Report.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Navigation project report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: unityagents in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (0.4.0)\n",
      "Requirement already satisfied: matplotlib in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (3.0.2)\n",
      "Requirement already satisfied: grpcio==1.11.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (1.11.0)\n",
      "Requirement already satisfied: numpy>=1.11.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (1.15.4)\n",
      "Requirement already satisfied: tensorflow==1.7.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (1.7.1)\n",
      "Requirement already satisfied: pyyaml in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (3.13)\n",
      "Requirement already satisfied: pytest>=3.2.2 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (4.0.2)\n",
      "Requirement already satisfied: docopt in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (0.6.2)\n",
      "Requirement already satisfied: Pillow>=4.2.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (5.3.0)\n",
      "Requirement already satisfied: protobuf==3.5.2 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (3.5.2)\n",
      "Requirement already satisfied: jupyter in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from unityagents) (1.0.0)\n",
      "Requirement already satisfied: python-dateutil>=2.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from matplotlib->unityagents) (2.7.5)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from matplotlib->unityagents) (1.0.1)\n",
      "Requirement already satisfied: cycler>=0.10 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from matplotlib->unityagents) (0.10.0)\n",
      "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from matplotlib->unityagents) (2.3.0)\n",
      "Requirement already satisfied: six>=1.5.2 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from grpcio==1.11.0->unityagents) (1.12.0)\n",
      "Requirement already satisfied: absl-py>=0.1.6 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorflow==1.7.1->unityagents) (0.6.1)\n",
      "Requirement already satisfied: termcolor>=1.1.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorflow==1.7.1->unityagents) (1.1.0)\n",
      "Requirement already satisfied: astor>=0.6.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorflow==1.7.1->unityagents) (0.7.1)\n",
      "Requirement already satisfied: tensorboard<1.8.0,>=1.7.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorflow==1.7.1->unityagents) (1.7.0)\n",
      "Requirement already satisfied: gast>=0.2.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorflow==1.7.1->unityagents) (0.2.0)\n",
      "Requirement already satisfied: wheel>=0.26 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorflow==1.7.1->unityagents) (0.32.3)\n",
      "Requirement already satisfied: attrs>=17.4.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from pytest>=3.2.2->unityagents) (18.2.0)\n",
      "Requirement already satisfied: py>=1.5.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from pytest>=3.2.2->unityagents) (1.7.0)\n",
      "Requirement already satisfied: atomicwrites>=1.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from pytest>=3.2.2->unityagents) (1.2.1)\n",
      "Requirement already satisfied: more-itertools>=4.0.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from pytest>=3.2.2->unityagents) (4.3.0)\n",
      "Requirement already satisfied: setuptools in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages/setuptools-28.8.0-py3.6.egg (from pytest>=3.2.2->unityagents) (28.8.0)\n",
      "Requirement already satisfied: pluggy>=0.7 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from pytest>=3.2.2->unityagents) (0.8.0)\n",
      "Requirement already satisfied: ipywidgets in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from jupyter->unityagents) (7.4.2)\n",
      "Requirement already satisfied: qtconsole in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from jupyter->unityagents) (4.4.3)\n",
      "Requirement already satisfied: nbconvert in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from jupyter->unityagents) (5.4.0)\n",
      "Requirement already satisfied: ipykernel in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from jupyter->unityagents) (5.1.0)\n",
      "Requirement already satisfied: notebook in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from jupyter->unityagents) (5.7.3)\n",
      "Requirement already satisfied: jupyter-console in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from jupyter->unityagents) (6.0.0)\n",
