example_of_usage.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### The following notebook demonstrates usage of library"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Firstly, proceed all useful imports."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "import torch\n",
    "\n",
    "from models import AgentA3C, AgentA2C\n",
    "\n",
    "import gym\n",
    "import time\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And define whether learning will be done on CPU or GPU."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "USE_GPU = False and torch.cuda.is_available()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Set random seed for reproducibility:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "SEED = 777\n",
    "np.random.seed(SEED)\n",
    "torch.manual_seed(SEED)\n",
    "if USE_GPU:\n",
    "    torch.cuda.manual_seed(SEED)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And define set of Atari games with different environment:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "games = [\n",
    "    'FrozenLake-v0',\n",
    "    'CartPole-v0',\n",
    "    'LunarLander-v2'\n",
    "]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then define game and model parameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "i_game = 0\n",
    "\n",
    "discount_gamma = 0.99\n",
    "n_episodes = 10*1000\n",
    "n_agents = 10\n",
    "\n",
    "env = gym.make(games[i_game])\n",
    "envs = [gym.make(games[i_game]) for _ in range(n_agents)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets create agent A2C and let him learn to play chosen game:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "agentA2C = AgentA2C(env, gamma=discount_gamma, lr=1e-4, save_returns=True)\n",
    "if USE_GPU:\n",
    "    agentA2C = agentA2C.cuda()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "episode    500 [act: -0.2115, crt: 57.6915, ent: 2.0275 rew_e: 0.005077], 2.5 ms/ep\n",
      "episode   1000 [act: -0.7905, crt: 2.1479, ent: 2.0772 rew_e: 0.001703], 2.2 ms/ep\n",
      "episode   1500 [act: -0.3336, crt: 0.8220, ent: 2.1989 rew_e: 0.003443], 2.1 ms/ep\n",
      "episode   2000 [act: -0.2686, crt: 0.4950, ent: 2.1631 rew_e: 0.005289], 2.1 ms/ep\n",
      "episode   2500 [act: -0.2929, crt: 0.2878, ent: 2.4460 rew_e: 0.007031], 2.2 ms/ep\n",
      "episode   3000 [act: -0.2736, crt: 0.2070, ent: 2.5178 rew_e: 0.015635], 2.2 ms/ep\n",
      "episode   3500 [act: -0.3471, crt: 0.3144, ent: 2.6941 rew_e: 0.015599], 2.2 ms/ep\n",
      "episode   4000 [act: -0.4274, crt: 0.3838, ent: 2.8442 rew_e: 0.024340], 2.2 ms/ep\n",
