notebooks/06_sudoku_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Summary\n",
    "\n",
    "In this notebook, we test the accuracy of a network trained to solve Sudoku puzzles on the *validation dataset*, comprised of puzzles generated using the same `sugen` progam that was used to generate the training dataset, and on the *test dataset*, comprised of 30 Sudoku puzzles collected from http://1sudoku.com.\n",
    "\n",
    "----"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "eoBMUoW2Hvhp",
    "toc-hr-collapsed": true
   },
   "source": [
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "TbKxMUZWHvhq"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/strokach/env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:526: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint8 = np.dtype([(\"qint8\", np.int8, 1)])\n",
      "/home/strokach/env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:527: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint8 = np.dtype([(\"quint8\", np.uint8, 1)])\n",
      "/home/strokach/env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:528: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint16 = np.dtype([(\"qint16\", np.int16, 1)])\n",
      "/home/strokach/env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:529: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint16 = np.dtype([(\"quint16\", np.uint16, 1)])\n",
      "/home/strokach/env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:530: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint32 = np.dtype([(\"qint32\", np.int32, 1)])\n",
      "/home/strokach/env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:535: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  np_resource = np.dtype([(\"resource\", np.ubyte, 1)])\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import sys\n",
    "import tempfile\n",
    "import time\n",
    "from collections import deque\n",
    "from pathlib import Path\n",
    "\n",
    "import tqdm\n",
    "from torch_geometric.data import DataLoader\n",
    "\n",
    "import matplotlib as mpl\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import pyarrow\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from matplotlib import cm\n",
    "from tensorflow.python.summary.summary_iterator import summary_iterator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 334
    },
    "colab_type": "code",
    "id": "rczy7pPiHvhs",
