Mohinga_Project_RL_One.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "Azi_bLWn6mAR"
      },
      "source": [
        "# Using RL to Model Cognitive Tasks\n",
        "\n",
        "**By Neurmatch Academy**\n",
        "\n",
        "__Content creators:__ Morteza Ansarinia, Yamil Vidal\n",
        "\n",
        "__Production editor:__ Spiros Chavlis\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "t0JJoEeC6mAb"
      },
      "source": [
        "<p align='center'><img src='https://github.com/NeuromatchAcademy/widgets/blob/master/sponsors.png?raw=True'/></p>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "9HbuVnWO6mAc"
      },
      "source": [
        "---\n",
        "# Objective\n",
        "\n",
        "- This project aims to use behavioral data to train an agent and then use the agent to investigate data produced by human subjects. Having a computational agent that mimics humans in such tests, we will be able to compare its mechanics with human data.\n",
        "\n",
        "- In another conception, we could fit an agent that learns many cognitive tasks that require abstract-level constructs such as executive functions. This is a multi-task control problem.\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "-hzSA9mE6mAd"
      },
      "source": [
        "---\n",
        "# Setup"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "execution": {},
        "id": "zigNT_hI6mAd",
        "outputId": "7b0074d9-0371-487a-d67c-941624573d8a",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[?25l\r\u001b[K     |█                               | 10 kB 27.5 MB/s eta 0:00:01\r\u001b[K     |██                              | 20 kB 30.3 MB/s eta 0:00:01\r\u001b[K     |███▏                            | 30 kB 32.5 MB/s eta 0:00:01\r\u001b[K     |████▏                           | 40 kB 11.0 MB/s eta 0:00:01\r\u001b[K     |█████▏                          | 51 kB 5.6 MB/s eta 0:00:01\r\u001b[K     |██████▎                         | 61 kB 6.6 MB/s eta 0:00:01\r\u001b[K     |███████▎                        | 71 kB 6.1 MB/s eta 0:00:01\r\u001b[K     |████████▍                       | 81 kB 6.4 MB/s eta 0:00:01\r\u001b[K     |█████████▍                      | 92 kB 7.1 MB/s eta 0:00:01\r\u001b[K     |██████████▍                     | 102 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |███████████▌                    | 112 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |████████████▌                   | 122 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |█████████████▋                  | 133 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |██████████████▋                 | 143 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |███████████████▋                | 153 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |████████████████▊               | 163 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |█████████████████▊              | 174 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |██████████████████▉             | 184 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |███████████████████▉            | 194 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |████████████████████▉           | 204 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |██████████████████████          | 215 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |███████████████████████         | 225 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |████████████████████████        | 235 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |█████████████████████████       | 245 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |██████████████████████████      | 256 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |███████████████████████████▏    | 266 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |████████████████████████████▏   | 276 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |█████████████████████████████▏  | 286 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |██████████████████████████████▎ | 296 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |███████████████████████████████▎| 307 kB 5.7 MB/s eta 0:00:01\r\u001b[K     |████████████████████████████████| 314 kB 5.7 MB/s \n",
            "\u001b[33mWARNING: dm-acme 0.4.0 does not provide the extra 'tensorflow'\u001b[0m\n",
            "\u001b[K     |████████████████████████████████| 3.8 MB 58.3 MB/s \n",
            "\u001b[K     |████████████████████████████████| 72 kB 676 kB/s \n",
            "\u001b[K     |████████████████████████████████| 342 kB 71.5 MB/s \n",
            "\u001b[K     |████████████████████████████████| 202 kB 72.4 MB/s \n",
            "\u001b[K     |████████████████████████████████| 145 kB 73.8 MB/s \n",
            "\u001b[K     |████████████████████████████████| 117 kB 57.3 MB/s \n",
            "\u001b[K     |████████████████████████████████| 6.4 MB 48.0 MB/s \n",
            "\u001b[K     |████████████████████████████████| 4.0 MB 46.8 MB/s \n",
            "\u001b[K     |████████████████████████████████| 668.3 MB 18 kB/s \n",
            "\u001b[K     |████████████████████████████████| 5.7 MB 60.0 MB/s \n",
            "\u001b[K     |████████████████████████████████| 462 kB 69.2 MB/s \n",
            "\u001b[K     |████████████████████████████████| 9.1 MB 24.4 MB/s \n",
            "\u001b[K     |████████████████████████████████| 217 kB 72.5 MB/s \n",
            "\u001b[K     |████████████████████████████████| 596 kB 74.2 MB/s \n",
            "\u001b[K     |████████████████████████████████| 51 kB 4.7 MB/s \n",
            "\u001b[K     |████████████████████████████████| 272 kB 68.0 MB/s \n",
            "\u001b[?25h  Building wheel for dm-acme (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "\u001b[K     |████████████████████████████████| 18.8 MB 5.3 MB/s \n",
            "\u001b[K     |████████████████████████████████| 88 kB 9.0 MB/s \n",
            "\u001b[K     |████████████████████████████████| 207 kB 59.1 MB/s \n",
            "\u001b[K     |████████████████████████████████| 4.9 MB 30.8 MB/s \n",
            "\u001b[K     |████████████████████████████████| 99 kB 8.4 MB/s \n",
            "\u001b[?25h  Building wheel for bsuite (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "\u001b[K     |████████████████████████████████| 254 kB 5.1 MB/s \n",
            "\u001b[K     |████████████████████████████████| 104 kB 5.1 MB/s \n",
            "\u001b[?25h"
          ]
        }
      ],
      "source": [
        "# @title Install dependencies\n",
        "# !pip install --upgrade pip setuptools wheel --quiet\n",
        "!pip install dm-acme[jax,tensorflow] --quiet\n",
        "!pip install dm-acme[envs] --quiet\n",
        "!pip install dm-sonnet --quiet\n",
        "!pip install trfl --quiet"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "execution": {},
        "id": "IOdC2K-56mAk"
      },
      "outputs": [],
      "source": [
        "# Imports\n",
        "\n",
        "import time\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import sonnet as snt\n",
        "import seaborn as sns\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "import dm_env\n",
        "\n",
        "import acme\n",
        "from acme import specs\n",
        "from acme import wrappers\n",
        "from acme import EnvironmentLoop\n",