      "episode   4500 [act: -0.6159, crt: 0.5561, ent: 2.9954 rew_e: 0.033295], 2.2 ms/ep\n",
      "episode   5000 [act: -0.6991, crt: 0.7157, ent: 3.0551 rew_e: 0.045412], 2.3 ms/ep\n",
      "episode   5500 [act: -0.6783, crt: 0.7174, ent: 2.8506 rew_e: 0.042356], 2.2 ms/ep\n",
      "episode   6000 [act: -0.5855, crt: 0.5015, ent: 2.9326 rew_e: 0.033037], 2.2 ms/ep\n",
      "episode   6500 [act: -0.7843, crt: 0.7324, ent: 3.2140 rew_e: 0.044045], 2.4 ms/ep\n",
      "episode   7000 [act: -0.8947, crt: 0.8170, ent: 3.0807 rew_e: 0.049913], 2.4 ms/ep\n",
      "episode   7500 [act: -0.6246, crt: 0.6051, ent: 3.0461 rew_e: 0.039960], 2.4 ms/ep\n",
      "episode   8000 [act: -0.8163, crt: 0.7755, ent: 3.1437 rew_e: 0.058076], 2.5 ms/ep\n",
      "episode   8500 [act: -0.9095, crt: 0.7859, ent: 3.2153 rew_e: 0.060319], 2.7 ms/ep\n",
      "episode   9000 [act: -0.8359, crt: 0.8552, ent: 3.2905 rew_e: 0.050856], 2.6 ms/ep\n",
      "episode   9500 [act: -0.8520, crt: 0.8937, ent: 3.1258 rew_e: 0.056853], 2.5 ms/ep\n",
      "episode  10000 [act: -1.1388, crt: 1.0138, ent: 3.3810 rew_e: 0.076521], 2.7 ms/ep\n",
      "Trained in 0.39 min\n"
     ]
    }
   ],
   "source": [
    "start = time.time()\n",
    "agentA2C.learn(n_episodes, n_steps=10, k_bootstrap=5, verbosity=500)\n",
    "print('Trained in {} min'.format(round((time.time() - start) / 60, 2)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Do the same actions with agent A3C:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "agentA3C = AgentA3C(envs, gamma=discount_gamma, lr=1e-4, save_returns=True)\n",
    "if USE_GPU:\n",
    "    agentA3C = agentA3C.cuda()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "episode    500 [act: 39.7597, crt: 36.0758, ent: 0.1489 rew_e: 0.025249], 10.3 ms/ep\n",
      "episode   1000 [act: 24.8768, crt: 24.8341, ent: 0.0907 rew_e: 0.027793], 13.1 ms/ep\n",
      "episode   1500 [act: 20.5710, crt: 19.3856, ent: 0.0784 rew_e: 0.028329], 12.2 ms/ep\n",
      "episode   2000 [act: 17.2013, crt: 15.6290, ent: 0.0731 rew_e: 0.026387], 13.5 ms/ep\n",
      "episode   2500 [act: 18.2714, crt: 16.5277, ent: 0.0754 rew_e: 0.034329], 14.7 ms/ep\n",
      "episode   3000 [act: 18.7764, crt: 17.2411, ent: 0.0751 rew_e: 0.053071], 15.8 ms/ep\n",
      "episode   3500 [act: 17.4256, crt: 15.9330, ent: 0.0708 rew_e: 0.062121], 14.6 ms/ep\n",
      "episode   4000 [act: 14.4060, crt: 13.7856, ent: 0.0608 rew_e: 0.077320], 15.9 ms/ep\n",
      "episode   4500 [act: 11.4899, crt: 10.3682, ent: 0.0519 rew_e: 0.072366], 16.6 ms/ep\n",
      "episode   5000 [act: 10.0895, crt: 9.4336, ent: 0.0485 rew_e: 0.076761], 17.0 ms/ep\n",
