
       " 288.344,\n",
       " 283.018,\n",
       " 292.7989999999999,\n",
       " 287.61799999999994,\n",
       " 286.4180000000001,\n",
       " 288.81200000000007,\n",
       " 297.75600000000003,\n",
       " 296.03400000000005,\n",
       " 288.994,\n",
       " 282.0310000000001,\n",
       " 293.71100000000007,\n",
       " 284.612,\n",
       " 294.389,\n",
       " 286.78499999999997,\n",
       " 294.29400000000004,\n",
       " 282.43399999999997,\n",
       " 291.70700000000005,\n",
       " 285.37300000000005,\n",
       " 286.1299999999998,\n",
       " 292.7749999999999,\n",
       " 288.18600000000004,\n",
       " 293.0220000000001,\n",
       " 290.4959999999999,\n",
       " 286.03299999999984,\n",
       " 283.31800000000004,\n",
       " 276.5959999999999]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fitness_y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 642,
   "id": "13c1b33e",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[19500,\n",
       " 39000,\n",
       " 58500,\n",
       " 78000,\n",
       " 97500,\n",
       " 117000,\n",
       " 136500,\n",
       " 156000,\n",
       " 175500,\n",
       " 195000,\n",
       " 214500,\n",
       " 234000,\n",
       " 253500,\n",
       " 273000,\n",
       " 292500,\n",
       " 312000,\n",
       " 331500,\n",
       " 351000,\n",
       " 370500,\n",
       " 390000,\n",
       " 409500,\n",
       " 429000,\n",
       " 448500,\n",
       " 468000,\n",
       " 487500,\n",
       " 507000,\n",
       " 526500,\n",
       " 546000,\n",
       " 565500,\n",
       " 585000,\n",
       " 604500,\n",
       " 624000,\n",
       " 643500,\n",
       " 663000,\n",
       " 682500,\n",
       " 702000,\n",
       " 721500,\n",
       " 741000,\n",
       " 760500,\n",
       " 780000,\n",
       " 799500,\n",
       " 819000,\n",
       " 838500,\n",
       " 858000,\n",
       " 877500,\n",
       " 897000,\n",
       " 916500,\n",
       " 936000,\n",
       " 955500,\n",
       " 975000,\n",
       " 994500,\n",
       " 1014000,\n",
       " 1033500,\n",
       " 1053000,\n",
       " 1072500,\n",
       " 1092000,\n",
       " 1111500,\n",
       " 1131000,\n",
       " 1150500,\n",
       " 1170000,\n",
       " 1189500,\n",
       " 1209000,\n",
       " 1228500,\n",
       " 1248000,\n",
       " 1267500,\n",
       " 1287000,\n",
       " 1306500,\n",
       " 1326000,\n",
       " 1345500,\n",
       " 1365000,\n",
       " 1384500,\n",
       " 1404000,\n",
       " 1423500,\n",
       " 1443000,\n",
       " 1462500]"
      ]
     },
     "execution_count": 642,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "steps_x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "7d38d37f",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "eb1238ff",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1080x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,15))\n",
    "plt.plot(steps_x, fitness_y, label=\"Predator Prey - Genetic Algorithm\")\n",
    "plt.xlabel(\"Time\")\n",
    "plt.ylabel(\"Avarage Reward\")\n",
    "plt.grid(True)\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "f63519a7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<__main__.NeuralModel at 0x202affd2460>"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "models[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "d098a14a",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_trained = models[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3b40010f",
   "metadata": {},
   "source": [
    "## Save model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "a29b0c85",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: Training/my_model_PP_3_4M_GA\\assets\n"