        "from acme.agents.tf import dqn\n",
        "from acme.utils import loggers\n",
        "from acme.agents.tf import r2d2\n",
        "from tqdm import tqdm"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "execution": {},
        "id": "VIvUNa5t6mAl"
      },
      "outputs": [],
      "source": [
        "# @title Figure settings\n",
        "from IPython.display import clear_output, display, HTML\n",
        "%matplotlib inline\n",
        "sns.set()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "-R0grYI-6mAm"
      },
      "source": [
        "---\n",
        "# Background\n",
        "\n",
        "- Cognitive scientists use standard lab tests to tap into specific processes in the brain and behavior. Some examples of those tests are Stroop, N-back, Digit Span, TMT (Trail making tests), and WCST (Wisconsin Card Sorting Tests).\n",
        "\n",
        "- Despite an extensive body of research that explains human performance using descriptive what-models, we still need a more sophisticated approach to gain a better understanding of the underlying processes (i.e., a how-model).\n",
        "\n",
        "- Interestingly, many of such tests can be thought of as a continuous stream of stimuli and corresponding actions, that is in consonant with the RL formulation. In fact, RL itself is in part motivated by how the brain enables goal-directed behaviors using reward systems, making it a good choice to explain human performance.\n",
        "\n",
        "- One behavioral test example would be the N-back task.\n",
        "\n",
        "  - In the N-back, participants view a sequence of stimuli, one by one, and are asked to categorize each stimulus as being either match or non-match. Stimuli are usually numbers, and feedback is given at both timestep and trajectory levels.\n",
        "\n",
        "  - The agent is rewarded when its response matches the stimulus that was shown N steps back in the episode. A simpler version of the N-back uses two-choice action schema, that is match vs non-match. Once the present stimulus matches the one presented N step back, then the agent is expected to respond to it as being a `match`.\n",
        "\n",
        "\n",
        "- Given a trained RL agent, we then find correlates of its fitted parameters with the brain mechanisms. The most straightforward composition could be the correlation of model parameters with the brain activities."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "Jr2eTAmn6mAo"
      },
      "source": [
        "## Datasets\n",
        "\n",
        "- HCP WM task ([NMA-CN HCP notebooks](https://github.com/NeuromatchAcademy/course-content/tree/master/projects/fMRI))\n",
        "\n",
        "Any dataset that used cognitive tests would work.\n",
        "Question: limit to behavioral data vs fMRI?\n",
        "Question: Which stimuli and actions to use?\n",
        "classic tests can be modeled using 1) bounded symbolic stimuli/actions (e.g., A, B, C), but more sophisticated one would require texts or images (e.g., face vs neutral images in social stroop dataset)\n",
        "The HCP dataset from NMA-CN contains behavioral and imaging data for 7 cognitive tests including various versions of N-back."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "hi23aJh46mAp"
      },
      "source": [
        "## N-back task\n",
        "\n",
        "In the N-back task, participants view a sequence of stimuli, one per time, and are asked to categorize each stimulus as being either match or non-match. Stimuli are usually numbers, and feedbacks are given at both timestep and trajectory levels.\n",
        "\n",
        "In a typical neuro setup, both accuracy and response time are measured, but here, for the sake of brevity, we focus only on accuracy of responses."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "0BCw1SqR6mAq"
      },
      "source": [
        "---\n",
        "# Cognitive Tests Environment\n",
        "\n",
        "First we develop an environment in that agents perform a cognitive test, here the N-back."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "H7aDZqad6mAr"
      },
      "source": [
        "## Human dataset\n",
        "\n",
        "We need a dataset of human perfoming a N-back test, with the following features:\n",
        "\n",
        "- `participant_id`: following the BIDS format, it contains a unique identifier for each participant.\n",
        "- `trial_index`: same as `time_step`.\n",
        "- `stimulus`: same as `observation`.\n",
        "- `response`: same as `action`, recorded response by the human subject.\n",
        "- `expected_response`: correct response.\n",
        "- `is_correct`: same as `reward`, whether the human subject responded correctly.\n",
        "- `response_time`: won't be used here.\n",
        "\n",
        "Here we generate a mock dataset with those features, but remember to **replace this with real human data.**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "mvgbdQuK6mAs"
      },
      "source": [
        "### Environment\n",
        "\n",
        "The following cell implments N-back envinronment, that we later use to train a RL agent on human data. It is capable of performing two kinds of simulation:\n",
        "- rewards the agent once the action was correct (i.e., a normative model of the environment).\n",
        "- receives human data (or mock data if you prefer), and returns what participants performed as the observation. This is more useful for preference-based RL."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "execution": {},
        "id": "ufg4-znW6mAs"
      },
      "outputs": [],
      "source": [
        "class NBack(dm_env.Environment):\n",
        "\n",
        "  ACTIONS = ['match', 'non-match']\n",
        "\n",
        "  def __init__(self,\n",
        "               N=2,\n",
        "               episode_steps=32,\n",
        "               stimuli_choices=list('ABCDEF'),\n",
        "               human_data=None,\n",
        "               seed=1,\n",
        "               prob = None\n",
        "               ):\n",
        "    \"\"\"\n",
        "    Args:\n",
        "      N: Number of steps to look back for the matched stimuli. Defaults to 2 (as in 2-back).\n",
        "      episode_steps\n",
        "      stimuli_choices\n",
        "      human_data\n",
        "      seed\n",
        "\n",
        "    \"\"\"\n",
        "    self.N = N\n",
        "    self.episode_steps = episode_steps\n",
        "    self.stimuli_choices = stimuli_choices\n",
        "    self.stimuli = np.empty(shape=episode_steps)  # will be filled in the `reset()`\n",
        "    self.n_s = len(self.stimuli_choices)\n",
        "    if prob is None:\n",
        "      self.prob = 0.5 * (self.n_s - 2)/ (self.n_s - 1)\n",
        "    else:\n",
        "      self.prob = prob\n",
        "    self._reset_next_step = True\n",
        "\n",
        "    # whether mimic humans or reward the agent once it responds optimally.\n",
        "    if human_data is None:\n",
        "      self._imitate_human = False\n",
        "      self.human_data = None\n",
        "      self.human_subject_data = None\n",
        "    else:\n",
        "      self._imitate_human = True\n",
        "      self.human_data = human_data\n",
