evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "JW4JqPPWt5rX"
   },
   "outputs": [],
   "source": [
    "from simulation import run_simulation\n",
    "import numpy as np\n",
    "from policies import *\n",
    "from environment import Servers\n",
    "import sys"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "eQI1IAY6t5re"
   },
   "outputs": [],
   "source": [
    "\n",
    "# Initialize basic variables\n",
    "num_requests = 15#int(input(\"num_requests: \"))\n",
    "num_servers = 4#int(input(\"num_servers: \"))\n",
    "num_trajectories_list = [0] + [10 ** i for i in range(1, 7)]\n",
    "start_trajectory_index = 0\n",
    "num_boot = 5\n",
    "# end_trajectory_index = 1\n",
    "\n",
    "# Initialize policies\n",
    "behavior_policy = ('UniRan', UniRan(num_servers))\n",
    "target_policy1 = ('LeastLoad_0.3', LeastLoad(num_servers, epsilon=0.3))\n",
    "target_policy2 = ('LeastLoad_0', LeastLoad(num_servers, epsilon=0))\n",
    "target_policy3 = ('EpsilonGreedy', EpsilonGreedy(num_servers, epsilon=0.3))\n",
    "target_policy4 = ('UCB1', UCB1(num_servers))\n",
    "\n",
    "# behavior_policy = ('EpsilonGreedy', EpsilonGreedy(num_servers, epsilon=0.3))\n",
    "# target_policy1 = ('LeastLoad_0.3', LeastLoad(num_servers, epsilon=0.3))\n",
    "# target_policy2 = ('LeastLoad_0', LeastLoad(num_servers, epsilon=0))\n",
    "# target_policy3 = ('UniRan', UniRan(num_servers))\n",
    "# target_policy4 = ('UCB1', UCB1(num_servers))\n",
    "\n",
    "\n",
    "target_policies = [target_policy1, target_policy2, target_policy3, target_policy4]\n",
    "num_target_policies = len(target_policies)\n",
    "\n",
    "policies = [target_policy1, target_policy2, target_policy3, target_policy4, behavior_policy]\n",
    "num_policies = 1 + num_target_policies\n",
    "\n",
    "target_policy_names = [policy[0] for policy in policies if policy != policies[-1]]\n",
    "\n",
    "# Initialize environment (servers)\n",
    "servers = Servers(num_policies=num_policies, num_servers=num_servers)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "W_VM-IoVEa2x"
   },
   "outputs": [],
   "source": [
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 102
    },
    "colab_type": "code",
    "id": "Jgz0grKLEU9p",
    "outputId": "58f45f0f-c4cb-4773-ae2c-4d5b7f89a385"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "99999  complete\n",
      "199999  complete\n",
      "299999  complete\n",
      "399999  complete\n",
      "499999  complete\n"
     ]
    }
   ],
   "source": [
    "trace_lst = []\n",
    "latency_lst = []\n",
    "for _ in range(5*10**5):\n",
    "    total_latency_for_each_policy, trace = \\\n",
    "                run_simulation(policies=policies, num_requests=num_requests, servers=servers)\n",
    "    trace_lst.append(trace)\n",
    "    latency_lst.append(total_latency_for_each_policy)\n",
    "    if _ % 100000 == 99999:\n",
    "        print(_, ' complete')\n",
    "with open('trace1.p', 'wb') as fp:\n",
    "    pickle.dump(trace_lst, fp)           \n",
    "with open('latency1.p', 'wb') as fp:\n",
    "    pickle.dump(latency_lst, fp)  \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 102
    },
    "colab_type": "code",
    "id": "ePdYu26v2i2d",
    "outputId": "62bf03d9-4f34-4f78-dec6-a6730f1a0b89"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "99999  complete\n",
      "199999  complete\n",
      "299999  complete\n",
      "399999  complete\n",
      "499999  complete\n"
     ]
    }
   ],
   "source": [
    "trace_lst = []\n",
    "latency_lst = []\n",
    "for _ in range(5*10**5):\n",
    "    total_latency_for_each_policy, trace = \\\n",
    "                run_simulation(policies=policies, num_requests=num_requests, servers=servers)\n",
    "    trace_lst.append(trace)\n",
    "    latency_lst.append(total_latency_for_each_policy)\n",