     ]
    }
   ],
   "source": [
    "path_best_model = 'Training/my_model_PP_3_4M_GA'\n",
    "model_trained.model.save(path_best_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cf87363b",
   "metadata": {},
   "source": [
    "## Load model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "ae0a1548",
   "metadata": {},
   "outputs": [],
   "source": [
    "path_best_model = 'Training/my_model_PP_3_4M_GA'\n",
    "model = keras.models.load_model(path_best_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "1dfd2e78",
   "metadata": {},
   "outputs": [],
   "source": [
    "obs = env.reset()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "9375fd6a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[2, 2, 2, 2, 2],\n",
       "       [0, 2, 2, 0, 2],\n",
       "       [2, 0, 1, 0, 2],\n",
       "       [2, 0, 0, 2, 0],\n",
       "       [0, 0, 2, 2, 0]])"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''\n",
    "0 - direita\n",
    "1 - esquerda\n",
    "2 - cima\n",
    "3 - baixo\n",
    "'''\n",
    "\n",
    "obs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "3070f252",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0. 0. 0. 1.]] 3\n"
     ]
    }
   ],
   "source": [
    "print(model_trained(obs.reshape(1, -1)).numpy(), model_trained(obs.reshape(1, -1)).numpy().argmax())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "a073874b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 0, 2, 0, 2],\n",
       "       [0, 0, 0, 2, 0],\n",
       "       [2, 0, 1, 0, 0],\n",
       "       [0, 0, 0, 0, 2],\n",
       "       [2, 0, 2, 0, 0]])"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "env.reset()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "b0d0c272",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[3, 2, 2, 3, 0, 0, 3, 1, 3, 2]"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[env.action_space.sample() for i in range(10)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "4a5aab7f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 1 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 1 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 2 0 2]] \n",
      "\n",
      "[0, 0, 0, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 2 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]] \n",
      "\n",
      "[0, 0, 2, 0]\n",
      "------------------------------\n",
      "2\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 2 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]] \n",
      "\n",
      "[0, 0, 2, 0]\n",
      "------------------------------\n",
      "2\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 2 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]] \n",
      "\n",
      "[0, 0, 2, 0]\n",
      "------------------------------\n",
      "2\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 2 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]] \n",
      "\n",
      "[0, 0, 2, 0]\n",
      "------------------------------\n",
      "2\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 2 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]] \n",
      "\n",
      "[0, 0, 2, 0]\n",
      "------------------------------\n",
      "2\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 2 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]] \n",
      "\n",
      "[0, 0, 2, 0]\n",
      "------------------------------\n",
      "2\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 2 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]] \n",
      "\n",
      "[0, 0, 2, 0]\n",
      "------------------------------\n",
      "2\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 2 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]] \n",
