main.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fist we import the required modules."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "3YAaLe0QEsP6"
   },
   "outputs": [],
   "source": [
    "import random\n",
    "import numpy as np\n",
    "\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "from IPython import display\n",
    "\n",
    "from unityagents import UnityEnvironment\n",
    "from UnityDQNAgent import UnityDQNAgent"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we make a unity envionment and use it to create an agent."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:unityagents:\n",
      "'Academy' started successfully!\n",
      "Unity Academy name: Academy\n",
      "        Number of Brains: 1\n",
      "        Number of External Brains : 1\n",
      "        Lesson number : 0\n",
      "        Reset Parameters :\n",
      "\t\t\n",
      "Unity brain name: BananaBrain\n",
      "        Number of Visual Observations (per agent): 0\n",
      "        Vector Observation space type: continuous\n",
      "        Vector Observation space size (per agent): 37\n",
      "        Number of stacked Vector Observation: 1\n",
      "        Vector Action space type: discrete\n",
      "        Vector Action space size (per agent): 4\n",
      "        Vector Action descriptions: , , , \n"
     ]
    }
   ],
   "source": [
    "env = UnityEnvironment(file_name=\"Banana_Linux/Banana.x86_64\")\n",
    "\n",
    "brain_name = env.brain_names[0]\n",
    "brain = env.brains[brain_name]\n",
    "\n",
    "env_info = env.reset(train_mode=True)[brain_name]\n",
    "action_size = brain.vector_action_space_size\n",
    "state_size = len(env_info.vector_observations[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "hyperparams = { 'batch_size':500,   # minibatch size\n",
    "                'gamma':0.99,       # discount factor\n",
    "                'lr':1e-4,          # learning rate\n",
    "                'update_every':2,   # how often we update the target network\n",
    "                'tau':1e-3,         # weight for the soft update of the target network\n",
    "                'eps_start':0.9,\n",
    "                'eps_end':0.01, \n",
    "                'eps_decay':0.98,\n",
    "                'seed':2147483647 } # random seed\n",
    "\n",
    "agent = UnityDQNAgent(env, brain_name=brain_name, \\\n",
    "                        state_size=state_size, action_size=action_size, **hyperparams)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following function will be run at every episode. We use it to plot our progress and check if we have solved the game."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def is_finished(scores, num_episodes):\n",
    "    plot_every = 10\n",
    "    avg_every = 100\n",
    "    winning_threshold = 13\n",
    "    \n",
    "    i = len(scores)\n",
    "    finished = (i >= avg_every and np.mean(scores[i-avg_every:i]) >= winning_threshold)\n",
    "\n",
    "    if not finished and (i+1) % plot_every != 0: return False\n",
    "    \n",
    "    mean_scores = [np.mean(scores[max(0, j-avg_every+1):j+1]) for j in range(i)]\n",
    "    plt.plot(scores, color='red')\n",
    "    plt.plot(mean_scores, color='blue')\n",
    "    display.display(plt.gcf())\n",
    "    print(\"\\r Episode {}/{}\\tAverage Score: {:.2f}\".format(i+1, num_episodes, mean_scores[-1]), end=\"\")\n",
    "    display.clear_output(wait=True)\n",
    "    if finished:\n",
    "        print(\"\\r Finished in episode {}/{}\\tAverage Score: {:.2f}\".format(i+1, num_episodes, mean_scores[-1]))\n",
    "    return finished"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now train the agent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 269
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 9,
     "status": "ok",
     "timestamp": 1569786071022,
     "user": {
      "displayName": "Mohammad Pedramfar",
      "photoUrl": "",
      "userId": "06489714970426098467"
     },
     "user_tz": -60
    },
    "id": "EsrLlmXFEsQS",
    "outputId": "f1b5e70d-01af-4110-e311-23f16c260e89",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      " Finished in episode 256/1000\tAverage Score: 13.05\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "scores = agent.train(500, is_finished=is_finished)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To observe the trained agent in action, we use the `agent.run()`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[16.0]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "agent.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "env.close()"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "LunarLander_DQL.ipynb",
   "provenance": []
  },
  "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}