    "    if _ % 100000 == 99999:\n",
    "        print(_, ' complete')\n",
    "with open('trace2.p', 'wb') as fp:\n",
    "    pickle.dump(trace_lst, fp)           \n",
    "with open('latency2.p', 'wb') as fp:\n",
    "    pickle.dump(latency_lst, fp)  \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Z8T3LimYQ9_l"
   },
   "outputs": [],
   "source": [
    "trace=pickle.load(open('trace1.p', 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "tFkiBS3va8tt",
    "outputId": "1dcbe6c5-7359-4c6d-96f5-b10ca8d69826"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "499999  complete\n"
     ]
    }
   ],
   "source": [
    "pol_rho1 = [np.zeros((len(trace), num_requests)), np.zeros((len(trace), num_requests)),\n",
    "           np.zeros((len(trace), num_requests)), np.zeros((len(trace), num_requests))]\n",
    "pol_r1 = [np.zeros((len(trace), num_requests)), np.zeros((len(trace), num_requests)),\n",
    "           np.zeros((len(trace), num_requests)), np.zeros((len(trace), num_requests))]\n",
    "for i in range(len(trace)):\n",
    "    for k in range(len(target_policies)):\n",
    "        target_policies[k][1].reset()\n",
    "    if i%500000 == 499999:\n",
    "        print(i, ' complete')\n",
    "    for j in range(num_requests):\n",
    "        state, action, r, old_p = trace[i][j]\n",
    "        server_load, lst_rewards = state\n",
    "        for k in range(len(target_policies)):\n",
    "            for a_, r_ in lst_rewards:\n",
    "                target_policies[k][1].update(a_, r_)\n",
    "            new_p = target_policies[k][1].choose_action(context=server_load, return_prob=True)\n",
    "            if j == 0:\n",
    "                pol_rho1[k][i,j] = new_p[action]/old_p[action]\n",
    "            else:\n",
    "                pol_rho1[k][i,j] = new_p[action]/old_p[action]*pol_rho1[k][i, j-1]\n",
    "            pol_r1[k][i,j] = r"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "vnoD3MBB22Lz"
   },
   "outputs": [],
   "source": [
    "trace = None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "cVAxHpSy274F"
   },
   "outputs": [],
   "source": [
    "trace=pickle.load(open('trace2.p', 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 187
    },
    "colab_type": "code",
    "id": "0WQt3opG3AZo",
    "outputId": "67057d37-a2ca-4bb7-d9b5-9296533b4352"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "49999  complete\n",
      "99999  complete\n",
      "149999  complete\n",
      "199999  complete\n",
      "249999  complete\n",
      "299999  complete\n",
      "349999  complete\n",
      "399999  complete\n",
      "449999  complete\n",
      "499999  complete\n"
     ]
    }
   ],
   "source": [
    "pol_rho2 = [np.zeros((len(trace), num_requests)), np.zeros((len(trace), num_requests)),\n",
    "           np.zeros((len(trace), num_requests)), np.zeros((len(trace), num_requests))]\n",
    "pol_r2 = [np.zeros((len(trace), num_requests)), np.zeros((len(trace), num_requests)),\n",
    "           np.zeros((len(trace), num_requests)), np.zeros((len(trace), num_requests))]\n",
    "for i in range(len(trace)):\n",
    "    for k in range(len(target_policies)):\n",
    "        target_policies[k][1].reset()\n",
    "    if i%50000 == 49999:\n",
    "        print(i, ' complete')\n",
    "    for j in range(num_requests):\n",
    "        state, action, r, old_p = trace[i][j]\n",
    "        server_load, lst_rewards = state\n",
    "        for k in range(len(target_policies)):\n",
    "            for a_, r_ in lst_rewards:\n",
    "                target_policies[k][1].update(a_, r_)\n",
    "            new_p = target_policies[k][1].choose_action(context=server_load, return_prob=True,\n",
    "                                                        behavior_policy=False, chosen_action=action)\n",
    "            if j == 0:\n",