      "Requirement already satisfied: markdown>=2.6.8 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorboard<1.8.0,>=1.7.0->tensorflow==1.7.1->unityagents) (3.0.1)\n",
      "Requirement already satisfied: bleach==1.5.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorboard<1.8.0,>=1.7.0->tensorflow==1.7.1->unityagents) (1.5.0)\n",
      "Requirement already satisfied: html5lib==0.9999999 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorboard<1.8.0,>=1.7.0->tensorflow==1.7.1->unityagents) (0.9999999)\n",
      "Requirement already satisfied: werkzeug>=0.11.10 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from tensorboard<1.8.0,>=1.7.0->tensorflow==1.7.1->unityagents) (0.14.1)\n",
      "Requirement already satisfied: nbformat>=4.2.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from ipywidgets->jupyter->unityagents) (4.4.0)\n",
      "Requirement already satisfied: ipython>=4.0.0; python_version >= \"3.3\" in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from ipywidgets->jupyter->unityagents) (5.0.0)\n",
      "Requirement already satisfied: widgetsnbextension~=3.4.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from ipywidgets->jupyter->unityagents) (3.4.2)\n",
      "Requirement already satisfied: traitlets>=4.3.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from ipywidgets->jupyter->unityagents) (4.3.2)\n",
      "Requirement already satisfied: pygments in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from qtconsole->jupyter->unityagents) (2.3.0)\n",
      "Requirement already satisfied: jupyter-client>=4.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from qtconsole->jupyter->unityagents) (5.2.4)\n",
      "Requirement already satisfied: jupyter-core in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from qtconsole->jupyter->unityagents) (4.4.0)\n",
      "Requirement already satisfied: ipython-genutils in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from qtconsole->jupyter->unityagents) (0.2.0)\n",
      "Requirement already satisfied: mistune>=0.8.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from nbconvert->jupyter->unityagents) (0.8.4)\n",
      "Requirement already satisfied: pandocfilters>=1.4.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from nbconvert->jupyter->unityagents) (1.4.2)\n",
      "Requirement already satisfied: defusedxml in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from nbconvert->jupyter->unityagents) (0.5.0)\n",
      "Requirement already satisfied: jinja2 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from nbconvert->jupyter->unityagents) (2.10)\n",
      "Requirement already satisfied: entrypoints>=0.2.2 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from nbconvert->jupyter->unityagents) (0.2.3)\n",
      "Requirement already satisfied: testpath in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from nbconvert->jupyter->unityagents) (0.4.2)\n",
      "Requirement already satisfied: tornado>=4.2 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from ipykernel->jupyter->unityagents) (5.1.1)\n",
      "Requirement already satisfied: terminado>=0.8.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from notebook->jupyter->unityagents) (0.8.1)\n",
      "Requirement already satisfied: pyzmq>=17 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from notebook->jupyter->unityagents) (17.1.2)\n",
      "Requirement already satisfied: prometheus-client in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from notebook->jupyter->unityagents) (0.5.0)\n",
      "Requirement already satisfied: Send2Trash in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from notebook->jupyter->unityagents) (1.5.0)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting prompt-toolkit<2.1.0,>=2.0.0 (from jupyter-console->jupyter->unityagents)\n",
      "  Using cached https://files.pythonhosted.org/packages/d1/e6/adb3be5576f5d27c6faa33f1e9fea8fe5dbd9351db12148de948507e352c/prompt_toolkit-2.0.7-py3-none-any.whl\n",
      "Requirement already satisfied: jsonschema!=2.5.0,>=2.4 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from nbformat>=4.2.0->ipywidgets->jupyter->unityagents) (2.6.0)\n",
      "Requirement already satisfied: pexpect; sys_platform != \"win32\" in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from ipython>=4.0.0; python_version >= \"3.3\"->ipywidgets->jupyter->unityagents) (4.6.0)\n",
      "Requirement already satisfied: simplegeneric>0.8 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from ipython>=4.0.0; python_version >= \"3.3\"->ipywidgets->jupyter->unityagents) (0.8.1)\n",
      "Requirement already satisfied: pickleshare in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from ipython>=4.0.0; python_version >= \"3.3\"->ipywidgets->jupyter->unityagents) (0.7.5)\n",