        "      self.human_subject_data = None\n",
        "\n",
        "    self._action_history = []\n",
        "\n",
        "  def reset(self):\n",
        "    self._reset_next_step = False\n",
        "    self._current_step = 0\n",
        "    self._action_history.clear()\n",
        "\n",
        "    # generate a random sequence instead of relying on human data\n",
        "    if self.human_data is None:\n",
        "      for i in range(self.episode_steps):\n",
        "          if i<self.N:\n",
        "            self.stimuli[i]=np.random.choice(self.n_s)\n",
        "          else:\n",
        "            rand = np.random.random()\n",
        "            if rand < self.prob:\n",
        "              self.stimuli[i]=self.stimuli[i-self.N]\n",
        "            else:\n",
        "              self.stimuli[i]=np.random.choice(self.n_s)\n",
        "    else:\n",
        "      self.human_subject_data = self.human_data.query('participant_id == participant_id.sample().iloc[0]',\n",
        "                                                engine='python').sort_values('trial_index')\n",
        "      self.stimuli = self.human_subject_data['stimulus'].to_list()\n",
        "      self.stimuli = np.array([ord(s) - ord('A') + 1 for s in self.stimuli]).astype(np.float32)\n",
        "\n",
        "    return dm_env.restart(self._observation())\n",
        "\n",
        "\n",
        "  def _episode_return(self):\n",
        "    if self._imitate_human:\n",
        "      return np.mean(self.human_subject_data['response'] == self._action_history)\n",
        "    else:\n",
        "      return 0.0\n",
        "\n",
        "  def step(self, action: int):\n",
        "    if self._reset_next_step:\n",
        "      return self.reset()\n",
        "\n",
        "    agent_action = NBack.ACTIONS[action]\n",
        "\n",
        "    if self._imitate_human:\n",
        "      human_action = self.human_subject_data['response'].iloc[self._current_step]\n",
        "      step_reward = 0. if (agent_action == human_action) else -1.\n",
        "    else:\n",
        "      expected_action = 'match' if (self.stimuli[self._current_step] == self.stimuli[self._current_step - self.N]) else 'non-match'\n",
        "      step_reward = 0. if (agent_action == expected_action) else -1.\n",
        "\n",
        "    self._action_history.append(agent_action)\n",
        "\n",
        "    self._current_step += 1\n",
        "\n",
        "    if self._current_step == self.stimuli.shape[0]:\n",
        "      self._reset_next_step = True\n",
        "      return dm_env.termination(reward=self._episode_return(),\n",
        "                                observation=self._observation())\n",
        "    else:\n",
        "      return dm_env.transition(reward=step_reward,\n",
        "                               observation=self._observation())\n",
        "\n",
        "  def observation_spec(self):\n",
        "    return dm_env.specs.BoundedArray(\n",
        "        shape=self.stimuli[0].shape,\n",
        "        dtype=self.stimuli[0].dtype,\n",
        "        name='nback_stimuli', minimum=0, maximum=len(self.stimuli_choices) + 1)\n",
        "\n",
        "  def action_spec(self):\n",
        "    return dm_env.specs.DiscreteArray(\n",
        "        num_values=len(NBack.ACTIONS),\n",
        "        dtype=np.int32,\n",
        "        name='action')\n",
        "\n",
        "  def _observation(self):\n",
        "    if self._current_step ==  self.episode_steps:\n",
        "      obs = self.stimuli[0]\n",
        "    else:\n",
        "      obs = self.stimuli[self._current_step]\n",
        "\n",
        "    return obs\n",
        "\n",
        "  def plot_state(self):\n",
        "    \"\"\"Display current state of the environment.\n",
        "\n",
        "     Note: `M` mean `match`, and `.` is a `non-match`.\n",
        "    \"\"\"\n",
        "    stimuli = self.stimuli[:self._current_step - 1]\n",
        "    actions = ['M' if a=='match' else '.' for a in self._action_history[:self._current_step - 1]]\n",
        "    return HTML(\n",
        "        f'<b>Environment ({self.N}-back):</b><br />'\n",
        "        f'<pre><b>Stimuli:</b> {\"\".join(map(str,map(int,stimuli)))}</pre>'\n",
        "        f'<pre><b>Actions:</b> {\"\".join(actions)}</pre>'\n",
        "    )\n",
        "\n",
        "  @staticmethod\n",
        "  def create_environment(prob = None):\n",
        "    \"\"\"Utility function to create a N-back environment and its spec.\"\"\"\n",
        "\n",
        "    # Make sure the environment outputs single-precision floats.\n",
        "    environment = wrappers.SinglePrecisionWrapper(NBack(prob = prob))\n",
        "\n",
        "    # Grab the spec of the environment.\n",
        "    environment_spec = specs.make_environment_spec(environment)\n",
        "\n",
        "    return environment, environment_spec"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "NQTnbj5V6mAt"
      },
      "source": [
        "### Define a random agent\n",
        "\n",
        "For more information you can refer to NMA-DL W3D2 Basic Reinforcement learning."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "execution": {},
        "id": "_h1bWf8D6mAu"
      },
      "outputs": [],
      "source": [
        "class RandomAgent(acme.Actor):\n",
        "\n",
        "  def __init__(self, environment_spec, action = None):\n",
        "    \"\"\"Gets the number of available actions from the environment spec.\"\"\"\n",
        "    self._num_actions = environment_spec.actions.num_values\n",
        "    self.action = action\n",
        "\n",
        "  def select_action(self, observation):\n",
        "    \"\"\"Selects an action uniformly at random.\"\"\"\n",
        "    if self.action is not None: \n",
        "      return self.action\n",
        "    return np.random.randint(self._num_actions)\n",
        "\n",
        "  def observe_first(self, timestep):\n",
        "    \"\"\"Does not record as the RandomAgent has no use for data.\"\"\"\n",
        "    pass\n",
        "\n",
        "  def observe(self, action, next_timestep):\n",
        "    \"\"\"Does not record as the RandomAgent has no use for data.\"\"\"\n",
        "    pass\n",
        "\n",
        "  def update(self):\n",
        "    \"\"\"Does not update as the RandomAgent does not learn from data.\"\"\"\n",
        "    pass"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Define Learnable Agents"
      ],
      "metadata": {
        "id": "_X_RIcItvaVU"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def dqn_make_network(action_spec: specs.DiscreteArray) -> snt.Module:\n",
        "  return snt.Sequential([\n",
        "      snt.Flatten(),\n",
        "      snt.nets.MLP([128, 128, action_spec.num_values]),\n",
        "  ])\n",
        "\n",
        "def r2d2_make_network(action_spec: specs.DiscreteArray) -> snt.Module:\n",
        "  return snt.DeepRNN([\n",
        "            snt.Flatten(),\n",
        "            snt.LSTM(hidden_size = 10),\n",
        "            snt.nets.MLP([128, 128, action_spec.num_values])\n",
        "            ])"
      ],
      "metadata": {
        "id": "T6A2pa9Zvcpz"
      },
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "135ZABGG6mAw"