    "                pol_rho2[k][i,j] = new_p[action]/old_p[action]\n",
    "            else:\n",
    "                pol_rho2[k][i,j] = new_p[action]/old_p[action]*pol_rho2[k][i, j-1]\n",
    "            pol_r2[k][i,j] = r"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "qkD69zNe3Ic8"
   },
   "outputs": [],
   "source": [
    "trace = None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "6NVXygF_3JVE"
   },
   "outputs": [],
   "source": [
    "pol_rho = np.concatenate((pol_rho1, pol_rho2), axis=1)\n",
    "pol_r = np.concatenate((pol_r1, pol_r2), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ntpqb9S7r5UQ"
   },
   "outputs": [],
   "source": [
    "with open('pol_rho.p', 'wb') as fp:\n",
    "    pickle.dump(pol_rho, fp)           \n",
    "with open('pol_r.p', 'wb') as fp:\n",
    "    pickle.dump(pol_r, fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "zX7_R7bRktJV"
   },
   "outputs": [],
   "source": [
    "pol_rho=pickle.load(open('pol_rho.p', 'rb'))\n",
    "pol_r=pickle.load(open('pol_r.p', 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Bqvvin8FlDXF"
   },
   "outputs": [],
   "source": [
    "def get_estimates(pol_rho, pol_r, n_samples):\n",
    "    D = n_samples\n",
    "    estimates = []\n",
    "    for i in range(4):\n",
    "        rhos = pol_rho[i][:D]\n",
    "        rs = pol_r[i][:D]\n",
    "        IS = np.sum(rhos[:,-1]*np.sum(rs, axis=1)/D)\n",
    "        SIS = np.sum(np.sum(rhos*rs/D, axis=1))\n",
    "        ws = np.sum(rhos, axis=0)\n",
    "        WIS = np.sum(rhos[:,-1]/ws[-1]*np.sum(rs, axis=1))\n",
    "        SWIS = np.sum(np.sum(rhos/ws*rs, axis=1))\n",
    "        estimates.append((IS, SIS, WIS, SWIS))\n",
    "    return estimates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 105
    },
    "colab_type": "code",
    "id": "IUxzzeiGZsmc",
    "outputId": "d353b41a-ef42-4298-b190-998c6bb63c24"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:10: RuntimeWarning: invalid value encountered in true_divide\n",
      "  # Remove the CWD from sys.path while we load stuff.\n",
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:11: RuntimeWarning: invalid value encountered in true_divide\n",
      "  # This is added back by InteractiveShellApp.init_path()\n"
     ]
    }
   ],
   "source": [
    "estimates = []\n",
    "for i in range(1, 7):\n",
    "  estimates.append(get_estimates(pol_rho, pol_r, 10**i))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 190
    },
    "colab_type": "code",
    "id": "dkeTIg5wZlNR",
    "outputId": "1d137af7-466e-4c9e-8eee-3d3a9e0f29bd"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:10: RuntimeWarning: invalid value encountered in true_divide\n",
      "  # Remove the CWD from sys.path while we load stuff.\n",
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:11: RuntimeWarning: invalid value encountered in true_divide\n",
      "  # This is added back by InteractiveShellApp.init_path()\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20  complete\n",
      "40  complete\n",
      "60  complete\n",
      "80  complete\n",
      "100  complete\n"
     ]
    }
   ],
   "source": [
    "boot_estimates = [estimates]\n",
    "for i in range(100):\n",
    "  if i % 20 == 19:\n",
    "    print(i+1, ' complete')\n",
    "  boot_ind = np.random.choice(range(10**6), 10**6)\n",
    "  boot_rho = pol_rho[:, boot_ind, :]\n",
    "  boot_r = pol_r[:, boot_ind, :]\n",
    "  estimates = []\n",
    "  for i in range(1, 7):\n",
    "    estimates.append(get_estimates(boot_rho, boot_r, 10**i))\n",
    "  boot_estimates.append(np.array(estimates))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nVsOQ7Ycaoho"
   },
   "outputs": [],
   "source": [
    "boot_estimates = np.array(boot_estimates)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "4ItVmfT14HOV"
   },