      "episode   5500 [act: 8.5990, crt: 8.1751, ent: 0.0459 rew_e: 0.089617], 17.4 ms/ep\n",
      "episode   6000 [act: 8.5338, crt: 8.3666, ent: 0.0464 rew_e: 0.087570], 17.5 ms/ep\n",
      "episode   6500 [act: 8.2364, crt: 7.8605, ent: 0.0458 rew_e: 0.096345], 17.6 ms/ep\n",
      "episode   7000 [act: 7.9797, crt: 7.8640, ent: 0.0474 rew_e: 0.101545], 18.0 ms/ep\n",
      "episode   7500 [act: 7.8239, crt: 7.2755, ent: 0.0456 rew_e: 0.130797], 17.7 ms/ep\n",
      "episode   8000 [act: 7.3822, crt: 7.0124, ent: 0.0444 rew_e: 0.135303], 17.4 ms/ep\n",
      "episode   8500 [act: 7.8480, crt: 7.2282, ent: 0.0454 rew_e: 0.134783], 16.4 ms/ep\n",
      "episode   9000 [act: 6.8548, crt: 6.6377, ent: 0.0423 rew_e: 0.159729], 17.5 ms/ep\n",
      "episode   9500 [act: 6.3263, crt: 6.0598, ent: 0.0393 rew_e: 0.188535], 16.8 ms/ep\n",
      "episode  10000 [act: 5.9401, crt: 5.4667, ent: 0.0363 rew_e: 0.203093], 17.2 ms/ep\n",
      "Trained in 2.64 min\n"
     ]
    }
   ],
   "source": [
    "start = time.time()\n",
    "agentA3C.learn(n_agents, n_episodes, n_steps=10, k_bootstrap=5, verbosity=500)\n",
    "print('Trained in {} min'.format(round((time.time() - start) / 60, 2)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see debug information with actor gain, critic loss, entropy and mean obtained reward for bunch of episodes. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now lets take a look at graphics of returns by episodes of two trained agents:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niBAhsiIUFiFChAgRIiv+PyQnMHCyA2NdAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11810c6a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "window_size = 50\n",
    "plt.title('Cumulative discounted rewards (smoothed)')\n",
    "plt.xlabel('Episode')\n",
    "plt.ylabel('Return')\n",
    "plt.grid()\n",
    "plt.plot(np.convolve(\n",
    "    agentA2C.returns_over_episodes, np.ones(window_size) / window_size, mode='valid'\n",
    "), label='A2C')\n",
    "\n",
    "plt.plot(np.convolve(\n",
    "    agentA3C.returns_over_episodes, np.ones(window_size) / window_size, mode='valid'\n",
    "), label='A3C')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Also we can see trained policy for both agents:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.colorbar.Colorbar at 