      "\n",
      "[0, 0, 2, 0]\n",
      "------------------------------\n",
      "2\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n",
      "3\n",
      "[[0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 2 0 0]] \n",
      "\n",
      "[[0 0 2 0 2]\n",
      " [2 0 2 0 0]\n",
      " [0 0 1 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]] \n",
      "\n",
      "[0, 0, 2, 0]\n",
      "------------------------------\n",
      "2\n",
      "[[0 0 2 0 2]\n",
      " [0 0 0 2 0]\n",
      " [2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]] \n",
      "\n",
      "[[2 0 0 0 0]\n",
      " [0 0 2 0 2]\n",
      " [2 0 1 0 0]\n",
      " [0 0 2 0 2]\n",
      " [0 0 0 2 0]] \n",
      "\n",
      "[0, 0, 2, 2]\n",
      "------------------------------\n"
     ]
    }
   ],
   "source": [
    "for i in range(20):\n",
    "    action = model_trained(obs.reshape(1, -1)).numpy().argmax()\n",
    "    print(action)\n",
    "    obs, _, _, _ = env.step(action)\n",
    "    print(env.array, '\\n')\n",
    "    print(env.get_obs(), '\\n')\n",
    "    print(env.get_neighboors())\n",
    "    print(\"-\" * 30)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 575,
   "id": "becb27c7",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ep: 0 || Score: 766\n",
      "Ep: 1 || Score: 608\n",
      "Ep: 2 || Score: 400\n",
      "Ep: 3 || Score: 296\n",
      "Ep: 4 || Score: 1181\n",
      "Ep: 5 || Score: 369\n",
      "Ep: 6 || Score: 1186\n",
      "Ep: 7 || Score: 600\n",
      "Ep: 8 || Score: 540\n",
      "Ep: 9 || Score: 602\n",
      "Ep: 10 || Score: 804\n",
      "Ep: 11 || Score: 1208\n",
      "Ep: 12 || Score: 804\n",
      "Ep: 13 || Score: 1072\n",
      "Ep: 14 || Score: 598\n",
      "Ep: 15 || Score: 379\n",
      "Ep: 16 || Score: 607\n",
      "Ep: 17 || Score: 452\n",
      "Ep: 18 || Score: 906\n",
      "Ep: 19 || Score: 1130\n",
      "Ep: 20 || Score: 296\n",
      "Ep: 21 || Score: 777\n",
      "Ep: 22 || Score: 796\n",
      "Ep: 23 || Score: 785\n",
      "Ep: 24 || Score: 668\n",
      "Ep: 25 || Score: 988\n",
      "Ep: 26 || Score: 864\n",
      "Ep: 27 || Score: 611\n",
      "Ep: 28 || Score: 600\n",
      "Ep: 29 || Score: 489\n",
      "Ep: 30 || Score: 586\n",
      "Ep: 31 || Score: 848\n",
      "Ep: 32 || Score: 982\n",
      "Ep: 33 || Score: 404\n",
      "Ep: 34 || Score: 634\n",
      "Ep: 35 || Score: 596\n",
      "Ep: 36 || Score: 920\n",
      "Ep: 37 || Score: 484\n",
      "Ep: 38 || Score: 1128\n",
      "Ep: 39 || Score: 961\n",
      "Ep: 40 || Score: 512\n",
      "Ep: 41 || Score: 528\n",
      "Ep: 42 || Score: 770\n",
      "Ep: 43 || Score: 976\n",
      "Ep: 44 || Score: 800\n",
      "Ep: 45 || Score: 591\n",
      "Ep: 46 || Score: 988\n",
      "Ep: 47 || Score: 1116\n",
      "Ep: 48 || Score: 598\n",
      "Ep: 49 || Score: 537\n",
      "Ep: 50 || Score: 595\n",
      "Ep: 51 || Score: 1155\n",
      "Ep: 52 || Score: 1158\n",
      "Ep: 53 || Score: 486\n",
      "Ep: 54 || Score: 764\n",
      "Ep: 55 || Score: 988\n",
      "Ep: 56 || Score: 1008\n",
      "Ep: 57 || Score: 925\n",
      "Ep: 58 || Score: 800\n",
      "Ep: 59 || Score: 591\n",
      "Ep: 60 || Score: 868\n",
      "Ep: 61 || Score: 449\n",
      "Ep: 62 || Score: 710\n",
      "Ep: 63 || Score: 504\n",
      "Ep: 64 || Score: 498\n",
      "Ep: 65 || Score: 647\n",
      "Ep: 66 || Score: 1528\n",
      "Ep: 67 || Score: 610\n",
      "Ep: 68 || Score: 455\n",
      "Ep: 69 || Score: 1006\n",
      "Ep: 70 || Score: 356\n",
      "Ep: 71 || Score: 445\n",
      "Ep: 72 || Score: 906\n",
      "Ep: 73 || Score: 709\n",