      },
      "source": [
        "### Get reward distribution and mean reward of the agent"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "execution": {},
        "id": "VPawgcGr6mAx"
      },
      "outputs": [],
      "source": [
        "def get_reward_distribution(agent, env, n_episodes = 25000,  episode_length = 32, log_loss = False):\n",
        "  # main loop\n",
        "  n_total_steps = 0\n",
        "  n_steps = n_episodes * episode_length\n",
        "  all_returns = []\n",
        "  for episode in range(n_episodes):\n",
        "    episode_steps = 0\n",
        "    episode_return = 0\n",
        "    episode_loss = 0\n",
        "\n",
        "    start_time = time.time()\n",
        "\n",
        "    timestep = env.reset()\n",
        "\n",
        "    # Make the first observation.\n",
        "    agent.observe_first(timestep)\n",
        "\n",
        "    # Run an episode\n",
        "    while not timestep.last():\n",
        "\n",
        "      # Generate an action from the agent's policy and step the environment.\n",
        "      action = agent.select_action(timestep.observation)\n",
        "      timestep = env.step(action)\n",
        "\n",
        "      # Have the agent observe the timestep and let the agent update itself.\n",
        "      agent.observe(action, next_timestep=timestep)\n",
        "      agent.update()\n",
        "\n",
        "      # Book-keeping.\n",
        "      episode_steps += 1\n",
        "      \n",
        "      n_total_steps += 1\n",
        "      print(n_total_steps)\n",
        "      episode_return += timestep.reward\n",
        "\n",
        "      if log_loss:\n",
        "        episode_loss += agent.last_loss\n",
        "\n",
        "      if n_steps is not None and n_total_steps >= n_steps:\n",
        "        break\n",
        "\n",
        "    # Collect the results and combine with counts.\n",
        "    steps_per_second = episode_steps / (time.time() - start_time)\n",
        "    result = {\n",
        "        'episode': episode,\n",
        "        'episode_length': episode_steps,\n",
        "        'episode_return': episode_return,\n",
        "    }\n",
        "    if log_loss:\n",
        "      result['loss_avg'] = episode_loss/episode_steps\n",
        "\n",
        "    all_returns.append(episode_return)\n",
        "\n",
        "    #display(env.plot_state())\n",
        "    # Log the given results.\n",
        "    # print(result)\n",
        "\n",
        "    if n_steps is not None and n_total_steps >= n_steps:\n",
        "      break\n",
        "\n",
        "  return all_returns, np.mean(all_returns), np.std(all_returns)"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Training and Evaluation Script"
      ],
      "metadata": {
        "id": "DMxo_yt5uN5d"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def train_and_eval(env, agent, train , n_episodes,  episode_length, log_loss = False):\n",
        "  if train:\n",
        "    print(\"here\")\n",
        "    loop = EnvironmentLoop(env, agent, logger=loggers.InMemoryLogger())\n",
        "    loop.run(n_episodes)\n",
        "\n",
        "    return pd.DataFrame(loop._logger._data)\n",
        "  \n",
        "  else:\n",
        "    all_returns, mean_return, std_return = get_reward_distribution(agent, env, n_episodes = 25000,  episode_length = 32, log_loss = False)\n",
        "    clear_output()\n",
        "    print(f'Mean Return: {mean_return} | Standard Deviation: {std_return}')\n",
        "    plt.figure()\n",
        "    sns.histplot(all_returns, stat=\"density\", kde=True, bins=12)\n",
        "    plt.xlabel('Return [a.u.]')\n",
        "    plt.ylabel('Density')\n",
        "    plt.show()\n",
        "\n",
        "    return None"
      ],
      "metadata": {
        "id": "_cxysHVguQOG"
      },
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "execution": {},
        "id": "_EGCq1_c6mAu"
      },
      "source": [
        "### Initialize the environments"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "metadata": {
        "execution": {},
        "id": "tuD4BchT6mAv",
        "outputId": "8cf4ee20-6671-44d1-8353-65f932b6dd6c",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "actions:\n",
            " DiscreteArray(shape=(), dtype=int32, name=action, minimum=0, maximum=1, num_values=2)\n",
            "observations:\n",
            " BoundedArray(shape=(), dtype=dtype('float32'), name='nback_stimuli', minimum=0.0, maximum=7.0)\n",
            "rewards:\n",
            " Array(shape=(), dtype=dtype('float32'), name='reward')\n"
          ]
        }
      ],
      "source": [
        "env_balanced, env_spec_balanced = NBack.create_environment()\n",
        "env_unbalanced, env_spec_unbalanced = NBack.create_environment(prob = 0.)\n",
        "\n",
        "print('actions:\\n', env_spec_balanced.actions)\n",
        "print('observations:\\n', env_spec_balanced.observations)\n",
        "print('rewards:\\n', env_spec_balanced.rewards)"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Define Agents"
      ],
      "metadata": {
        "id": "ukAvPj94hwOJ"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "random_agent = RandomAgent(env_spec_balanced)\n",
        "match_agent = RandomAgent(env_spec_balanced, 0)\n",
        "unmatch_agent = RandomAgent(env_spec_balanced, 1)\n",
        "\n",
        "dqn_agent_balanced = dqn.DQN(\n",
        "    environment_spec=env_spec_balanced,\n",
        "    network=dqn_make_network(env_spec_balanced.actions),\n",
        "    epsilon=[0.5],\n",
        "    learning_rate = 0.0001,\n",
        "    min_replay_size = 32,\n",
        "    logger=loggers.InMemoryLogger(),\n",
        "    checkpoint=False,\n",
        ")\n",
        "\n",
        "dqn_agent_unbalanced = dqn.DQN(\n",
        "    environment_spec=env_spec_unbalanced,\n",
        "    network=dqn_make_network(env_spec_unbalanced.actions),\n",
        "    epsilon=[0.5],\n",
        "    learning_rate = 0.0001,\n",
        "    min_replay_size = 32,\n",
        "    logger=loggers.InMemoryLogger(),\n",
        "    checkpoint=False,\n",
        ")\n",
        "\n",
        "drqn_agent_balanced = r2d2.R2D2(\n",
        "    environment_spec=env_spec_balanced,\n",
        "    network=r2d2_make_network(env_spec_balanced.actions),\n",
        "    learning_rate = 0.0001,\n",
        "    burn_in_length=10,\n",
        "    trace_length=2,\n",
        "    replay_period=10,\n",
        "    epsilon=[0.05],\n",
        "    logger=loggers.InMemoryLogger(),\n",
        "    min_replay_size = 32,\n",
        "    checkpoint=False,\n",
        ")\n",
        "\n",
        "drqn_agent_unbalanced = r2d2.R2D2(\n",
        "    environment_spec=env_spec_unbalanced,\n",
        "    network=r2d2_make_network(env_spec_unbalanced.actions),\n",
        "    learning_rate = 0.0001,\n",
        "    burn_in_length=10,\n",
        "    trace_length=2,\n",
        "    replay_period=10,\n",
        "    epsilon=[0.05],\n",
        "    logger=loggers.InMemoryLogger(),\n",
        "    min_replay_size = 32,\n",
        "    checkpoint=False,\n",
        ")"
      ],
      "metadata": {
        "id": "XcvCnGfDhxIC"
      },
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "n_episodes = 25000\n",