      "Requirement already satisfied: decorator in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from ipython>=4.0.0; python_version >= \"3.3\"->ipywidgets->jupyter->unityagents) (4.3.0)\n",
      "Requirement already satisfied: MarkupSafe>=0.23 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from jinja2->nbconvert->jupyter->unityagents) (1.1.0)\n",
      "Requirement already satisfied: ptyprocess; os_name != \"nt\" in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from terminado>=0.8.1->notebook->jupyter->unityagents) (0.6.0)\n",
      "Requirement already satisfied: wcwidth in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from prompt-toolkit<2.1.0,>=2.0.0->jupyter-console->jupyter->unityagents) (0.1.7)\n",
      "\u001b[31mipython 5.0.0 has requirement prompt-toolkit<2.0.0,>=1.0.3, but you'll have prompt-toolkit 2.0.7 which is incompatible.\u001b[0m\n",
      "Installing collected packages: prompt-toolkit\n",
      "  Found existing installation: prompt-toolkit 1.0.15\n",
      "    Uninstalling prompt-toolkit-1.0.15:\n",
      "      Successfully uninstalled prompt-toolkit-1.0.15\n",
      "Successfully installed prompt-toolkit-2.0.7\n",
      "Requirement already satisfied: torch in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (1.0.0)\n",
      "Requirement already satisfied: numpy in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (1.15.4)\n",
      "Requirement already satisfied: matplotlib in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (3.0.2)\n",
      "Requirement already satisfied: cycler>=0.10 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from matplotlib) (0.10.0)\n",
      "Requirement already satisfied: numpy>=1.10.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from matplotlib) (1.15.4)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from matplotlib) (1.0.1)\n",
      "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from matplotlib) (2.3.0)\n",
      "Requirement already satisfied: python-dateutil>=2.1 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from matplotlib) (2.7.5)\n",
      "Requirement already satisfied: six in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from cycler>=0.10->matplotlib) (1.12.0)\n",
      "Requirement already satisfied: setuptools in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages/setuptools-28.8.0-py3.6.egg (from kiwisolver>=1.0.1->matplotlib) (28.8.0)\n",
      "Collecting prompt-toolkit==1.0.15\n",
      "  Using cached https://files.pythonhosted.org/packages/04/d1/c6616dd03701e7e2073f06d5c3b41b012256e42b72561f16a7bd86dd7b43/prompt_toolkit-1.0.15-py3-none-any.whl\n",
      "Requirement already satisfied: wcwidth in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from prompt-toolkit==1.0.15) (0.1.7)\n",
      "Requirement already satisfied: six>=1.9.0 in /home/ammi/Development/deep-monkey/lib/python3.6/site-packages (from prompt-toolkit==1.0.15) (1.12.0)\n",
      "\u001b[31mjupyter-console 6.0.0 has requirement prompt-toolkit<2.1.0,>=2.0.0, but you'll have prompt-toolkit 1.0.15 which is incompatible.\u001b[0m\n",
      "Installing collected packages: prompt-toolkit\n",
      "  Found existing installation: prompt-toolkit 2.0.7\n",
      "    Uninstalling prompt-toolkit-2.0.7:\n",
      "      Successfully uninstalled prompt-toolkit-2.0.7\n",
      "Successfully installed prompt-toolkit-1.0.15\n"
     ]
    }
   ],
   "source": [
    "!python -m pip install unityagents\n",
    "!python -m pip install torch\n",
    "!python -m pip install numpy\n",
    "!python -m pip install matplotlib\n",
    "!python -m pip install 'prompt-toolkit==1.0.15'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from unityagents import UnityEnvironment\n",
    "import numpy as np\n",
    "\n",
    "# import model\n",
    "from agent import Agent, DQN, DoubleDQN\n",
    "import torch\n",
    "from deep_monkey import *\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "import glob\n",
    "import os"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Setup environment\n",
    "UnityEnvironment contains several brains which control the actions of the agents. In this case we are interested only in the first brain as there is only one agent "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "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=\"../bin/Banana_Linux/Banana.x86_64\")\n",
    "\n",
    "# Extract brain object and its name,\n",
    "# useful for obtaining feedback from the environment\n",
    "brain_name = env.brain_names[0]\n",
    "brain = env.brains[brain_name]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training a vanilla DQN agent\n",
    "The model we are going to train is a vanilla DQN model. The following hyperparameters were chosen for the algorithm:\n",
    "```\n",