0x1182d9630>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAASYAAAMaCAYAAAAr4+slAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzt3Xu0ZHV55vHvQzfIpUHAVsOlFZaD\nSYiamLTkYsY4opOWRHGWTgJRVySOPbOyNDrRGMxFHHM3s3JZGXPpREUNQhCMYQyKxqBOElEaRCIg\nBhGlEUVAEQRpuvudP6rQY+f0uUjX2W+d+n7WqkVd9tn1Huh+eN9f7V07VYUkdbLP0AVI0u4MJknt\nGEyS2jGYJLVjMElqx2CS1I7B1ESS1yT56/H9RyS5K8maTnVJK8Vg2suS3JDknnGwfDHJmUnWLWcf\nVfW5qlpXVTsnVafUmcE0Gc+oqnXA9wMbgV8buJ4Vl2Tt0DVoehlME1RVNwHvBh4DkOTIJBckuT3J\ndUleNN/PJTkmSd3/lzvJ4UnelOTzSb6c5J3j5z+R5Blzfm7fJLcmefwe9ntykiuSfDXJp5NsWk5d\n422fmeSqJF9J8oEk3z3ntRuS/HKSK4GvJVk7fu6XklyZ5GtJ3pDk4UneneTOJP+Q5LA5+3h7ki8k\nuSPJh5J8z5zXzkzy50neN/7ZDyZ55JL+Y2iqGEwTlGQDcBLwsfFT5wDbgCOB5wC/neQpS9jVW4ED\nge8BHgb84fj5twDPm7PdScDNVfUxdpPkhPH2vwQcCjwJuGE5dSV5NHA28DLgocCFwP9Nst+czU4F\nfgI4tKp2jJ97NvA04NHAMxiF9a+M97EP8Atzfv7dwHHj3/Ny4Kzdyngu8BvAeuCKeV7XalBV3vbi\njdFf9ruArwCfBf4UOADYAOwEDp6z7e8AZ47vvwb46/H9Y4AC1gJHALuAw+Z5ryOBO4FDxo/PA165\nh7r+AvjDeZ5fTl2/Dpw7Z7t9gJuAJ8/53X9unn8fz53z+Hzgz+Y8fgnwzj3UfOj438ODx4/PBM6Z\n8/q6ce0bhv7v7m3v3uyYJuNZVXVoVT2yqn6+qu5hFCK3V9Wdc7b7LHDUIvvaMP65L+/+QlV9Hvhn\n4NlJDgWezp47iA3Ap+d5fjl1HTl+7f733wXcuNu2N87zc1+cc/+eeR6vA0iyJsnvjsfMr/LNjm79\nfPuvqruA28d1aRUxmFbO54HDkxw857lHMOo4FnLj+OcO3cPrb2Y0zv1X4MM1Wtfa034e9QDr+jzw\njTWdJGEUeHO3fSBfV/EzwMnAU4EHM+ocATJnmw1z3n8dcPi4Lq0iBtMKqaobgX8BfifJ/kkeB7wQ\nWPAYoaq6mdG6y58mOWy8wP2kOZu8k9Gnfy9ltIa0J28ATktyYpJ9khyV5LuWWde5wE+M97Ev8HLg\n3vHP7w0Hj/d3G6M1td+eZ5uTkvzoeF3rN4BLxr+DVhGDaWWdyqgL+Dzwt8AZVfUPS/i55wP3AZ8E\nbmG0+AzAeEw8HzgWeMeedlBVHwVOY7RwfgfwQb7Z/Syprqq6llF39ifArYwWsp9RVduX8DssxVsY\njYo3AVcDl8yzzduAMxiNcD/Aty7+a5VIlV8UN+2SvBp4dFWt6r+kSc4EtlXVzB0XNms8CG7KJTmc\n0ej1/KFrkfYWR7kpNj4Q8kbg3VX1oaHrkfYWRzlJ7dgxSWrHYJLUjsEkqR2DSVI7BpOkdgwmSe0Y\nTJLaMZgktWMwSWrHYJLUjsEkqR2DSVI7BpOkdgwmSe0YTJLaMZgktWMwSWrHYJLUjsEkqR2DSVI7\nBpOkdmY2mJJsSnJtkuuSnD50PYtJ8sYktyT5xNC1LEWSDUkuTnJ1kquSvHTomhYzvkT6R5N8fFzz\n/xq6plk1k5dvSrIG+BTwNGAbcClwalVdPWhhC0jyJOAu4C1V9Zih61lMkiOAI6rq8iQHA5cBz2r+\n7zjAQVV1V5J9gX8CXlpV812qXBM0qx3TCcB1VXV9VW0HzgFOHrimBY0vaHn70HUsVVXdXFWXj+/f\nCVwDHDVsVQurkbvGD/cd32bv/9wNzGowHcXoCrb320bzvzTTLMkxwOOBjwxbyeKSrElyBXAL8L6q\nal/zajSrwaQVkmQdcD7wsqr66tD1LKaqdlbV9wFHAyckaT82r0azGkw3ARvmPD56/Jz2ovE6zfnA\nWVX1jqHrWY6q+gpwMbBp6Fpm0awG06XAcUmOTbIfcApwwcA1rSrjheQ3ANdU1R8MXc9SJHlokkPH\n9w9g9OHIJ4etajbNZDBV1Q7gxcBFjBZlz62qq4atamFJzgY+DHxnkm1JXjh0TYt4IvB84ClJrhjf\nThq6qEUcAVyc5EpG//N6X1W9a+CaZtJMHi4gqbeZ7Jgk9WYwSWrHYJLUjsEkqR2DSVI7Mx1MSTYP\nXcNyWfPkTVu9q9FMBxMwjX8ArXnypq3eVWfWg0lSQ1NxgOV++xxQB6w9eK/vd/uue9hvnwP2+n4B\ndq570ET2e9+9d7Hvg9bt9f3u3G+v7/Kb+77na6w54KC9vt997tvruwRgx9e/xtr99369AHffvu3W\nqnroRHa+iqwduoClOGDtwfzI+p8auoxlueNHjxm6hGW545g1Q5ewbOtu2jV0Ccv20be94rND1zAN\nHOUktWMwSWrHYJLUjsEkqR2DSVI7BpOkdgwmSe0YTJLaMZgktWMwSWrHYJLUjsEkqR2DSVI7BpOk\ndgwmSe0YTJLaGSSYkmxKcm2S65KcPkQNkvpa8WBKsgZ4PfB04Hjg1CTHr3QdkvoaomM6Abiuqq6v\nqu3AOcDJA9Qhqakhguko4MY5j7eNn5MkoPHFCMYXHdwMsP+avX9VEKm7H/9PB9Vtt+8ctIbLrrz3\noqratNLvO0Qw3QRsmPP46PFz36KqtgBbAB6838P6X2NK2stuu30nH73oEYPWsOaIf1s/xPsOMcpd\nChyX5Ngk+wGnABcMUIekpla8Y6qqHUleDFwErAHeWFVXrXQdUncF7GL6rp23NwyyxlRVFwIXDvHe\nkvrzyG9J7bT9VE5SsbNmc5SzY5LUjh2T1NRo8Xs2j5SxY5LUjsEkqR1HOamxWT2OyY5JUjt2TFJT\nRbGzXPyWpBYMJkntOMpJjXkckyQ1YTBJasdRTmqqgJ2OcpLUgx2T1NisLn5PRzDt3MWuO+8auorl\nmbI/T4d+etircXw7vvgEG/7Vyv+yktqZjo5JmkEFnpIiSV0YTJLacZSTGpvNb2OyY5LUkB2T1FRR\nHvktSV0YTJLacZSTuirYOZuTnB2TpH7smKSmRlfinU12TJLaMZgkteMoJ7UVdpKhixiEHZOkdgwm\nSe04yklNFbDL45gkqQc7JqkxF78lqQmDSVI7jnJSU6Mr8TrKSVILdkxSY7vKjkmSWjCYJLXjKCc1\n5eK3JDViMElqx1FOaqoIO2e0d5jN31pSa3ZMUmMexyRJTRhMktpxlJOa8jgmSWrEYJLUjqOc1FbY\nWbPZO8zmby2pNTsmqakCds1o7zCbv7Wk1gwmSe1Mxyi371pyxMOGrmJZth88XZl/76HTd7zMUR/c\nPnQJy/bpZW7vcUyS1MR0dEzSDKrycAFJasNgktSOo5zU2C4XvyWpB4NJUjuOclJTo+9jms3eYTZ/\na0mt2TFJbXkckyS1YTBJasdRTmrK72OSpEbsmKTGdnolXknqwWCS1I6jnNRUEY/8lqQuVjyYkmxI\ncnGSq5NcleSlK12DpN6GGOV2AC+vqsuTHAxcluR9VXX1ALVIre3ylJSVUVU3V9Xl4/t3AtcAR610\nHZL6GnTxO8kxwOOBj8zz2mZgM8D+aw9Z0bqkDvzakwEkWQecD7ysqr66++tVtaWqNlbVxv3WHLDy\nBUoazCDBlGRfRqF0VlW9Y4gaJPW14qNckgBvAK6pqj9Y6feXpkURT0lZQU8Eng88JckV49tJA9Qh\nqakV75iq6p9gRq9JI2lJPCVFaszvY5KkJuyYpKaq8GIEktSFwSSpHUc5qa2wa0Y/wLZjktSOHZPU\nVOHityS1YTBJasdRTmrM72OSpCYMJkntOMpJTRVhl9/HJEk92DFJjc3q4vdUBFPdu52d131m6DKW\n5YDveujQJSzLgV+qoUtYthv+yxT+pX3P0AVMhyn8LytptZuKjkmaRYVX4pWkNuyYpLbCTr/2RJJ6\nMJgkteMoJzXl4rckNWIwSWrHUU5qzE/lJKkJOyapqaq4+C1JXRhMkh6QJJuSXJvkuiSnz/P6I5Jc\nnORjSa5MctJi+3SUkxrrfl25JGuA1wNPA7YBlya5oKqunrPZrwHnVtWfJTkeuBA4ZqH99v6tJXV3\nAnBdVV1fVduBc4CTd9umgEPG9x8MfH6xndoxSXogjgJunPN4G/CDu23zGuC9SV4CHAQ8dbGd2jFJ\nTRWwiwx6A9Yn2Trntvnb+FVOBc6sqqOBk4C3Jlkwe+yYJC3k1qrauMDrNwEb5jw+evzcXC8ENgFU\n1YeT7A+sB27Z004NJqmttF/8Bi4FjktyLKNAOgX4md22+RxwInBmku8G9ge+tNBO2//Wkvqqqh3A\ni4GLgGsYffp2VZLXJnnmeLOXAy9K8nHgbOAFVbXg1S/smCQ9IFV1IaNDAOY+9+o5968GnricfRpM\nUlOj72PyJF5JasGOSWpsVq/EO5u/taTWDCZJ7TjKSU0VcfFbkrowmCS14ygnNbZrRnuH2fytJbVm\nxyQ1VQU7XfyWpB4MJkntOMpJjXkckyQ1YcckNTU68ns2e4fZ/K0ltWYwSWrHUU5qbCcufktSCwaT\npHYc5aSmZvliBFMRTI9+3N1cdNEVQ5exLD9+5NAVrH6P/vuhK1i+zw1dwJSYimCSZpPHMUlSGwaT\npHYc5aTGdnkckyT1YDBJasdRTmrKr9aVpEbsmKTGPI5JkpowmCS14ygnNTX6al0XvyWpBTsmqTGP\n/JakJgwmSe04yklNzfI3WNoxSWpnsGBKsibJx5K8a6gaJPU05Cj3UuAa4JABa5Ba85SUFZTkaOAn\ngL8a4v0l9TZUx/RHwCuBg/e0QZLNwGaARxzlGr1mUHnk94pJ8pPALVV12ULbVdWWqtpYVRsf+pA1\nK1SdpA6GGOWeCDwzyQ3AOcBTkvz1AHVIamrFZ6SqehXwKoAkTwZeUVXPW+k6pO4KT0mRpDYGXVWu\nqg8AHxiyBqkzF78lqQmDSVI7HiAkNeVJvJLUiMEkqR1HOakxRzlJasKOSWrKyzdJUiMGk6R2HOWk\nxjyJV5KaMJgkteMoJ3VVHsckSW3YMUlNzfJJvFMRTJ+68kB+/OgfGLqMZbnzp58wdAnLsv2Q6fsL\ncO+h01czrztv6AqmgqOcpHamomOSZtWsjnJ2TJLasWOSmvIkXklqxGCS1I6jnNRYOcpJUg8Gk6R2\nHOWkxvw+Jklqwo5Jaqr82hNJ6sNgktSOo5zUmMcxSVITdkxSW57EK0ltGEyS2nGUkxpz8VuSmjCY\nJLXjKCc1NcvXlbNjktSOHZPUVY1O5J1FdkyS2jGYJLXjKCc15jdYSlITBpOkdhzlpKYKT0mRpDbs\nmKS2/D4mSWrDYJLUjqOc1JinpEhSE3ZMUmMeLiBJTRhMktpxlJOaqnKUk6Q2DCZJ7TjKSY3N6ikp\nUxFMxz72Ts688INDl7Espz1qzdAlqKGrhy5gSkxFMEmzyiO/JakJg0lSO45yUmMexyRJTdgxSU0V\nsWOSpC4MJkntOMpJjc3oYUx2TJL6MZgkteMoJ3Xl9zFJUh92TFJnM7r6bcckqR2DSdIDkmRTkmuT\nXJfk9D1s81NJrk5yVZK3LbZPRzmpse6L30nWAK8HngZsAy5NckFVXT1nm+OAVwFPrKovJ3nYYvu1\nY5L0QJwAXFdV11fVduAc4OTdtnkR8Pqq+jJAVd2y2E4HCaYkhyY5L8knk1yT5IeHqEPSotYn2Trn\ntnm3148CbpzzeNv4ubkeDTw6yT8nuSTJpsXedKhR7o+B91TVc5LsBxw4UB1Saw2+WvfWqtr4APex\nFjgOeDJwNPChJI+tqq/s6QdWvGNK8mDgScAbAKpq+0IFSmrtJmDDnMdHj5+baxtwQVXdV1WfAT7F\nKKj2aIhR7ljgS8CbknwsyV8lOWiAOqTWitHi95C3JbgUOC7JsePp5xTggt22eSejbokk6xmNdtcv\ntNMhgmkt8P3An1XV44GvAf/uI8Ykm++fa2+7fddK1yhpCapqB/Bi4CLgGuDcqroqyWuTPHO82UXA\nbUmuBi4Gfqmqbltov0OsMW0DtlXVR8aPz2OeYKqqLcAWgO993L7DT9qS5lVVFwIX7vbcq+fcL+AX\nx7clWfFgqqovJLkxyXdW1bXAiXgdQOnfK6D5cUyTMtSnci8BzhrPpNcDpw1Uh6SGBgmmqroCeKAf\nQUqrXoPDBQbhkd+S2jGYJLXjSbxSZ45yktSDwSSpHUc5qS0vES5JbdgxSZ25+C1JPRhMktpxlJO6\n8kq8ktSHHZPUmYvfktSDwSSpHUc5qTUXvyWphanomD755e/gP779ZUOXsSzH/uDXhy5hWdZ+dbrq\nBbjvIVN4ndR/HLqA6TAVwSTNLD+Vk6Qe7JikzuyYJKkHg0lSO45yUlczfCVeOyZJ7RhMktpxlJMa\n8xLhktSEHZPUmR2TJPVgMElqx1FO6szjmCSpBzsmqbG4+C1JPRhMktpxlJO6KjyOSZK6MJgkteMo\nJ7UVj2OSpC7smKTOXPyWpB4MJkntOMpJnTnKSVIPdkxSZ3ZMktSDwSSpHUc5qSuvxCtJfSwpmJIc\nmOTXk/zl+PFxSX5ysqVJmlVL7ZjeBNwL/PD48U3Ab06kIknfkBr2NpSlBtOjqup1wH0AVXU3MJvD\nr6SJW+ri9/YkBzA+qiLJoxh1UJImaUaPY1pqML0GeA+wIclZwBOB0yZVlKTZtqRgqqr3JrkM+CFG\nI9xLq+rWiVYmaWYt9VO591fVbVX191X1rqq6Ncn7J12cpNm0YMeUZH/gQGB9ksP45oL3IcBRE65N\n0oxabJT778DLgCOBy/hmMH0V+D8TrEvSDFswmKrqj4E/TvKSqvqTFapJ0tisXiJ8qYvff5LkMcDx\nwP5znn/LpAqba/+b7ua4X71iJd5qr7ntpx8/dAnLsv8d+w1dwrLd+tgpPNXzH4cuYDos6b9skjOA\nJzMKpguBpwP/BKxIMEkzy5N4F/Qc4ETgC1V1GvC9wIMnVpWkmbbUYLqnqnYBO5IcAtwCbJhcWZJm\n2VKH9K1JDgX+ktGnc3cBH55YVZLG38c0dBHDWOri98+P7/55kvcAh1TVlZMrS9IsW/KR3/ffr6ob\nqupKj/yWVkANfBuIR35Lame5R37f70488lvShCwWTP8CnAs8Z3yQ5c8CzwZuAN424dqkmTerR34v\ntsb0F8C941B6EvA7wJuBO4Atky5O0mxarGNaU1W3j+//NLClqs4Hzk8yXeeISJoai3VMa5LcH14n\n8q1n+kzhiUrSlPFTuXmdDXwwya3APcD/A0jyHxiNc5K01y32tSe/NT5e6QjgvVV1f4buA7xk0sVJ\nM29GF78XHceq6pJ5nvvUZMqRJC8RLqkhF7Clpoa+Gu6Q7JgktTNIx5TkfwL/jdHS3r8Cp1XV14eo\nRWrNb7BcGUmOAn4B2FhVjwHWAKesdB2S+hpqlFsLHDA+ePNA4PMD1SGpoRUPpqq6CfjfwOeAm4E7\nquq9K12HNBVm9MjvIUa5w4CTgWMZfZ3KQUmeN892m5NsTbJ1O/eudJmSBjTEKPdU4DNV9aWqug94\nB/Aju29UVVuqamNVbdyPB614kZKGM8Sncp8DfijJgYzOvzsR2DpAHVJ7Hse0QqrqI8B5wOWMDhXY\nB7/bSdIcgxzHVFVnAGcM8d7SVLFjkqQeDCZJ7XgSr9SVJ/FKUh8Gk6R2HOWkzhzlJKkHOyapMzsm\nSerBYJLUjqOc1JjHMUlSEwaTpHYMJkntGEyS2nHxW+rMxW9J6mE6OqYE9t136CqW5e4jpusKqjsf\ntGboEpbtYZffN3QJy3bt0AVMiekIJmkW+X1MktSHHZPUmR2TJPVgMElqx1FO6sxRTpJ6sGOSmgoe\nLiBJbRhMktpxlJM6c5STpB4MJkntOMpJXXkSryT1YcckdWbHJEnLl2RTkmuTXJfk9AW2e3aSSrJx\nsX0aTJK+bUnWAK8Hng4cD5ya5Ph5tjsYeCnwkaXs12CSOquBb4s7Abiuqq6vqu3AOcDJ82z3G8Dv\nAV9fyk4NJkkPxFHAjXMebxs/9w1Jvh/YUFV/v9SduvgtaSHrk2yd83hLVW1Z6g8n2Qf4A+AFy3lT\ng0lqrMFxTLdW1UKL1TcBG+Y8Pnr83P0OBh4DfCAJwHcAFyR5ZlXNDbxv4Sgn6YG4FDguybFJ9gNO\nAS64/8WquqOq1lfVMVV1DHAJsGAogR2T1NvwHdOCqmpHkhcDFwFrgDdW1VVJXgtsraoLFt7D/Awm\nSQ9IVV0IXLjbc6/ew7ZPXso+HeUktWPHJHW19GOJVh07Jknt2DFJjTU4XGAQdkyS2jGYJLXjKCd1\n5ignST0YTJLacZSTGvNTOUlqwo5J6syOSZJ6MJgkteMoJ3XlSbyS1Icdk9RUxrdZZMckqR2DSVI7\nUzHKHfuYr/LWC98zdBnL8rxj7xm6BK0GLn5LUg8Gk6R2pmKUk2aVJ/FKUhN2TFJndkyS1IPBJKkd\nRzmpM0c5SerBYJLUjqOc1FV5HJMktWHHJHVmxyRJPRhMktpxlJMac/FbkpqwY5I6s2Pau5K8Mckt\nST4x57nDk7wvyb+N/3nYpN5f0vSa5Ch3JrBpt+dOB95fVccB7x8/lqRvMbFgqqoPAbfv9vTJwJvH\n998MPGtS7y+tBqlhb0NZ6cXvh1fVzeP7XwAevqcNk2xOsjXJ1ttu27Uy1UlqYbBP5apqwSuzV9WW\nqtpYVRsf8hA/PJRmyUp/KvfFJEdU1c1JjgBuWeH3l6bHgv/rXt1WuhW5APjZ8f2fBf5uhd9f0hSY\nWMeU5GzgycD6JNuAM4DfBc5N8kLgs8BPTer9pVVhRjumiQVTVZ26h5dOnNR7SlodXFWW1I6npEhN\nBU/ilaQ27JikzuyYJKkHg0lSO45yUmOp2Zzl7JgktWMwSWrHUU7qypN4JakPOyapMY/8lqQmDCZJ\n7TjKSZ05yklSD1PRMV1/5Tqe+8gnDV3Gstxx6sahS1iW+9Zl6BKWbcf+01czf3TO0BVMhakIJmlW\n+amcJDVhxyR1ZsckST0YTJLacZSTuioXvyWpDTsmqTM7JknqwWCS1I6jnNSUV+KVpEYMJkntOMpJ\nnXldOUnqwY5JaszFb0lqwmCS1I6jnNSVV+KVpD7smKTGsmvoCoZhxySpHYNJUjuOclJnLn5LUg8G\nk6R2HOWkxjwlRZKasGOSuir82hNJ6sJgktSOo5zUmIvfktSEwSSpHUc5qTNHOUnqwY5Jasor8UpS\nIwaTpHYc5aSuqjwlRZK6sGOSGnPxW5KaMJgktTM9o9yunUNXsCw7DsjQJax6a78+A3PODPyK87Fj\nktSOwSSpnekZ5aQZ5KdyktSEHZPUVQG7ZrNlsmOS1I7BJKkdRzmps9mc5OyYJPVjMElqx1FOaszj\nmCSpCTsmqTO/wVKSejCYJLXjKCc15uK3JDVhMEldVYPbEiTZlOTaJNclOX2e138xydVJrkzy/iSP\nXGyfEwumJG9MckuST8x57veTfHJc4N8mOXRS7y9p8pKsAV4PPB04Hjg1yfG7bfYxYGNVPQ44D3jd\nYvudZMd0JrBpt+feBzxmXOCngFdN8P0lTd4JwHVVdX1VbQfOAU6eu0FVXVxVd48fXgIcvdhOJxZM\nVfUh4PbdnntvVe0YP1xSgdKsCpCqQW9LcBRw45zH28bP7ckLgXcvttMhP5X7OeBvBnx/SYtbn2Tr\nnMdbqmrLt7OjJM8DNgI/tti2gwRTkl8FdgBnLbDNZmAzwP4cuEKVSdrNrVW1cYHXbwI2zHl89Pi5\nb5HkqcCvAj9WVfcu9qYrHkxJXgD8JHBi1Z57xXEqbwE4JIfP6NEcmnm7hi5gUZcCxyU5llEgnQL8\nzNwNkjwe+AtgU1XdspSdrmgwJdkEvJJRat692PaSequqHUleDFwErAHeWFVXJXktsLWqLgB+H1gH\nvD0JwOeq6pkL7XdiwZTkbODJjGbUbcAZjD6FexDwvnGBl1TV/5hUDdK0W+IC9KCq6kLgwt2ee/Wc\n+09d7j4nFkxVdeo8T79hUu8nafXwyG9J7XgSr9TVMk4LWW3smCS1Y8cktVV+g6UkdWEwSWrHUU5q\nzG+wlKQmDCZJ7TjKSZ35qZwk9WDHJHVVkP5fezIRdkyS2jGYJLXjKCd15uK3JPVgMElqZ3pGudFX\n8U6NtV+frhb8jkdN3/+jDr9mBj6ymq4/RnvN9P1plLTqTU/HJM2gabgYwSTYMUlqx2CS1I6jnNSZ\no5wk9WDHJHVVwAwcETEfOyZJ7RhMktpxlJOaCuVxTJLUhcEkqR1HOakzRzlJ6sGOSerMjkmSejCY\nJLXjKCd15SkpktSHHZPUmEd+S1ITBpOkdhzlpM4c5SSpB4NJUjuOclJb5SgnSV3YMUldFXZMktSF\nwSSpHUc5qTNP4pWkHgwmSe04ykmN