        "episode_length = 32"
      ],
      "metadata": {
        "id": "x4yOQ6aY6P9K"
      },
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Random Agent"
      ],
      "metadata": {
        "id": "tI_K_1zDyaz7"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Balanced N-Back"
      ],
      "metadata": {
        "id": "SURbpHPeydLy"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "train_and_eval(env_balanced, random_agent, False, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "id": "7KqHQnnXw-oX",
        "outputId": "2ab69afd-1fed-4f61-bcd7-8f11100a05a1"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -15.52376 | Standard Deviation: 2.800956169310759\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Un-Balanced N-Back"
      ],
      "metadata": {
        "id": "sPQgEENVyf8O"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "train_and_eval(env_unbalanced, random_agent, False, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "id": "4eDnhTWYyBbM",
        "outputId": "edf8601a-0e75-4586-cfdb-10e380cfccb6"
      },
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -15.48716 | Standard Deviation: 2.7800566782711464\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Matching Agent"
      ],
      "metadata": {
        "id": "KHlNxPbMyhzS"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Balanced N-Back"
      ],
      "metadata": {
        "id": "uhblKbMfyjmh"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "train_and_eval(env_balanced, match_agent,False, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "id": "iSwcN1cHyEJf",
        "outputId": "f1d5f851-6d62-4373-9c32-754115de6d52"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -16.18132 | Standard Deviation: 2.74579734459774\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Un-Balanced N-Back"
      ],
      "metadata": {
        "id": "JO5ZCYnHylS0"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "train_and_eval(env_unbalanced, match_agent,False, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "id": "AfdlslUeyEg3",
        "outputId": "1bb82082-e41f-458c-d658-f8989b304370"
      },
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -25.82692 | Standard Deviation: 2.0930559747890163\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Un-Matching Agent"
      ],
      "metadata": {
        "id": "v9KqkvKRyoO4"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Balanced N-Back"
      ],
      "metadata": {
        "id": "-jRSzfu2yqQP"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "train_and_eval(env_balanced, unmatch_agent, False, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "id": "CF2AnpQiyJxl",
        "outputId": "2bba3094-dd2e-4095-caa7-9fd8c5ab72fd"
      },
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -14.8332 | Standard Deviation: 2.754534036820021\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Un-Balanced N-Back"
      ],
      "metadata": {
        "id": "lziRjJDYyrD2"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "train_and_eval(env_unbalanced, unmatch_agent, False, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "id": "eULQxh4KyKGB",
        "outputId": "7f41185b-3326-4318-be24-0dcbf9fc8810"
      },
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -5.15156 | Standard Deviation: 2.077081983552888\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## DQN Agent"
      ],
      "metadata": {
        "id": "ji6I5b0zys59"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Balanced N-Back"
      ],
      "metadata": {
        "id": "1tCSFK6RyuPb"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "logs_dqn_balanced = train_and_eval(env_balanced, dqn_agent_balanced, True, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TR4-WWlQyOAc",
        "outputId": "61f5b569-dabc-4309-9b51-d2cde9e4fff3"
      },
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "here\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "mean_dqn = np.mean(logs_dqn_balanced['episode_return'])\n",
        "std_dqn = np.std(logs_dqn_balanced['episode_return'])\n",
        "\n",
        "print(f'Mean Return: {mean_dqn} | Standard Deviation: {std_dqn}')\n",
        "plot = logs_dqn_balanced.episode_return.hist(bins=12, density=True)\n",
        "plot.set_xlabel('Return [a.u.]')\n",
        "plot.set_ylabel('Density')\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 321
        },
        "id": "OyJ9qlJIx1GN",
        "outputId": "7ab33fe2-492c-4948-9ecd-1d86d17d9e10"
      },
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -15.36624 | Standard Deviation: 2.765217579576696\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0, 0.5, 'Density')"
            ]
          },
          "metadata": {},
          "execution_count": 24
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Un-Balanced N-Back"
      ],
      "metadata": {
        "id": "JRApxz2xyvIn"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "logs_dqn_unbalanced = train_and_eval(env_unbalanced, dqn_agent_unbalanced, True, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "id": "RbHSBs9JyOPL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "932e11db-3d59-41bd-8dcf-53151eaf7e24"
      },
      "execution_count": 25,
      "outputs": [
        {
          "metadata": {
            "tags": null
          },
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "here\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "mean_dqn_ub = np.mean(logs_dqn_unbalanced['episode_return'])\n",
        "std_dqn_ub = np.std(logs_dqn_unbalanced['episode_return'])\n",
        "print(f'Mean Return: {mean_dqn_ub} | Standard Deviation: {std_dqn_ub}')\n",
        "\n",
        "plot_dqn_ub = logs_dqn_unbalanced.episode_return.hist(bins=12, density=True)\n",
        "plot_dqn_ub.set_xlabel('Return [a.u.]')\n",
        "plot_dqn_ub.set_ylabel('Density')\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 321
        },
        "id": "D50ezbuAooDF",
        "outputId": "91099506-b27a-4366-d425-b6ccbafc5d9e"
      },
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -10.86908 | Standard Deviation: 2.9555270179106805\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0, 0.5, 'Density')"
            ]
          },
          "metadata": {},
          "execution_count": 26