    "outputId": "8e4673d7-4cc7-4f8e-ffc4-de94302e5fab"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/strokach/env/lib/python3.7/site-packages/Bio/KDTree/__init__.py:27: BiopythonDeprecationWarning: Bio.KDTree has been deprecated, and we intend to remove it in a future release of Biopython. Please use Bio.PDB.kdtrees instead, which is functionally very similar.\n",
      "  BiopythonDeprecationWarning,\n"
     ]
    }
   ],
   "source": [
    "import proteinsolver\n",
    "import proteinsolver.datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "try:\n",
    "    inline_rc\n",
    "except NameError:\n",
    "    inline_rc = mpl.rcParams.copy()\n",
    "    \n",
    "mpl.rcParams.update({\"font.size\": 12})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "dsEY3dtLHvhy",
    "toc-hr-collapsed": true
   },
   "source": [
    "## Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "UNIQUE_ID = \"c8de7e56\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PosixPath('/tmp/strokach')"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "DATA_ROOT = Path(tempfile.gettempdir())\n",
    "DATA_ROOT.mkdir(exist_ok=True)\n",
    "DATA_ROOT"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PosixPath('sudoku_analysis')"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "NOTEBOOK_NAME = \"sudoku_analysis\"\n",
    "NOTEBOOK_PATH = Path(NOTEBOOK_NAME)\n",
    "NOTEBOOK_PATH.mkdir(exist_ok=True)\n",
    "NOTEBOOK_PATH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PosixPath('/home/kimlab1/database_data/datapkg_output_dir')"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "DATAPKG_DATA_DIR = Path(f\"~/datapkg_data_dir\").expanduser().resolve()\n",
    "DATAPKG_DATA_DIR"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device(type='cpu')"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device = torch.device(\"cuda:2\" if torch.cuda.is_available() else \"cpu\")\n",
    "device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/home/kimlab1/database_data/datapkg_output_dir'"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "proteinsolver.settings.data_url = DATAPKG_DATA_DIR.as_posix()\n",
    "proteinsolver.settings.data_url"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets[\"sudoku_valid_0\"] = proteinsolver.datasets.SudokuDataset4(\n",
    "    root=DATA_ROOT.joinpath(\"sudoku_valid_0\"), subset=f\"valid_0\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets[\"sudoku_test\"] = proteinsolver.datasets.SudokuDataset2(\n",
    "    root=DATA_ROOT.joinpath(\"sudoku_test\"),\n",
    "    data_url=DATAPKG_DATA_DIR.joinpath(\n",