      "Ep: 74 || Score: 1168\n",
      "Ep: 75 || Score: 581\n",
      "Ep: 76 || Score: 578\n",
      "Ep: 77 || Score: 501\n",
      "Ep: 78 || Score: 1002\n",
      "Ep: 79 || Score: 670\n",
      "Ep: 80 || Score: 682\n",
      "Ep: 81 || Score: 706\n",
      "Ep: 82 || Score: 581\n",
      "Ep: 83 || Score: 506\n",
      "Ep: 84 || Score: 536\n",
      "Ep: 85 || Score: 594\n",
      "Ep: 86 || Score: 485\n",
      "Ep: 87 || Score: 1194\n",
      "Ep: 88 || Score: 556\n",
      "Ep: 89 || Score: 914\n",
      "Ep: 90 || Score: 1008\n",
      "Ep: 91 || Score: 389\n",
      "Ep: 92 || Score: 458\n",
      "Ep: 93 || Score: 667\n",
      "Ep: 94 || Score: 310\n",
      "Ep: 95 || Score: 500\n",
      "Ep: 96 || Score: 984\n",
      "Ep: 97 || Score: 606\n",
      "Ep: 98 || Score: 457\n",
      "Ep: 99 || Score: 612\n",
      "Avg score: 717.47\n"
     ]
    }
   ],
   "source": [
    "episodes = 100\n",
    "avg_total = 0\n",
    "for ep in range(episodes):\n",
    "    obs = env.reset().reshape(1, -1)\n",
    "    done = False\n",
    "    score = 0\n",
    "    while not done:\n",
    "        action = model(obs.reshape(1, -1)).numpy().argmax()\n",
    "        obs, rew, done, info = env.step(action)\n",
    "        if render:\n",
    "            env.render()\n",
    "        score += rew\n",
    "    avg_total += score\n",
    "    print(f\"Ep: {ep} || Score: {score}\")\n",
    "print(f\"Avg score: {avg_total / episodes}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 576,
   "id": "aa372ed3",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ep: 0 || Score: 537\n",
      "Ep: 1 || Score: 469\n",
      "Ep: 2 || Score: 523\n",
      "Ep: 3 || Score: 561\n",
      "Ep: 4 || Score: 660\n",
      "Ep: 5 || Score: 504\n",
      "Ep: 6 || Score: 572\n",
      "Ep: 7 || Score: 513\n",
      "Ep: 8 || Score: 446\n",
      "Ep: 9 || Score: 438\n",
      "Ep: 10 || Score: 688\n",
      "Ep: 11 || Score: 558\n",
      "Ep: 12 || Score: 543\n",
      "Ep: 13 || Score: 561\n",
      "Ep: 14 || Score: 454\n",
      "Ep: 15 || Score: 505\n",
      "Ep: 16 || Score: 496\n",
      "Ep: 17 || Score: 529\n",
      "Ep: 18 || Score: 471\n",
      "Ep: 19 || Score: 417\n",
      "Ep: 20 || Score: 897\n",
      "Ep: 21 || Score: 456\n",
      "Ep: 22 || Score: 451\n",
      "Ep: 23 || Score: 624\n",
      "Ep: 24 || Score: 645\n",
      "Ep: 25 || Score: 601\n",
      "Ep: 26 || Score: 505\n",
      "Ep: 27 || Score: 490\n",
      "Ep: 28 || Score: 530\n",
      "Ep: 29 || Score: 644\n",
      "Ep: 30 || Score: 496\n",
      "Ep: 31 || Score: 521\n",
      "Ep: 32 || Score: 565\n",
      "Ep: 33 || Score: 410\n",
      "Ep: 34 || Score: 388\n",
      "Ep: 35 || Score: 588\n",
      "Ep: 36 || Score: 407\n",
      "Ep: 37 || Score: 653\n",
      "Ep: 38 || Score: 477\n",
      "Ep: 39 || Score: 353\n",
      "Ep: 40 || Score: 613\n",
      "Ep: 41 || Score: 476\n",
      "Ep: 42 || Score: 525\n",
      "Ep: 43 || Score: 464\n",
      "Ep: 44 || Score: 384\n",
      "Ep: 45 || Score: 572\n",
      "Ep: 46 || Score: 497\n",
      "Ep: 47 || Score: 499\n",
      "Ep: 48 || Score: 559\n",
      "Ep: 49 || Score: 539\n",
      "Ep: 50 || Score: 428\n",
      "Ep: 51 || Score: 454\n",
      "Ep: 52 || Score: 303\n",
      "Ep: 53 || Score: 503\n",
      "Ep: 54 || Score: 656\n",
      "Ep: 55 || Score: 551\n",
      "Ep: 56 || Score: 421\n",
      "Ep: 57 || Score: 590\n",
      "Ep: 58 || Score: 376\n",
      "Ep: 59 || Score: 448\n",
      "Ep: 60 || Score: 422\n",
      "Ep: 61 || Score: 676\n",
      "Ep: 62 || Score: 692\n",
      "Ep: 63 || Score: 506\n",
      "Ep: 64 || Score: 473\n",
      "Ep: 65 || Score: 552\n",
      "Ep: 66 || Score: 332\n",