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZEAAAEMCAYAAAAF2YvKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3df3RTdZ7/8WcTftWhHEynLem2yMjsYHagyoqyKD9cKU2VlDCcrXWrzI4/isdh5AzLIgWZloLoll2Xowz4g1EctpyR03UOpbHWbldmaMcRZJcB9kTcHaZs5TT9QWsPFMTSNN8/OOZrLLTpldz+yOvxV5L7ubnvd/Pj1Xxu7k1MIBAIICIiYoBloAsQEZGhSyEiIiKGKURERMQwhYiIiBimEBEREcMUIiIiYphCREREDBsx0AWY6bPPLtDd3f/DYuLjx9La2hGBioaGaO4/mnuH6O5fvXdgscRw443f6nVsVIVId3fAUIh8uW40i+b+o7l3iO7+1XvfNJ0lIiKGmRYidXV15OTk4HQ6ycnJ4fTp0z3G1NbWsmTJEqZOnUpxcfFV7+dPf/oTt9566zWXi4iIeUwLkcLCQnJzc3nvvffIzc2loKCgx5jU1FQ2b97MY489dtX78Pv9FBYWkp6eHulyRUQkDKaESGtrK16vF5fLBYDL5cLr9dLW1hYy7qabbsLhcDBixNV31bz22mvcc889TJo0KdIli4hIGEzZse7z+UhKSsJqtQJgtVpJTEzE5/Nhs9nCuo+TJ09SW1vL7t272bFjh6E64uPHGloPICEhzvC6w0E09x/NvUN096/e+zYkvp11+fJlfvazn/H8888Hg8iI1tYOQ9+2SEiIo6XlvOHtDnXR3H809w7R3b96P4/FEtPnP9+mhIjdbqepqQm/34/VasXv99Pc3Izdbg9r/ZaWFurr61m2bBkA586dIxAI0NHRwaZNmyJZuoiI9MKUEImPj8fhcODxeHC73Xg8HhwOR9hTWcnJyRw6dCh4fdu2bVy8eJE1a9ZEqmSRYSduXCxjRvf/Jd+fKZ1LX3Rx/tzn/d6GDF2mTWdt2LCB/Px8duzYwbhx44Jf0c3Ly2PFihVMmzaNI0eO8Pd///d0dHQQCAR455132Lx5M3PmzDGrTJFha8zoEWStKovoNspfcBOdE0DRKyaafh5X+0SMieb+h1PvCQlxpoTIcPp7DZde+qs/+0R0xLqIiBimEBEREcMUIiIiYphCREREDFOIiIiIYQoRERExTCEiIiKGKURERMQwhYiIiBimEBEREcMUIiIiYphCREREDFOIiIiIYQoRERExTCEiIiKGKURERMQwhYiIiBimEBEREcMUIiIiYphCREREDFOIiIiIYaaFSF1dHTk5OTidTnJycjh9+nSPMbW1tSxZsoSpU6dSXFwcsmz79u0sXLiQrKwslixZQk1NjUmVi4jItYwwa0OFhYXk5ubidrspKyujoKCA3bt3h4xJTU1l8+bNVFZW0tnZGbIsLS2NRx99lNjYWE6ePMnDDz9MbW0tY8aMMasFERH5GlM+ibS2tuL1enG5XAC4XC68Xi9tbW0h42666SYcDgcjRvTMtjlz5hAbGwvAlClTCAQCtLe3R754ERG5JlM+ifh8PpKSkrBarQBYrVYSExPx+XzYbLZ+39++ffuYOHEiEyZM6Nd68fFj+72tLyUkxBledziI5v6juXcjhtPfazj10l/h9m7adNb1cvjwYV588UXeeOONfq/b2tpBd3eg3+slJMTR0nK+3+sNF9Hc/3Dq3aw3xOH09xouvfTXl71bLDF9/vNtynSW3W6nqakJv98PgN/vp7m5Gbvd3q/7OXr0KKtXr2b79u3cfPPNkShVRET6wZQQiY+Px+Fw4PF4APB4PDgcjn5NZR0/fpyVK1fy0ksv8f3vfz9SpYqISD+Y9hXfDRs2UFJSgtPppKSkhKKiIgDy8vI4ceIEAEeOHGHu3Lns2rWLt956i7lz5wa/yltUVMSlS5coKCjA7Xbjdrv55JNPzCpfRESuwrR9IpMnT6a0tLTH7Tt37gxenjFjBgcPHrzq+m+//XbEahMREWN0xLqIiBimEBEREcMUIiIiYphCREREDFOIiIiIYQoRERExTCEiIiKGKURERMQwhYiIiBimEBEREcMUIiIiYtiQ+z0RERm8Oi/7TfndkktfdHH+3OcR3470TSEiItfNqJFWslaVRXw75S+4ic6fixp8FCIiAyxuXCxjRuulKEOTnrkiA2zM6BGm/fcucr1px7qIiBimEBEREcMUIiIiYphCREREDFOIiIiIYQoRERExzLQQqaurIycnB6fTSU5ODqdPn+4xpra2liVLljB16lSKi4tDlvn9foqKikhPT2fBggWUlpaaVLmIiFyLaSFSWFhIbm4u7733Hrm5uRQUFPQYk5qayubNm3nsscd6LCsvL6e+vp6qqir27t3Ltm3bOHPmjBmli4jINZgSIq2trXi9XlwuFwAulwuv10tbW1vIuJtuugmHw8GIET2PgayoqCA7OxuLxYLNZiM9PZ3KykozyhcRkWsw5Yh1n89HUlISVqsVAKvVSmJiIj6fD5vNFvZ9JCcnB6/b7XYaGxv7VUd8/Nh+jf8qM04qN5hFc//R3PtgZsbjEs2Pfbi9R9VpT1pbO+juDvR7vYSEOFpaovd0b9Hcvxm9R/Mb1TdhxuMS7c97iyWmz3++TZnOstvtNDU14ff7gSs7yZubm7Hb7f26j4aGhuB1n8/HhAkTrnutIiISPlNCJD4+HofDgcfjAcDj8eBwOMKeygLIzMyktLSU7u5u2traqK6uxul0RqpkEREJg2nfztqwYQMlJSU4nU5KSkooKioCIC8vjxMnTgBw5MgR5s6dy65du3jrrbeYO3cuNTU1ALjdblJSUsjIyOCBBx5g+fLlpKammlW+iIhchWn7RCZPnnzVYzt27twZvDxjxgwOHjx41fWtVmsweEREZHDQEesiImKYQkRERAxTiIiIiGEKERERMUwhIiIihilERETEMIWIiIgYphARERHDFCIiImKYQkRERAxTiIiIiGEKERERMUwhIiIihilERETEMIWIiIgYphARERHDwg6R6upqurq6IlmLiIgMMWGHyEsvvcTs2bPZuHEjx44di2RNIiIyRIQdIvv37+fNN99k9OjRPPXUUzidTnbs2MGZM2ciWZ+IiAxi/doncsstt7BmzRp++9vfUlhYSGVlJQsWLOChhx5i//79dHd3R6pOEREZhEb0d4X6+nr279/P/v37iYmJYcWKFdjtdvbs2UNVVRU///nPI1GniIgMQmGHyJ49eygrK+P//u//uO+++9iyZQu33XZbcLnT6eSuu+665vp1dXXk5+fT3t7O+PHjKS4uZtKkSSFj/H4/zz77LDU1NcTExLBs2TKys7MBaG1tZe3atfh8Prq6upg5cybr169nxIh+56CIiFwnYb8DHzx4kEceeYT58+czatSoHstjY2PZtm3bNdcvLCwkNzcXt9tNWVkZBQUF7N69O2RMeXk59fX1VFVV0d7ezuLFi5k1axYpKSm88sorTJ48mddee43Lly+Tm5tLVVUV999/fz/aFRGR6ynsfSJ33nkn9913X48A2bVrV/Dy7Nmzr7pua2srXq8Xl8sFgMvlwuv10tbWFjKuoqKC7OxsLBYLNpuN9PR0KisrAYiJieHChQt0d3fT2dnJ5cuXSUpKCrd8ERGJgLA/iWzfvp3HHnusx+0vv/wyjzzySK/r+nw+kpKSsFqtAFitVhITE/H5fNhstpBxycnJwet2u53GxkYAfvzjH/PUU08xe/ZsPv/8cx566CFuv/32cMsHID5+bL/Gf1VCQpzhdYeDaO4/mnsfzMx4XKL5sQ+39z5D5Pe//z1wZX/Fhx9+SCAQCC47c+YM3/rWtwyW2D+VlZVMmTKFX/7yl1y4cIG8vDwqKyvJzMwM+z5aWzvo7g70PfBrEhLiaGk53+/1hoto7t+M3qP5jeqbMONxifbnvcUS0+c/332GyDPPPANAZ2cn69atC94eExNDQkIC69ev77Mgu91OU1MTfr8fq9WK3++nubkZu93eY1xDQwNpaWlA6CeTkpISnnvuOSwWC3Fxcdx7770cOnSoXyEiIiLXV58h8v777wPw9NNPs2XLFkMbiY+Px+Fw4PF4cLvdeDweHA5HyFQWQGZmJqWlpWRkZNDe3k51dTV79uwBICUlhYMHD5KWlkZnZye///3vWbBggaF6RETk+gh7x7rRAPnShg0bKCkpwel0UlJSQlFREQB5eXmcOHECALfbTUpKChkZGTzwwAMsX76c1NRUANatW8d//ud/kpWVxeLFi5k0aRIPPPDAN6pJRES+mV4/idx33328++67AMybN4+YmJirjvvNb37T54YmT55MaWlpj9t37twZvGy1WoPh8nUTJ04M+SaYiIgMvF5DZNOmTcHL//RP/xTxYkREZGjpNURmzJgRvHznnXdGvBgRERlawt4nsmvXLj7++GMA/vCHP3DPPfdw7733cvTo0YgVJyIig1vYIfLmm2+SkpICwAsvvMCPfvQjnnzySZ577rmIFSciIoNb2CFy/vx54uLi6Ojo4JNPPmHp0qVkZ2dTV1cXyfpERGQQC/u0J3a7nf/6r//ij3/8IzNmzMBqtdLR0RE8lYmIiESfsEPk6aefZsWKFYwaNYqXXnoJgAMHDjBt2rSIFSciIoNb2CEyb948amtrQ27LzMzUaUdERKJYv37R6fz589TV1XHhwoWQ22fNmnVdixIRkaEh7BD59a9/zcaNG7nhhhsYM2ZM8PaYmBj+4z/+IyLFiYjI4BZ2iGzdupUXX3yRefPmRbIeEREZQsL+iq/f77/mLxeKiEh0CjtE8vLyePnll+nu7o5kPSIiMoSEPZ315ptvcvbsWX7xi18wfvz4kGXhnMVXRESGn7BDRGfxFRGRrws7RHQWXxER+bqw94l0dnaydetW5s+fz+233w5AbW0tJSUlEStOREQGt7BD5LnnnuN//ud/+Od//ufgLxz++Z//Ob/61a8iVpyIiAxuYU9nVVdXU1VVxQ033IDFciV7kpKSaGpqilhxIiIyuIX9SWTkyJH4/f6Q29ra2np8U0tERKJH2CGSmZnJmjVr+PTTTwFobm5m48aNLFy4MGLFiYjI4BZ2iKxcuZLU1FQWLVrEuXPncDqdJCQksHz58rDWr6urIycnB6fTSU5ODqdPn+4xxu/3U1RURHp6OgsWLKC0tDRkeUVFBVlZWbhcLrKysjh79my45YuISASEvU+kvr6e73znOzzxxBP4/X7S09OZMmVK2BsqLCwkNzcXt9tNWVkZBQUF7N69O2RMeXk59fX1VFVV0d7ezuLFi5k1axYpKSmcOHGCn//85/zyl78kISGB8+fPM2rUqPA7FRGR667PTyKBQIC1a9eSlZXFq6++yoEDBygtLeUHP/gBa9euJRAI9LmR1tZWvF4vLpcLAJfLhdfrpa2tLWRcRUUF2dnZWCwWbDYb6enpVFZWAleOmH/00UdJSEgAIC4ujtGjR/e7YRERuX76/CSyd+9eDh8+zN69e0lLSwvefvz4cVatWsVbb73F3/7t3/Z6Hz6fj6SkpOBP6VqtVhITE/H5fNhstpBxycnJwet2u53GxkYATp06RUpKCg899BAXL15kwYIFPPnkk8GvG4cjPn5s2GO/LiEhzvC6w0E09x/NvQ9mZjwu0fzYh9t7nyFSVlbG+vXrQwIEIC0tjXXr1vHqq6/2GSLXg9/v55NPPmHXrl10dnby+OOPk5yczOLFi8O+j9bWDrq7+/7k9HUJCXG0tJzv93rDRTT3b0bv0fxG9U2Y8bhE+/PeYonp85/vPqezTp06xR133HHVZXfccQenTp3qsyC73U5TU1PwK8J+v5/m5mbsdnuPcQ0NDcHrPp+PCRMmAJCcnExmZiajRo1i7NixzJ8/n+PHj/e5bRERiZw+Q8Tv9zN27NWTaOzYsWGdGj4+Ph6Hw4HH4wHA4/HgcDhCprLgyteIS0tL6e7upq2tjerqapxOJ3BlP0ptbS2BQIDLly/z4Ycfcsstt/S5bRERiZw+p7O6urr48MMPr7kD/esHIF7Lhg0byM/PZ8eOHYwbN47i4mLgyu+UrFixgmnTpuF2uzl27BgZGRkALF++nNTUVAAWLlzIf//3f3P//fdjsViYPXs2f/M3fxPWtkVEJDL6DJH4+HjWrVt3zeVf/zRxLZMnT+5x3AfAzp07g5etVitFRUVXXd9isbB27VrWrl0b1vZERCTy+gyR999/34w6RERkCAr7iHUREZGvU4iIiIhhYZ/2RCQadV726zgOkV4oRER6MWqklaxVZRHdRvkL7ojev0gkaTpLREQMU4iIiIhhChERETFMISIiIoYpRERExDCFiIiIGKYQERERwxQiIiJimEJEREQMU4iIiIhhOu2JiAw5ZpzTrPNyeD+4F+0UIiIy5OicZoOHprNERMQwhYiIiBimEBEREcMUIiIiYphpIVJXV0dOTg5Op5OcnBxOnz7dY4zf76eoqIj09HQWLFhAaWlpjzF/+tOfuPXWWykuLjahahER6Y1pIVJYWEhubi7vvfceubm5FBQU9BhTXl5OfX09VVVV7N27l23btnHmzJngcr/fT2FhIenp6WaVLSIivTAlRFpbW/F6vbhcLgBcLhder5e2traQcRUVFWRnZ2OxWLDZbKSnp1NZWRlc/tprr3HPPfcwadIkM8oWEZE+mHKciM/nIykpCavVCoDVaiUxMRGfz4fNZgsZl5ycHLxut9tpbGwE4OTJk9TW1rJ792527NhhqI74+LGGe4j0gU2DXbT3L9Epmp/34fY+JA42vHz5Mj/72c94/vnng0FkRGtrB93dgX6vl5AQR0vLecPbHeqiuf9ofhMRovp539JyHoslps9/vk0JEbvdTlNTE36/H6vVit/vp7m5Gbvd3mNcQ0MDaWlpwP//ZNLS0kJ9fT3Lli0D4Ny5cwQCATo6Oti0aZMZLYiIyFWYEiLx8fE4HA48Hg9utxuPx4PD4QiZygLIzMyktLSUjIwM2tvbqa6uZs+ePSQnJ3Po0KHguG3btnHx4kXWrFljRvkiInINpn07a8OGDZSUlOB0OikpKaGoqAiAvLw8Tpw4AYDb7SYlJYWMjAweeOABli9fTmpqqlkliohIP5m2T2Ty5MlXPe5j586dwctWqzUYLr156qmnrmttIiJijI5YFxERwxQiIiJimEJEREQMU4iIiIhhChERETFMISIiIoYpRERExDCFiIiIGKYQERERwxQiIiJimEJEREQMU4iIiIhhChERETFMISIiIoYpRERExDCFiIiIGKYQERERwxQiIiJimEJEREQMU4iIiIhhChERETFshFkbqqurIz8/n/b2dsaPH09xcTGTJk0KGeP3+3n22WepqakhJiaGZcuWkZ2dDcD27dupqKjAYrEwcuRIVq5cyZw5c8wqXwahuHGxjBlt2lNYRK7CtFdgYWEhubm5uN1uysrKKCgoYPfu3SFjysvLqa+vp6qqivb2dhYvXsysWbNISUkhLS2NRx99lNjYWE6ePMnDDz9MbW0tY8aMMasFGWTGjB5B1qqyiG6j/AV3RO9fZKgzZTqrtbUVr9eLy+UCwOVy4fV6aWtrCxlXUVFBdnY2FosFm81Geno6lZWVAMyZM4fY2FgApkyZQiAQoL293YzyRUTkGkz5JOLz+UhKSsJqtQJgtVpJTEzE5/Nhs9lCxiUnJwev2+12Ghsbe9zfvn37mDhxIhMmTOhXHfHxYw12AAkJcYbXHQ6ivX+JTtH8vA+39yE3oXz48GFefPFF3njjjX6v29raQXd3oN/rJSTE0dJyvt/rDReDtf9ofoGLOQbj894MX77mLZaYPv/5NmU6y26309TUhN/vB67sQG9ubsZut/cY19DQELzu8/lCPm0cPXqU1atXs337dm6++WYzShcRkV6YEiLx8fE4HA48Hg8AHo8Hh8MRMpUFkJmZSWlpKd3d3bS1tVFdXY3T6QTg+PHjrFy5kpdeeonvf//7ZpQtIiJ9MG06a8OGDeTn57Njxw7GjRtHcXExAHl5eaxYsYJp06bhdrs5duwYGRkZACxfvpzU1FQAioqKuHTpEgUFBcH73LJlC1OmTDGrBRER+RrTQmTy5MmUlpb2uH3nzp3By1arlaKioquu//bbb0esNhERMUZHrIuIiGEKERERMUwhIiIihilERETEMIWIiIgYphARERHDFCIiImKYQkRERAwbcidgFBExQ+dlf8RP8nnpiy7On/s8otuINIWIiMhVjBppNeVHz4b6eYI1nSUiIoYpRERExDBNZ8l1FzculjGj9dQSiQZ6pct1N2b0iIjPJcOV+WQRGViazhIREcMUIiIiYphCREREDFOIiIiIYQoRERExTCEiIiKGKURERMQw044TqaurIz8/n/b2dsaPH09xcTGTJk0KGeP3+3n22WepqakhJiaGZcuWkZ2d3ecyCZ/RAwEjfSI6ERmaTAuRwsJCcnNzcbvdlJWVUVBQwO7du0PGlJeXU19fT1VVFe3t7SxevJhZs2aRkpLS6zIJnxkHAuogQJHoYUqItLa24vV62bVrFwAul4tNmzbR1taGzWYLjquoqCA7OxuLxYLNZiM9PZ3Kykoef/zxXpeFy2KJMdzDN1l3sEm8MXZYbMOs7QyXbZi1HfUSPjNONw/wxRdddHRc6tc6FktMWO97poSIz+cjKSkJq9UKgNVqJTExEZ/PFxIiPp+P5OTk4HW73U5jY2Ofy8J1443fMtxDfPxYw+sONq+vzxgW2zBrO8NlG2ZtR70MPqNHj2D06P69h4X7nqcd6yIiYpgpIWK322lqasLv9wNXdpI3Nzdjt9t7jGtoaAhe9/l8TJgwoc9lIiIyMEwJkfj4eBwOBx6PBwCPx4PD4QiZygLIzMyktLSU7u5u2traqK6uxul09rlMREQGRkwgEAiYsaFTp06Rn5/PuXPnGDduHMXFxdx8883k5eWxYsUKpk2bht/vZ+PGjfzud78DIC8vj5ycHIBel4mIyMAwLURERGT40Y51ERExTCEiIiKGKURERMQwhYiIiBimEOlFUVERmZmZLFq0iAcffJATJ04Ely1dupT58+fjdrtxu928/fbbA1jp9ddb72fPnuXRRx/F6XSyaNEijh07NoCVRkZZWRlZWVn8xV/8BSUlJSHL8vPzmTt3bvCxf/nllweoysjorffPP/+cn/70pyxYsIDMzEwOHDgwQFWaY7g/1ldTV1dHTk4OTqeTnJwcTp8+3fsKAbmm999/P9DZ2Rm8PH/+/OCyhx9+OPD+++8PVGkR11vv+fn5ge3btwcCgUDgo48+CixYsCDQ3d09IHVGyieffBL43//938Dq1asD//qv/xqybM2aNT1uG056633btm2BZ555JhAIBAJ1dXWBu+66K9DR0TEQZZpiuD/WV7N06dLAvn37AoFAILBv377A0qVLex2vTyK9+Ou//mtGjhwJwG233UZjYyPd3d0DXJU5euu9srKSBx98EIAZM2YwatSokE8qw8H3vvc9vvvd72KxRN9LpLfe33333eDxWZMmTWLq1KkcPHjQ7BIlQr48Wa7L5QKunCzX6/XS1tZ2zXWi7xVi0J49e7jnnntCXlhbtmwhKyuLf/iHf6CpqWkAq4usr/b+2WefEQgEQs42YORkmEPdrl27yMrK4sc//jGnTp0a6HJM09DQwJ/92Z8Fr0fDYx9Nj3VvJ8u9FtN+T2Qw+sEPfhByPq6v+uCDD4J/yHfeeYfy8nL27NkTXL5lyxbsdjt+v59XX32Vn/70p/zqV78ype7r4Zv0PhyE2//VrFy5koSEBCwWC/v27ePxxx+nurq613UGk2/S+3DT199iqD/WpjBhim1Iq6qqCsyfPz/w6aefXnPM+fPnAw6HI+D3+02sLPKu1futt94aaG1tDV5fuHBh4NixY2aXZ4pw5sTvvPPOwJkzZ0yqyDxX6/3+++8PHD9+PHh92bJlgYqKCrNLGzDD9bH+0tmzZwO33357oKurKxAIBAJdXV2B22+/PeT1/nWazurFgQMHeP7553n99ddDfkGxq6uLs2fPBq+/8847fO973xtW8+fX6h2unAzzrbfeAuDIkSNcunSJqVOnDkSZA+KrU5c1NTVYLBaSkpIGsCLzZGZmsnfvXgBOnz7NiRMnmDNnzgBXFTnR9liHe7Lcr9K5s3rxV3/1V4wcOTLkD/jmm28yevRoHn74YS5fvgxAYmIizzzzDDfffPNAlXrdXav3G2+8kZaWFlavXk1DQwOjR4+mqKiIv/zLvxzAaq8/j8fDli1bOHfuHCNHjiQ2NpY33niD7373u/zoRz+itbWVmJgYxo4dy9NPP81tt9020CVfN731fvHiRfLz8/n444+xWCysXr2a9PT0gS45Yob7Y3011zpZ7rUoRERExLDhM/8iIiKmU4iIiIhhChERETFMISIiIoYpRERExDCFiIiIGKYQERnEDh06xC233ML06dNNOdFhfn4+aWlpzJ07N+LbkuFBISJR59577yUtLY3p06dz9913k5+fz4ULF8Jad+nSpZSWlka4wlCJiYkcPXrUlDf2f/zHf2Tnzp0R344MHwoRiUqvvPIKR48eZd++fXi9Xl577TVTttvV1WXKdkTMohCRqJaQkMDs2bP5+OOPg7f94Q9/4MEHH2TGjBksWrSIQ4cOAbB161aOHDnCxo0bmT59Ohs3buTMmTNMmTIlJBy++mnl17/+NQ8++CDPPfccM2fOZNu2beTn51NUVMSyZcuYPn062dnZ1NfXh11zfX09P/zhD5k5cyYzZ85k1apVnDt37qpjDx061OMTzL333ssHH3wQ9vZEeqMQkajW2NhITU0NEydOBK6ccO+JJ57gySef5PDhw6xZs4YVK1bQ1tbGypUrmTFjBgUFBRw9epSCgoKwtnH8+HFSU1P53e9+x5NPPglARUUFP/nJT/joo4+YOHEiW7duDbvmQCDAE088QU1NDe+++y6NjY1s27at/82LXAcKEYlKy5cvZ/r06cybNw+bzcaKFSuAK78vPnfuXObNmxMhjVAAAAJwSURBVIfFYuHuu+9m6tSp/Pa3vzW8rcTERJYuXcqIESMYM2YMAOnp6aSlpTFixAgWLVoU8kmoLzfddBN33303o0aNwmaz8cgjj/DRRx8Zrk/km4jqH6WS6LV9+3buuusuDh8+zKpVq/jss88YN24cDQ0NVFZWcuDAgeDYrq4uZs6caXhbEyZM6HHbt7/97eDlMWPGcPHixbDv7+zZs2zevJkjR45w4cIFAoEA48aNM1yfyDehEJGoduedd7JkyRKKi4vZsWMHdrsdt9vNs88+G9b6N9xwAwCXLl1i7NixALS0tISMiYmJua41/8u//AsxMTGUl5czfvx4qqur2bhx41XHxsbGcunSpeB1v9/f6+9li/SXprMk6v3d3/0dH3zwASdPnmTRokUcOHCAmpoa/H4/X3zxBYcOHQr+jvi3v/1tPv300+C6NpuNpKQkysrK8Pv9/Nu//VvI8ki4cOECN9xwA3FxcTQ1NfGLX/zimmO/853v8MUXX/Cb3/yGy5cv8/LLL9PZ2RnR+iS6KEQk6tlsNtxuN9u3b8dut7Njxw5effVVZs2axbx583j99dfp7u4G4Ic//CHvvfced9xxR/DTyqZNm3j99deZOXMmf/zjH5k+fXpE6/3JT36C1+tlxowZLFu2jIyMjJDljz/+OK+88goAcXFxFBYWsn79eubOnUtsbGzI9Nr+/ftZuHBhROuV4U0/SiUyiH300Uc89thjjBo1iq1bt0b8p2jXrVtHZWUl8fHx/Pu//3tEtyXDg0JEREQM03SWiIgYphARERHDFCIiImKYQkRERAxTiIiIiGEKERERMUwhIiIihv0/AU//qlZKhWwAAAAASUVORK5CYII=\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## DRQN Agent"
      ],
      "metadata": {
        "id": "VYENVrXxywgf"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Balanced N-Back"
      ],
      "metadata": {
        "id": "PI2ZXu7Kyxyc"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "logs_drqn_balanced = train_and_eval(env_balanced, drqn_agent_balanced, True, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "id": "FKEeEbjByWrp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f29250e5-8c42-4fd4-bc70-1a98554069f3"
      },
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "here\n",
            "INFO:tensorflow:Assets written to: /root/acme/2d5f4468-0d66-11ed-9aaa-0242ac1c0002/snapshots/network/assets\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "mean_drqn = np.mean(logs_drqn_balanced['episode_return'])\n",
        "std_drqn = np.std(logs_drqn_balanced['episode_return'])\n",
        "\n",
        "print(f'Mean Return: {mean_drqn} | Standard Deviation: {mean_drqn}')\n",
        "\n",
        "plot_drqn = logs_drqn_balanced.episode_return.hist(bins=12, density = True)\n",
        "plot_drqn.set_xlabel('Return [a.u.]')\n",
        "plot_drqn.set_ylabel('Density')\n",
        "logs_dqn_balanced.tail()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 536
        },
        "id": "7Ab9UTVDRlFK",
        "outputId": "c828832c-a56b-4643-ced3-58ab5e0320b0"
      },
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -12.623 | Standard Deviation: -12.623\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "       episode_length episode_return  steps_per_second  episodes   steps\n",
              "24995              32          -16.0        354.970056     24996  799872\n",
              "24996              32          -14.0        286.602623     24997  799904\n",
              "24997              32          -18.0        316.785098     24998  799936\n",
              "24998              32          -17.0        343.858826     24999  799968\n",
              "24999              32          -16.0        334.936410     25000  800000"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-5f50a211-4c6c-4c9e-b6bd-d032efa4296e\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>episode_length</th>\n",
              "      <th>episode_return</th>\n",
              "      <th>steps_per_second</th>\n",
              "      <th>episodes</th>\n",
              "      <th>steps</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>24995</th>\n",
              "      <td>32</td>\n",
              "      <td>-16.0</td>\n",
              "      <td>354.970056</td>\n",
              "      <td>24996</td>\n",
              "      <td>799872</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24996</th>\n",
              "      <td>32</td>\n",
              "      <td>-14.0</td>\n",
              "      <td>286.602623</td>\n",
              "      <td>24997</td>\n",
              "      <td>799904</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24997</th>\n",
              "      <td>32</td>\n",
              "      <td>-18.0</td>\n",
              "      <td>316.785098</td>\n",
              "      <td>24998</td>\n",
              "      <td>799936</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24998</th>\n",
              "      <td>32</td>\n",
              "      <td>-17.0</td>\n",
              "      <td>343.858826</td>\n",
              "      <td>24999</td>\n",
              "      <td>799968</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24999</th>\n",
              "      <td>32</td>\n",
              "      <td>-16.0</td>\n",
              "      <td>334.936410</td>\n",
              "      <td>25000</td>\n",