    "- gamma           = 0.99\n",
    "- tau             = 1e-3\n",
    "- learning_rate   = 1e-3\n",
    "- learning_period = 4\n",
    "- epsilon_max     = 1\n",
    "- epsilon_min     = 0.01\n",
    "```\n",
    "The two neural networks used to estimate the q function have the following hyperparameters:\n",
    "- 2 fully connected layers\n",
    "- 64 units per layer\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Episode 100\tAverage Score: 1.31\n",
      "Episode 200\tAverage Score: 5.22\n",
      "Episode 300\tAverage Score: 9.27\n",
      "Episode 400\tAverage Score: 10.00\n",
      "Episode 500\tAverage Score: 11.98\n",
      "Episode 575\tAverage Score: 13.01\n",
      "Environment solved in 475 episodes!\tAverage Score: 13.01\n"
     ]
    }
   ],
   "source": [
    "# reset the environment\n",
    "state = env.reset(train_mode=True)\n",
    "observation = state[brain_name].vector_observations[0]\n",
    "\n",
    "# create the agent\n",
    "agent = DQN(state_size = len(observation), \n",
    "            action_size = brain.vector_action_space_size,\n",
    "            seed = 0)\n",
    "\n",
    "# train the agent\n",
    "agent, episodes_n, avg_score, scores = train_agent(agent,env,stop_avg_score=13)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Save agent checkpoint, plot the evolution of the average score during training and save the graph to file."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "tag = generate_filename_tag(agent, episodes_n, avg_score)\n",
    "checkpoint_filename = save_checkpoint(agent, tag)\n",
    "fig = plot_score(scores)\n",
    "fig.savefig(\"plots/training_plot_\" + tag + \".pdf\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Trained vanilla DQN demo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Score: 4.0\n"
     ]
    }
   ],
   "source": [
    "# reset the environment\n",
    "state = env.reset(train_mode=False)\n",
    "observation = state[brain_name].vector_observations[0]\n",
    "\n",
    "# load the weights from file\n",
    "agent = Agent(state_size = len(observation), \n",
    "              action_size = brain.vector_action_space_size,\n",
    "              seed = 0,\n",
    "              checkpoint_filename=checkpoint_filename)\n",
    "\n",
    "_ = demo_agent(agent, env)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training a DoubleDQN agent\n",
    "In this section we will repeat the same sequence but we will train a DoubleDQN agent, save its weights and generate a score plot with training values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Episode 100\tAverage Score: 0.91\n",
      "Episode 200\tAverage Score: 4.65\n",
      "Episode 300\tAverage Score: 8.59\n",
      "Episode 400\tAverage Score: 10.44\n",
      "Episode 489\tAverage Score: 13.02\n",
      "Environment solved in 389 episodes!\tAverage Score: 13.02\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": [
    "# reset the environment\n",
    "state = env.reset(train_mode=True)\n",
    "observation = state[brain_name].vector_observations[0]\n",
    "\n",
    "# create the agent\n",
    "agent = DoubleDQN(state_size = len(observation), \n",
    "            action_size = brain.vector_action_space_size,\n",
    "            seed = 0)\n",
    "\n",
    "# train the agent\n",
    "agent, episodes_n, avg_score, scores = train_agent(agent,env,stop_avg_score=13)\n",
    "\n",
    "# save checkpoint, plot scores and save plot\n",
    "tag = generate_filename_tag(agent, episodes_n, avg_score)\n",
    "checkpoint_filename = save_checkpoint(agent, tag)\n",
    "fig = plot_score(scores)\n",
    "fig.savefig(\"plots/training_plot_\" + tag + \".pdf\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Trained DoubleDQN demo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Score: 17.0\n"
     ]
    }
   ],
   "source": [
    "# reset the environment\n",
    "state = env.reset(train_mode=False)\n",
    "observation = state[brain_name].vector_observations[0]\n",
    "\n",
    "# load the weights from file\n",
    "agent = Agent(state_size = len(observation), \n",
    "              action_size = brain.vector_action_space_size,\n",
    "              seed = 0,\n",
    "              checkpoint_filename=checkpoint_filename)\n",
    "\n",
    "_ = demo_agent(agent, env)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "env.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Future development\n",
    "Agent's learning performances (number of episodes to solve the task) should be improved implementing the following:\n",
    "- Prioritized experience replay\n",
    "- Dueling DQN\n",
    "\n",
    "In addition, the same task should be solved starting from raw pixel observations."
   ]
  }
 ],
 "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}