      "Ep: 67 || Score: 442\n",
      "Ep: 68 || Score: 499\n",
      "Ep: 69 || Score: 452\n",
      "Ep: 70 || Score: 588\n",
      "Ep: 71 || Score: 412\n",
      "Ep: 72 || Score: 639\n",
      "Ep: 73 || Score: 493\n",
      "Ep: 74 || Score: 331\n",
      "Ep: 75 || Score: 405\n",
      "Ep: 76 || Score: 558\n",
      "Ep: 77 || Score: 484\n",
      "Ep: 78 || Score: 592\n",
      "Ep: 79 || Score: 498\n",
      "Ep: 80 || Score: 511\n",
      "Ep: 81 || Score: 407\n",
      "Ep: 82 || Score: 653\n",
      "Ep: 83 || Score: 489\n",
      "Ep: 84 || Score: 499\n",
      "Ep: 85 || Score: 513\n",
      "Ep: 86 || Score: 689\n",
      "Ep: 87 || Score: 353\n",
      "Ep: 88 || Score: 642\n",
      "Ep: 89 || Score: 531\n",
      "Ep: 90 || Score: 504\n",
      "Ep: 91 || Score: 635\n",
      "Ep: 92 || Score: 481\n",
      "Ep: 93 || Score: 501\n",
      "Ep: 94 || Score: 642\n",
      "Ep: 95 || Score: 470\n",
      "Ep: 96 || Score: 470\n",
      "Ep: 97 || Score: 451\n",
      "Ep: 98 || Score: 556\n",
      "Ep: 99 || Score: 581\n",
      "Avg score: 516.28\n"
     ]
    }
   ],
   "source": [
    "episodes = 100\n",
    "avg_total = 0\n",
    "for ep in range(episodes):\n",
    "    obs = env.reset().reshape(1, -1)\n",
    "    done = False\n",
    "    score = 0\n",
    "    while not done:\n",
    "        action = model(obs.reshape(1, -1)).numpy().argmax()\n",
    "        obs, rew, done, info = env.step(env.action_space.sample())\n",
    "        if render:\n",
    "            env.render()\n",
    "        score += rew\n",
    "    avg_total += score\n",
    "    print(f\"Ep: {ep} || Score: {score}\")\n",
    "print(f\"Avg score: {avg_total / episodes}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "d345a471",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\gabri\\anaconda3\\lib\\site-packages\\torch\\utils\\tensorboard\\__init__.py:4: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.\n",
      "  if not hasattr(tensorboard, \"__version__\") or LooseVersion(\n"
     ]
    }
   ],
   "source": [
    "### Carregando modelo\n",
    "import os\n",
    "from stable_baselines3 import PPO\n",
    "from stable_baselines3.common.vec_env import DummyVecEnv\n",
    "from stable_baselines3.common.evaluation import evaluate_policy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "8d35b023",
   "metadata": {},
   "outputs": [],
   "source": [
    "PPO_path = os.path.join('Training', 'Saved Models', 'PP_PPO_1_2M')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "f941fe31",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = PPO.load(PPO_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "0dba0fce",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<stable_baselines3.ppo.ppo.PPO at 0x202b4718b50>"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "b5015ee8",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ep: 0 || Score: 192\n",
      "Ep: 1 || Score: 186\n",
      "Ep: 2 || Score: 137\n",
      "Ep: 3 || Score: 266\n",
      "Ep: 4 || Score: 129\n",
      "Ep: 5 || Score: 255\n",
      "Ep: 6 || Score: 141\n",
      "Ep: 7 || Score: 211\n",
      "Ep: 8 || Score: 231\n",
      "Ep: 9 || Score: 286\n",
      "Ep: 10 || Score: 153\n",
      "Ep: 11 || Score: 131\n",
      "Ep: 12 || Score: 190\n",
      "Ep: 13 || Score: 268\n",
      "Ep: 14 || Score: 191\n",
      "Ep: 15 || Score: 420\n",
      "Ep: 16 || Score: 236\n",
      "Ep: 17 || Score: 117\n",
      "Ep: 18 || Score: 290\n",
      "Ep: 19 || Score: 129\n",
      "Ep: 20 || Score: 175\n",
      "Ep: 21 || Score: 274\n",
      "Ep: 22 || Score: 152\n",
      "Ep: 23 || Score: 146\n",
      "Ep: 24 || Score: 161\n",
      "Ep: 25 || Score: 169\n",