   "outputs": [],
   "source": [
    "latency1=pickle.load(open('latency1.p', 'rb'))\n",
    "latency2=pickle.load(open('latency2.p', 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "1YMSMXqTapb_"
   },
   "outputs": [],
   "source": [
    "latency = np.concatenate((latency1, latency2), axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "cNTY-hr34ide"
   },
   "outputs": [],
   "source": [
    "true_val = latency.mean(axis=0)[:4]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "r31jjE0JbKym"
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 538
    },
    "colab_type": "code",
    "id": "9stvfoNvYRA_",
    "outputId": "e6ebc28e-fa9f-4498-94c2-ba733fced0b8"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "a = true_val[0]\n",
    "x = [10**i for i in range(1,7)]\n",
    "for i in range(4): \n",
    "  error = np.abs((a - boot_estimates[:,:,0,i])/a)\n",
    "  plt.plot(x, np.mean(error, axis=0))\n",
    "plt.xscale('log')\n",
    "plt.legend(['IS', 'SIS', 'WIS', 'SWIS'])\n",
    "plt.title('Least Load 0.3')\n",
    "plt.show()\n",
    "\n",
    "fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2)\n",
    "axs = [ax1, ax2, ax3, ax4]\n",
    "titles = ['IS', 'SIS', 'WIS', 'SWIS']\n",
    "for i in range(4): \n",
    "  error = np.abs((a - boot_estimates[:,:,0,i])/a)\n",
    "  axs[i].plot(x, np.mean(error, axis=0))\n",
    "  axs[i].errorbar(x, np.mean(error, axis=0), np.std(error, axis=0))\n",
    "  axs[i].set_xscale('log')\n",
    "  axs[i].title.set_text(titles[i])\n",
    "  axs[i].set_ylim(0,1)\n",
    "  axs[i].set_xticks([])\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 285
    },
    "colab_type": "code",
    "id": "Au4o0qKrY7Ff",
    "outputId": "ec927936-ca39-4de6-f515-8aa923d73732"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "a = true_val[1]\n",
    "for i in [1]: \n",
    "  error = np.abs((a - boot_estimates[:,:,1,i])/a)\n",
    "  plt.plot(x, np.mean(error, axis=0))\n",
    "  plt.errorbar(x, np.mean(error, axis=0), np.std(error, axis=0), ecolor='r')\n",
    "plt.xscale('log')\n",
    "plt.ylim(0,1)\n",
    "plt.legend(['SIS'])\n",
    "plt.title('Least Load 0')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 538
    },
    "colab_type": "code",
    "id": "zK9nI8b_jpeZ",
    "outputId": "107d3fbd-2a9b-4926-c65c-0460e587ae5e"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "a = true_val[2]\n",
    "for i in range(4): \n",
    "  error = np.abs((a - boot_estimates[:,:,2,i])/a)\n",
    "  plt.plot(x, np.mean(error, axis=0))\n",
    "plt.xscale('log')\n",
    "plt.legend(['IS', 'SIS', 'WIS', 'SWIS'])\n",
    "plt.title('Epsilon Greedy')\n",
    "plt.show()\n",
    "\n",
    "fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2)\n",
    "axs = [ax1, ax2, ax3, ax4]\n",
    "titles = ['IS', 'SIS', 'WIS', 'SWIS']\n",
    "for i in range(4): \n",
    "  error = np.abs((a - boot_estimates[:,:,2,i])/a)\n",
    "  axs[i].plot(x, np.mean(error, axis=0))\n",
    "  axs[i].errorbar(x, np.mean(error, axis=0), np.std(error, axis=0))\n",
    "  axs[i].set_xscale('log')\n",
    "  axs[i].title.set_text(titles[i])\n",
    "  axs[i].set_ylim(0,1)\n",
    "  axs[i].set_xticks([])\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 285
    },
    "colab_type": "code",
    "id": "VULReZCmkOgw",
    "outputId": "7a0a36c3-fbf4-4058-8a54-cb7b64cfad0b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "a = true_val[3]\n",
    "for i in [1]: \n",
    "  error = np.abs((a - boot_estimates[:,:,3,i])/a)\n",
    "  plt.plot(x, np.mean(error, axis=0))\n",
    "  plt.errorbar(x, np.mean(error, axis=0), np.std(error, axis=0), ecolor='r')\n",
    "plt.xscale('log')\n",
    "plt.ylim((0, 1))\n",
    "plt.legend(['SIS'])\n",
    "plt.title('UCB1')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "evaluation.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}