    "        \"deep-protein-gen\", \"sudoku\", \"sudoku_test.csv.gz\"\n",
    "    ).as_posix(),\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "dfs = []\n",
    "\n",
    "for tb_file in Path(f\"sudoku_train/runs/{UNIQUE_ID}.xxx/\").glob(\"*\"):\n",
    "    data = []\n",
    "    for e in summary_iterator(tb_file.as_posix()):\n",
    "        for v in e.summary.value:\n",
    "            row = (e.step, v.tag, v.simple_value)\n",
    "            data.append(row)\n",
    "            if v.tag == \"accuracy_mv\":\n",
    "                row = (e.step, \"wall_time\", e.wall_time)\n",
    "                data.append(row)\n",
    "    df = (\n",
    "        pd.DataFrame(data, columns=[\"datapoint\", \"tag\", \"simple_value\"])\n",
    "        .pivot(index=\"datapoint\", columns=\"tag\", values=\"simple_value\")\n",
    "        .reset_index()\n",
    "    )\n",
    "    dfs.append(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "95281\n"
     ]
    }
   ],
   "source": [
    "df = pd.concat(dfs, sort=False)\n",
    "print(len(df))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "92813\n"
     ]
    }
   ],
   "source": [
    "df = df.sort_values(\"wall_time\", ascending=False).drop_duplicates(subset=[\"datapoint\"], keep=\"first\").sort_values(\"wall_time\", ascending=True)\n",
    "print(len(df))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "assert len(df) == len(df[\"datapoint\"].drop_duplicates())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fg, ax = plt.subplots(figsize=(5, 4))\n",
    "ax.plot(df[\"datapoint\"], df[\"accuracy_m\"], color=cm.get_cmap(\"tab10\")(0), alpha=1, label=\"training\")\n",
    "ax.plot(df[\"datapoint\"], df[\"accuracy_mv\"], color=cm.get_cmap(\"tab10\")(1), alpha=0.8, label=\"validation\")\n",
    "# ax.hlines(0.7, 0, df[\"datapoint\"].max(), linestyle='--')\n",
    "ax.legend(loc=\"lower right\")\n",
    "ax.set_xlabel(\"Number of training data points\")\n",
    "ax.set_ylabel(\"Accuracy on missing numbers\")\n",
    "ax.ticklabel_format(axis=\"x\", style=\"sci\", scilimits=(0, 0))\n",
    "\n",
    "# ax.set_ylim(0, 0.82)\n",
    "fg.savefig(NOTEBOOK_PATH / f\"{UNIQUE_ID}-training-curve.svg\", bbox_inches=\"tight\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%run sudoku_train/{UNIQUE_ID}/model.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[PosixPath('sudoku_train/c8de7e56/e1-s400000-d2400000-amv07269.state')]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "state_files = sorted(\n",
    "    Path(\"sudoku_train\").joinpath(UNIQUE_ID).glob(\"*.state\"),\n",