      "Ep: 26 || Score: 169\n",
      "Ep: 27 || Score: 223\n",
      "Ep: 28 || Score: 256\n",
      "Ep: 29 || Score: 201\n",
      "Ep: 30 || Score: 246\n",
      "Ep: 31 || Score: 202\n",
      "Ep: 32 || Score: 158\n",
      "Ep: 33 || Score: 155\n",
      "Ep: 34 || Score: 255\n",
      "Ep: 35 || Score: 120\n",
      "Ep: 36 || Score: 254\n",
      "Ep: 37 || Score: 237\n",
      "Ep: 38 || Score: 307\n",
      "Ep: 39 || Score: 229\n",
      "Ep: 40 || Score: 246\n",
      "Ep: 41 || Score: 187\n",
      "Ep: 42 || Score: 224\n",
      "Ep: 43 || Score: 312\n",
      "Ep: 44 || Score: 220\n",
      "Ep: 45 || Score: 183\n",
      "Ep: 46 || Score: 126\n",
      "Ep: 47 || Score: 277\n",
      "Ep: 48 || Score: 217\n",
      "Ep: 49 || Score: 236\n",
      "Ep: 50 || Score: 126\n",
      "Ep: 51 || Score: 268\n",
      "Ep: 52 || Score: 170\n",
      "Ep: 53 || Score: 182\n",
      "Ep: 54 || Score: 118\n",
      "Ep: 55 || Score: 200\n",
      "Ep: 56 || Score: 125\n",
      "Ep: 57 || Score: 135\n",
      "Ep: 58 || Score: 165\n",
      "Ep: 59 || Score: 173\n",
      "Ep: 60 || Score: 288\n",
      "Ep: 61 || Score: 202\n",
      "Ep: 62 || Score: 149\n",
      "Ep: 63 || Score: 82\n",
      "Ep: 64 || Score: 268\n",
      "Ep: 65 || Score: 253\n",
      "Ep: 66 || Score: 154\n",
      "Ep: 67 || Score: 162\n",
      "Ep: 68 || Score: 156\n",
      "Ep: 69 || Score: 161\n",
      "Ep: 70 || Score: 274\n",
      "Ep: 71 || Score: 210\n",
      "Ep: 72 || Score: 272\n",
      "Ep: 73 || Score: 234\n",
      "Ep: 74 || Score: 176\n",
      "Ep: 75 || Score: 248\n",
      "Ep: 76 || Score: 221\n",
      "Ep: 77 || Score: 282\n",
      "Ep: 78 || Score: 304\n",
      "Ep: 79 || Score: 171\n",
      "Ep: 80 || Score: 152\n",
      "Ep: 81 || Score: 128\n",
      "Ep: 82 || Score: 122\n",
      "Ep: 83 || Score: 334\n",
      "Ep: 84 || Score: 145\n",
      "Ep: 85 || Score: 242\n",
      "Ep: 86 || Score: 215\n",
      "Ep: 87 || Score: 192\n",
      "Ep: 88 || Score: 115\n",
      "Ep: 89 || Score: 217\n",
      "Ep: 90 || Score: 180\n",
      "Ep: 91 || Score: 192\n",
      "Ep: 92 || Score: 140\n",
      "Ep: 93 || Score: 214\n",
      "Ep: 94 || Score: 181\n",
      "Ep: 95 || Score: 255\n",
      "Ep: 96 || Score: 153\n",
      "Ep: 97 || Score: 138\n",
      "Ep: 98 || Score: 310\n",
      "Ep: 99 || Score: 192\n",
      "Avg score: 202.22\n"
     ]
    }
   ],
   "source": [
    "episodes = 100\n",
    "avg_total = 0\n",
    "for ep in range(episodes):\n",
    "    obs = env.reset().reshape(1, -1)\n",
    "    done = False\n",
    "    score = 0\n",
    "    while not done:\n",
    "        action = model.predict(obs.reshape(1, -1))[0]\n",
    "        obs, rew, done, info = env.step(env.action_space.sample())\n",
    "        if render:\n",
    "            env.render()\n",
    "        score += rew\n",
    "    avg_total += score\n",
    "    print(f\"Ep: {ep} || Score: {score}\")\n",
    "print(f\"Avg score: {avg_total / episodes}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "670d5900",
   "metadata": {},
   "source": [
    "## Conclusion:\n",
    "\n",
    "We use the Genetic Algorithm and this work well. But, we can try to implemment another model to optimization policy and verify if our model to reach more average reward.\n",
    "\n",
    "So, let me go ahead and try to Implemment PPO algorithm in another Notebook! :)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "67f4b974",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.12"
  },
  "vscode": {
   "interpreter": {
    "hash": "f0772a89b7b7552231f8b4808588399a8029470ff763161a0caeccf5fec49431"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}