              "      <td>800000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-5f50a211-4c6c-4c9e-b6bd-d032efa4296e')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-5f50a211-4c6c-4c9e-b6bd-d032efa4296e button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-5f50a211-4c6c-4c9e-b6bd-d032efa4296e');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 28
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Un-Balanced N-Back"
      ],
      "metadata": {
        "id": "h7hLLNggyyRG"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "logs_drqn_unbalanced = train_and_eval(env_unbalanced, drqn_agent_unbalanced, True, n_episodes = n_episodes,  episode_length = episode_length, log_loss = False)"
      ],
      "metadata": {
        "id": "y0Kz9jPtyW78",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d5360e6f-1027-43b4-9ab8-339e6ecab98d"
      },
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "here\n",
            "INFO:tensorflow:Assets written to: /root/acme/2d5f4468-0d66-11ed-9aaa-0242ac1c0002/snapshots/network/assets\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "mean_drqn_ub = np.mean(logs_drqn_unbalanced['episode_return'])\n",
        "std_drqn_ub = np.std(logs_drqn_unbalanced['episode_return'])\n",
        "\n",
        "print(f'Mean Return: {mean_drqn_ub} | Standard Deviation: {mean_drqn_ub}')\n",
        "\n",
        "plot_drqn = logs_drqn_unbalanced.episode_return.hist(bins=12, density = True)\n",
        "plot_drqn.set_xlabel('Return [a.u.]')\n",
        "plot_drqn.set_ylabel('Density')\n",
        "logs_drqn_unbalanced.tail()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 536
        },
        "id": "KaX22X7ht2R9",
        "outputId": "cb29ea26-88db-4f4a-99a5-1a8b06249714"
      },
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mean Return: -7.64784 | Standard Deviation: -7.64784\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "       episode_length episode_return  steps_per_second  episodes   steps\n",
              "24995              32           -8.0        342.566942     24996  799872\n",
              "24996              32          -15.0        367.830656     24997  799904\n",
              "24997              32          -10.0        363.534672     24998  799936\n",
              "24998              32           -6.0        368.094957     24999  799968\n",
              "24999              32           -4.0        336.529510     25000  800000"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-0b5ba265-9c93-462b-bdd5-d280d9acbb9e\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>episode_length</th>\n",
              "      <th>episode_return</th>\n",
              "      <th>steps_per_second</th>\n",
              "      <th>episodes</th>\n",
              "      <th>steps</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>24995</th>\n",
              "      <td>32</td>\n",
              "      <td>-8.0</td>\n",
              "      <td>342.566942</td>\n",
              "      <td>24996</td>\n",
              "      <td>799872</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24996</th>\n",
              "      <td>32</td>\n",
              "      <td>-15.0</td>\n",
              "      <td>367.830656</td>\n",
              "      <td>24997</td>\n",
              "      <td>799904</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24997</th>\n",
              "      <td>32</td>\n",
              "      <td>-10.0</td>\n",
              "      <td>363.534672</td>\n",
              "      <td>24998</td>\n",
              "      <td>799936</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24998</th>\n",
              "      <td>32</td>\n",
              "      <td>-6.0</td>\n",
              "      <td>368.094957</td>\n",
              "      <td>24999</td>\n",
              "      <td>799968</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24999</th>\n",
              "      <td>32</td>\n",
              "      <td>-4.0</td>\n",
              "      <td>336.529510</td>\n",
              "      <td>25000</td>\n",
              "      <td>800000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-0b5ba265-9c93-462b-bdd5-d280d9acbb9e')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-0b5ba265-9c93-462b-bdd5-d280d9acbb9e button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-0b5ba265-9c93-462b-bdd5-d280d9acbb9e');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 30
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def moving_average(interval, window_size):\n",
        "    window = np.ones(int(window_size)) / float(window_size)\n",
        "    return np.convolve(interval, window, 'same')  "
      ],
      "metadata": {
        "id": "uro52jc69oCH"
      },
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Balanced Logs\n",
        "dqn_b_ma = moving_average(logs_dqn_balanced['episode_return'], 100)\n",
        "drqn_b_ma = moving_average(logs_drqn_balanced['episode_return'], 100)\n",
        "\n",
        "# dqn_b = list(logs_dqn_balanced['episode_return'])\n",
        "# drqn_b = list(logs_drqn_balanced['episode_return'])\n",
        "\n",
        "plt.plot(dqn_b_ma, color='blue')\n",
        "plt.plot(drqn_b_ma, color='yellow')\n",
        "\n",
        "# plt.plot(dqn_b, color='blue')\n",
        "# plt.plot(drqn_b, color='yellow')\n",
        "\n",
        "plt.ylabel(\"episode_reward\")\n",
        "plt.xlabel(\"episode (N_steps per episode = 32)\")\n",
        "plt.legend(['DQN','DRQN'])\n",
        "plt.title(\"Equal match and non-match\")"
      ],
      "metadata": {
        "id": "gLIG5S-K1tB-",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 319
        },
        "outputId": "05792fdf-3185-404b-d331-888113b27f60"
      },
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Equal match and non-match')"
            ]
          },
          "metadata": {},
          "execution_count": 32
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Un-Balanced Logs\n",
        "dqn_u_ma = moving_average(logs_dqn_unbalanced['episode_return'], 100)\n",
        "drqn_u_ma = moving_average(logs_drqn_unbalanced['episode_return'], 100)\n",
        "\n",
        "# dqn_u = list(logs_dqn_unbalanced['episode_return'])\n",
        "# drqn_u = list(logs_drqn_unbalanced['episode_return'])\n",
        "\n",
        "plt.plot(dqn_u_ma, color='blue')\n",
        "plt.plot(drqn_u_ma, color='yellow')\n",
        "\n",
        "# plt.plot(dqn_u, color='blue')\n",
        "# plt.plot(drqn_u, color='yellow')\n",
        "\n",
        "plt.ylabel(\"episode_reward\")\n",
        "plt.xlabel(\"episode (N_steps per episode = 32)\")\n",
        "plt.legend(['DQN','DRQN'])\n",
        "plt.title(\"Unequal match and non-match\")"
      ],
      "metadata": {
        "id": "v1CkmCXg1uLI",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 319
        },
        "outputId": "ca791850-d716-4171-aa34-744fe9db533c"
      },
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Unequal match and non-match')"
            ]
          },
          "metadata": {},
          "execution_count": 33
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [],
      "name": "Mohinga_Project_RL_One.ipynb",
      "provenance": []
    },
    "kernel": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "nbformat": 4,
  "nbformat_minor": 0
}