+e0CktSEHZPUmR2TJPVgMElqx1FO6qqAXY5yktSCHZPUll97\nIkltGEyS2nGUkzqb0VFuOoIpIfvtN3QVy3LP+ulqRg/75PSdxv71QzN0CZqQ6frbI2kmTEfHJM2q\nGR3l7JgktWPHJHXlkd+S1IfBJKkdRzmprYKavsM49gY7Jknt2DFJnXm4gCT1YDBJasdRTurK45gk\nqQ+DSVI7jnJSZ34qJ0k92DFJndkx7V1J3pjkliSfmOe1lyepJOsn9f6SptckR7kzgU27P5lkA/Cf\ngc9N8L0lTbGJBVNVfQi4fZ6X/hB4JaOjNCTt0fi6ckPeBrKii99JTgZuqqqPr+T7SpouK7b4neRA\n4FcYjXFL2X4zsBlgfw6cYGWSulnJT+UeBRwLfDwJwNHA5UlOqKov7L5xVW0BtgAcss9DHPs0ewrY\nNZvfx7RiwVRV/wo87P7HSW4ANlbVrStVg6TpMMnDBc4GPgx8Z5JtSV44qfeSVq0ZXfyeWMdUVacu\n8voxk3pvSdPNU1IkteMpKVLcUVZRAAADSklEQVRnnpIiST3YMUltld9gKUldGEyS2nGUk7oqKK/E\nK0k9GEyS2nGUkzrzUzlJ6sGOSerMI78lqQeDSVI7jnJSV1Uz+w2WdkyS2rFjkjqb0cXvqQimnYcd\nyFdOevzQZSzL4ddsH7qEZdlx0JqhS1i2g744m2POLHCUk9TOVHRM0qwqF78lqQeDSVI7jnJSW8Ne\n221IdkyS2rFjkroq/NoTSerCYJLUjqOc1JkXI5CkHgwmSe04yklNFVB+KidJPdgxSV1VufgtSV0Y\nTJLacZSTGnPxW5KasGOSOnPxW5J6sGOSmrqTL1/0D3Xe+oHLuHWINzWYpKaqatPQNQzFUU5SOwaT\npHYMJkntGEyS2jGYJLVjMElqx2CS1I7BJKkdg0lSOwaTpHYMJkntGEyS2jGYJLVjMElqx2CS1I7B\nJKkdg0lSOwaTpHYMJkntGEyS2jGYJLWTqv6XIE7yJeCzE9j1ega6PM0DYM2TN8l6H1lVD53QvleN\nqQimSUmytao2Dl3Hcljz5E1bvauRo5ykdgwmSe3MejBtGbqAb8OiNSd5VpJK8l2LbPeCJEfOefxX\nSY7fG0XuZtr+PU9bvavOTK8xrVZJ/gY4EvjHqjpjge0+ALyiqrauVG3SUsx6x7TqJFkH/CjwQuCU\nOc//cpJ/TfLxJL+b5DnARuCsJFckOSDJB5JsHG9/6nj7TyT5vTn7uSvJb433c0mSh6/wr6gZYDCt\nPicD76mqTwG3JfmBJE8fP/+DVfW9wOuq6jxgK/Dcqvq+qrrn/h2Mx7vfA54CfB/whCTPGr98EHDJ\neD8fAl60Yr+ZZobBtPqcCpwzvn/O+PFTgTdV1d0AVXX7Ivt4AvCBqvpSVe0AzgKeNH5tO/Cu8f3L\ngGP2XunSyNqhC9Dek+RwRl3OY5MUsAYo4O178W3uq28uTO7EP0OaADum1eU5wFur6pFVdUxVbQA+\nA9wBnJbkQPhGgAHcCRw8z34+CvxYkvVJ1jDquj44+fKlEYNpdTkV+NvdnjsfOAK4ANia5ArgFePX\nzgT+/P7F7/t/oKpuBk4HLgY+DlxWVX834dqlb/BwAUnt2DFJasdgktSOwSSpHYNJUjsGk6R2DCZJ\n7RhMktoxmCS18/8BXHlZ7bPF6ioAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1180a8eb8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "policy = agentA2C.get_policy()\n",
    "plt.matshow(policy)\n",
    "plt.title('Policy colormap')\n",
    "plt.xlabel('Action')\n",
    "plt.ylabel('State')\n",
    "plt.colorbar()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.colorbar.Colorbar at 0x118288be0>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAASYAAAMaCAYAAAAr4+slAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x114617668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "policy = agentA3C.get_policy()\n",
    "plt.matshow(policy)\n",
    "plt.title('Policy colormap')\n",
    "plt.xlabel('Action')\n",
    "plt.ylabel('State')\n",
    "plt.colorbar()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can find more information in source code (commented)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}