    "    key=lambda s: (int(s.stem.split(\"-\")[3].strip(\"amv\")), int(s.stem.split(\"-\")[2].strip(\"d\"))),\n",
    ")\n",
    "\n",
    "state_files[-5:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PosixPath('sudoku_train/c8de7e56/e1-s400000-d2400000-amv07269.state')"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "state_file = state_files[-1]\n",
    "state_file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "net = Net(\n",
    "    x_input_size=13, adj_input_size=3, hidden_size=162, output_size=9, batch_size=8\n",
    ").to(device)\n",
    "\n",
    "net.load_state_dict(torch.load(state_file, map_location=device))\n",
    "net.eval()\n",
    "net = net.to(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Test plots"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Oneshot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "edge_attr = next(iter(datasets[\"sudoku_valid_0\"])).edge_attr.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = []\n",
    "for data in datasets[\"sudoku_test\"]:\n",
    "    data = data.to(device)\n",
    "    is_missing = data.x == 9\n",
    "    output = net(data.x, data.edge_index, edge_attr)\n",
    "    output = torch.softmax(output, dim=1)\n",
    "    _, predicted = output.max(dim=1)\n",
    "    num_correct = float((predicted[is_missing] == data.y[is_missing]).sum())\n",
    "    num_total = float(is_missing.sum())\n",
    "    results.append({\"fraction_correct\": num_correct / num_total, \"num_correct\": num_correct, \"num_total\": num_total})\n",
    "oneshot_results_df = pd.DataFrame(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVwAAAEUCAYAAABnHY4tAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3deZwU9Z3/8deHmcg1MyCHTLgGnIgGVDxAjeEwaIwx3hdyCJ5EXY27uj+JGhGPEHSjZBOPRIMKKoq7ARWNR4wXGi/MoitmPJBBECGAwnAfw+f3R9X09gxz1Mx0VzfN+/l49GO6v1Vd9flWV3+m+ltV36+5OyIikn4tMh2AiMjuQglXRCQmSrgiIjFRwhURiYkSrohITJRwRURikp/pAJqqU6dO3qtXr0yHISKS8N57761y9851Td9lE26vXr2YN29epsMQEUkws8X1TVeTgohITJRwRURiooQrIhITJVwRkZgo4YqIxCS2qxTM7DLgXOAA4FF3Pzdp2tHAXUBP4G3gXHev92xfQ1588UU2bdrUnEWI7FZat27NMccck+kwclqcl4UtA24BfgS0rio0s07ALOBCYA5wMzATOKI5K9u0aRMnnnhicxYhsluZM2dOpkPIebElXHefBWBmA4DuSZNOAxa4+3+F0ycCq8xsP3cviys+EZF0y4YbH/oB71e9cPcNZrYwLK+WcM1sHDAOoGfPnnHGKCI5rtfPn6l3evnknzR7Hdlw0qwAWFujbC1QWHNGd7/X3Qe4+4DOneu8e263s2LFCoYMGUJhYSFXXXVV2tf34x//mGnTpqV9PSK5JhsS7nqgqEZZEbAuXSs86qij2HPPPdmyZUu6VhGre++9l06dOlFRUcHtt9+e0mVPnDiR0aNHVyt79tlnGTt2bErXk83MjM8++6xR75kxYwYlJSW0bduWU045ha+//rrOeefPn8/gwYNp164d3bt356abbkpMe/nllznggANo3749HTt25NRTT+XLL79MTL/66qvp0aMHRUVFlJSU8Mtf/jIxbe7cuRQUFFR7mBl/+tOfGlUXSZ1sSLgLgP5VL8ysLVAalqdceXk5c+fOxcx46qmn0rEKtm/fnpbl1mXx4sX07dsXM8uKeDKptrrGXf8FCxbw05/+lIceeogVK1bQpk0bLr300jrnHzlyJEOGDOHrr7/m1Vdf5Z577knsm3379uX5559nzZo1LFu2jH322YdLLrkk8d4LLriAsrIyKioq+Nvf/saMGTOYNWsWAIMHD2b9+vWJx9NPP01BQQHHHXdcejeA1Cm2hGtm+WbWCsgD8syslZnlA7OB/c3s9HD6BOCDdJ0wmz59OkcccQTnnntutZ/Fb731FsXFxVRWVibKZs+ezYEHHgjAjh07mDx5MqWlpXTs2JGzzjorcdRSXl6OmTF16lR69uzJsGHDADjzzDMpLi6mXbt2DBkyhAUL/u9/yOrVqznxxBMpKipi4MCB/OIXv2DQoEGJ6WVlZfzwhz+kQ4cO7Lvvvjz++OO11qeqHrfddhsFBQW8+OKLTJw4kTPOOIPRo0dTVFTEgw8+yDvvvMP3vvc92rdvz7e//W0uu+wytm7dmljOggULEuvr0qULkyZN4rnnnmPSpEnMnDmTgoIC+vcP/i8eddRR/PGPf0xsl1tuuYWSkhL22msvxowZw9q1a6ttl2nTptGzZ086depU7Qispk2bNnHVVVdRUlJCu3btGDRoUOLSvqeeeop+/frRvn17jjrqKP7xj38k3terVy9uvfVWDjzwQNq2bcv27dtrLVu2bBmnn346nTt3pnfv3vz2t79NLKOyspJJkyZRWlpKYWEhhx56KEuWLGHIkCEA9O/fn4KCAmbOnFln/FUeeeQRTjzxRIYMGUJBQQE333wzs2bNYt262n+0lZeXM2rUKPLy8igtLWXQoEGJfaVLly507do1MW9eXl61o+19992Xtm3bJl63aNGizqPxadOmccYZZ1SbX2Lm7rE8gImA13hMDKcdQ3CCbBPwCtCroeUdeuihXp+nnnqq1vLS0lK/6667fN68eZ6fn+/Lly9PTNt77739hRdeSLw+44wz/Fe/+pW7u0+ZMsUPP/xwX7JkiW/evNnHjRvnZ599tru7L1q0yAE/55xzfP369b5x40Z3d586dapXVFT45s2b/YorrvD+/fsnlj18+HAfPny4b9iwwRcsWODdu3f373//++7uvn79eu/evbvff//9vm3bNn/vvfe8Y8eO/uGHH9Zap7Fjx/p1112XeH3DDTd4fn6+z5492ysrK33jxo0+b948f/PNN33btm2+aNEi32+//XzKlCnu7l5RUeHFxcX+61//2jdt2uQVFRX+1ltvJZY1atSoausbOnSo33fffYk6lpaW+sKFC33dunV+6qmn+ujRo6ttlwsvvNA3btzo8+fP9z322MM/+uijWutx6aWX+tChQ33p0qW+fft2f+ONN3zz5s3+8ccfe5s2bfyFF17wrVu3+q233uqlpaW+ZcsWd3cvKSnx/v37+xdffJHY9jXLKisr/ZBDDvEbb7zRt2zZ4gsXLvTevXv7c8895+7ut912m++///5eVlbmO3bs8Pnz5/uqVavc3R3wTz/9tFqs7dq187lz59Zaj5NOOsknT55craxt27Y+b968Wue/5pprfPz48b5161YvKyvzbt26+TvvvJOYvnjxYm/Xrp2bmefn5/sDDzxQ7f2/+tWvvG3btg547969fcmSJTutY8OGDV5QUOAvv/xyrTG41/2d2V2UjH+63kcUwDyvLw/WNzGbH01JuHPnzvX8/HxfuXKlu7vvu+++fscddySmX3fddX7eeee5e5CE2rRp4+Xl5e7uvt9++/mLL76YmHfZsmWen5+fSGCAL1y4sM54vvnmGwd8zZo1vn37ds/Pz/eysrJq665KuI899pgPGjSo2vvHjRvnEydOrHXZtSXcwYMH1xmLe/AP5JRTTnF39xkzZvhBBx1U63wNJdxhw4b5XXfdlZhWVla203ZJTgADBw70Rx99dKf1VFZWeqtWrXz+/Pk7Tbvpppv8zDPPrDZv165dE8mjpKTEp06dWu09Ncveeust79GjR7V5Jk2a5Oeee667u/fp08efeOKJWrdBbQm3PsOGDfN77rmnWllyvDW98cYbXlpa6nl5eQ74hAkTap1v9erVPnnyZH/zzTd3mrZjxw7/+9//7hMmTPCKioqdpk+fPt179erlO3bsqDNuJdz0J9xsaMONzbRp0zj22GPp1KkTELSdJTcrjBw5klmzZrFlyxZmzZrFIYccQklJCRC0k5566qm0b9+e9u3b893vfpe8vDxWrFiReH+PHj0SzysrK/n5z39OaWkpRUVFVHWWvmrVKlauXMn27durzZ/8fPHixbz99tuJdbVv355HHnmE5cuXR65r8vIAPvnkE0444QSKi4spKiri2muvZdWqVQAsWbKE0tLSyMtOtmzZssQ2AigpKWH79u3VtktxcXHieZs2bVi/fv1Oy1m1ahWbN2+uNY6a62jRogU9evSodvKoZn1rli1evJhly5ZV26aTJk1KxNmcbVBTQUEBFRUV1coqKiooLNzpwhu+/vprjjvuOCZMmMDmzZtZsmQJzz//PHffffdO83bo0IGxY8dy8skn79QubWYcfPDBtG7dmhtuuGGn906bNo0xY8bU2c4v8dhtEu6mTZt4/PHHefXVVykuLqa4uJgpU6bw/vvv8/77wWXAffv2paSkhGeffZYZM2YwcuTIxPt79OjBs88+y5o1axKPzZs3061bt8Q8yTvzjBkzePLJJ3nxxRdZu3Yt5eXlQPCLonPnzuTn57N06dLE/EuWLKm2rqFDh1Zb1/r167nnnnsi17fmF+uSSy5hv/3249NPP6WiooJJkyZVNfXQo0cPFi5cGGk5NXXt2pXFi//vLuwvvviC/Px8unTpEjlWgE6dOtGqVata46i5DndnyZIldW772sp69OhB7969q23TdevW8ec//zkxva5t0Fj9+vVL7FMAn3/+OVu2bKFPnz47zfv555+Tl5fHmDFjyM/Pp3v37px99tmJuGravn07//znP3dK6MnTa9ZjyZIlvPLKK4wZM6YZtZJU2G0S7hNPPEFeXh4fffQR8+fPZ/78+fzjH/9g8ODBTJ8+PTHfyJEj+e1vf8trr73GmWeemSi/+OKLue666xJf/JUrV/Lkk0/Wub5169bRsmVLOnbsyMaNG7n22msT0/Ly8jjttNOYOHEiGzdupKysrFoMJ5xwAp988gkPPfQQ27ZtY9u2bbz77rvVThQ11rp16ygqKqKgoICysrJqyfuEE05g+fLl/OY3v2HLli2sW7eOt99+GwhO2pSXl7Njx45alztixAimTJnCokWLWL9+Pddeey3Dhw8nP79x99S0aNGC888/nyuvvJJly5ZRWVnJm2++yZYtWzjrrLN45pln+Otf/8q2bdu4/fbbadmyJUceeWTk5R922GEUFRVx6623smnTJiorK/nwww959913Abjwwgu5/vrr+fTTT3F3PvjgA1avXp3YBp9//nnkdY0aNYo5c+Ywd+5cNmzYwIQJEzjttNNqPcLt06cP7s6MGTPYsWMHy5cvZ+bMmYkTlLNmzeLjjz9mx44drFy5kiuvvJKDDz6YDh06sGPHDv7whz/wzTff4O6888473HXXXRx99NHV1vHQQw9x5JFHpuwIXpput0m406ZN47zzzqNnz56JI9zi4mIuu+wyHnnkkcRPtBEjRvDKK68wbNiwRNMDwBVXXMFJJ53EscceS2FhIUcccUQiKdVmzJgxlJSU0K1bN/r27csRR1TvGuLOO+9k7dq1FBcXc8455zBixAhatmwJQGFhIS+88AKPPfYYXbt2pbi4mPHjxzfruuFf//rXzJgxg8LCQi666CKGDx+emFZYWMhf/vIX5syZQ3FxMfvssw8vv/wyQOKfTseOHTnkkEN2Wu7555/POeecw5AhQ+jduzetWrXid7/7XZNjPOCAAxg4cCAdOnRg/Pjx7Nixg3333ZeHH36Yyy+/nE6dOjFnzhzmzJnDHnvsEXnZeXl5zJkzh/nz59O7d286derEhRdemLii4sorr+Sss87i2GOPpaioiAsuuCBxhcTEiRMZO3Ys7du3T1wtUlBQwNy5c2tdV79+/fj973/PqFGj2GuvvVi3bl21JoKLL76Yiy++GICioiJmzZrFlClT2HPPPTnooIPYf//9ue666wD48ssvOe644ygsLOSAAw6gRYsWzJ49O7Gs2bNnJ66sGD16NJdffjmXX355tXimT5++W103nc2s6mflrmbAgAFe35hmc+bM2aU6rxk/fjzLly/XHVySMbvadybVUnFrr5m95+4D6pq+2xzhZpuysjI++OCDxE/BqVOncuqpp2Y6LBFJo2zovGa3tG7dOkaMGMGyZcvYa6+9uOqqqzj55JMzHZaIpJESboYMHDiw0ffni8iuTU0KIiIxUcIVEYmJEq6ISExytg23devWGqNJpBFat27d8EzSLDmbcDX6qIhkGzUpiIjERAlXRCQmSrgiIjFRwhURiYkSrohITJRwRURiooQrIhITJVwRkZgo4YqIxEQJV0QkJkq4IiIxUcIVEYmJEq6ISEyUcEVEYqKEKyISEyVcEZGYKOGKiMRECVdEJCZKuCIiMVHCFRGJSdYkXDPrZWZ/NrNvzGy5md1pZjk7yKWI7H6yJuECdwP/BL4NHAQMBS7NaEQiIimUTQm3N/C4u2929+XAc0C/DMckIpIy2ZRw/xM428zamFk34McESTfBzMaZ2Twzm7dy5cqMBCki0lTZlHBfJTiirQCWAvOAJ5JncPd73X2Auw/o3LlzBkIUEWm6rEi4ZtYCeB6YBbQFOgF7ArdmMi4RkVTKioQLdAB6AHe6+xZ3Xw08AByf2bBERFInKxKuu68CFgGXmFm+mbUHxgLvZzYyEZHUiZRwzewHZtY7fP5tM5tmZvebWXEKYzkNOA5YCXwGbAf+LYXLFxHJqKg3FtwN/Ch8fnv4dztwL3BSKgJx9/nAUalYlohINoqacLu5+xfhnV8/AkqArcCytEUmIpJjoibcCjPrAuwPfOTu681sD+Bb6QtNRCS3RE24vwPeBfYA/jUs+z5Qlo6gRERyUaSE6+63mtlsoNLdF4bFXwIXpi0yEZEc05jLwhYBXc1sePj6S+Dz1IckIpKbol4WdgDwCXAfMDUsHgrcn6a4RERyTtQj3HuACe6+H7AtLHsVGJSWqEREclDUhNsPeDh87gDuvgFonY6gRERyUdSEWw4cmlxgZocR3BEmIiIRRL0s7HrgGTP7PbCHmV0DXAxclLbIRERyTKQjXHd/mqBD8M4EbbclwGnu/kIaYxMRySmRB2l097+jMcZERJqszoRrZjdFWYC7T0hdOCIiuau+I9wesUUhIrIbqDPhuvt5cQYiIpLrot5p9pKZ7dRvgpk9k/qQRERyU9STZkcCxWZ2EHCFu1eG5YPTE5aISO6JeuPDVuBwoBfwopl1CMstHUGJiOSiyL2Fufs64ETgLWCemR1IeJuviIg0LGqTggG4uwPXmNn7wItAq3QFJiKSa6Im3AuSX7j7Y2b2MSkaQFJEZHcQtUlhv7AJIcHd/wfYkvqQRERyU9SEez3wFzM7s0b5tSmOR0QkZ0VNuJuBY4HbzOzmpHJdpSAiElHUhOvu/j5wGDDYzJ40swJ0lYKISGRRE27VVQorgWOAr4B3gG+lKS4RkZwTNeE+WPXE3be7+8XAfxJckysiIhFE7YD88lrK/uDuP0h9SCIiuSnSdbhmNr2uae4+JnXhiIjkrqg3Piys8boYOAN4JLXhiIjkrkgJ191vrFlmZlOBG1IekYhIjorceU0t5gNDUxWIiEiui9qGO6xGURvgbOCjlEckIpKjorbhTq3xegPBEe6IVAZjZmcTNFP0BJYD57r73FSuQ0QkU6K24fZOdyBm9kPgVmA4wU0V3073OkVE4hT1CBczaw/8BOgKLAP+7O7fpDCWG4Gb3L3qZoovU7hsEZGMizqI5DCgHPgZMBC4HFhkZkenIggzywMGAJ3N7DMzW2pmd5pZ6xrzjTOzeWY2b+XKlalYtYhIbKJepXAnMM7dD3f3s9z9COAi4K4UxdGFoF+GMwgGpjwIOBj4RfJM7n6vuw9w9wGdO3dO0apFROIRNeF2Bf5Uo2w2wQ0QqbAp/Ps7d//K3VcBdwDHp2j5IiIZFzXhTgf+pUbZJWF5s4VtwUtRd48iksOinjQ7BLjEzK4mOJnVDdgLeNvMXquayd2HNCOWB4DLzew5YBvwr8DTzVieiEhWiZpw7wsf6XQz0An4hGCEiceBX6Z5nSIisYl6He60dAfi7tuAS8OHiEjOaU5fCiIi0ghKuCIiMVHCFRGJSZ0J18zeSnqufm9FRJqpviPcPmbWKnx+VRzBiIjksvquUngS+MTMyoHWydfbJmvmtbciIruNOhOuu59nZoOAXgQd1tTsE1dERBqh3utw3f114HUz2yOOa3FFRHJZ1Bsf7jezHwDnENzW+yXwsLu/lM7gRERySdT+cC8EZhIMezML+AqYYWYXpTE2EZGcErUvhauBH7r7+1UFZjaToMvGdPexICKSE6Le+NCRnUfo/RjokNpwRERyV9SE+zpwh5m1ATCztsB/AH9LV2AiIrkmasK9GDgQWGtmK4A1QH/gp+kKTEQk10S9SuErYKiZdScctdfdl6Y1MhGRHBN5mHSAMMkq0YqINIF6CxMRiYkSrohITBpMuGbWwsyGmdkecQQkIpKrGky47r4DeNLdt8YQj4hIzorapPCamR2R1khERHJc1KsUFgPPmtmTwBLAqya4+4R0BCYikmuiJtzWwBPh8+5pikVEJKdFvfHhvHQHIiKS6yLf+GBm3wXOALq4+2Vmti/Q0t0/SFt0IiI5JGp/uGcCrxF0Pj4mLC4E7khTXCIiOSfqVQo3EfSHezFQGZa9T9CBjYiIRBA14e5FkGDh/65Q8KTnIiLSgKgJ9z2C8cySnQ28k9pwRERyV9STZj8DXjCzC4C2ZvY80Ac4Nm2RiYjkmKiXhZWZ2X7ACcDTBDc/PO3u69MZnIhILol8WZi7bzSzN4BFBB2QK9mKiDRC1MvCeprZXKAceAYoN7PXzawkncGJiOSSqCfNphGcOGvv7nsBewLvhuUpY2b7mNlmM3s4lcsVEckGUZsUDgWOdfdtAO6+3szGA6tTHM9dBIlcRCTnRD3CfQs4rEbZAODNVAViZmcTjAb811QtU0Qkm9R5hGtmNyW9XAj82cyeIbhCoQdwPDAjFUGYWRHB3WxHAxfUM984YBxAz549U7FqEalFr58/0+A85ZN/EkMkuaW+I9weSY9WwCxgC8FdZ1uA2WF5KtwMTHX3JfXN5O73uvsAdx/QuXPnFK1aRCQedR7hxtUlo5kdBBwDHBzH+kREMqUx3TO2Ab4DFCSXu/vfmhnDUUAv4AszI1x+npn1dfdDmrlsEZGsESnhmtkY4E5gK7ApaZIDzW1MvRd4LOn1vxMk4EuauVwRkawS9Qj3NuB0d/9LqgNw943AxqrXZrYe2OzuK1O9LhGRTIqacLcCr6QxjgR3nxjHekRE4hb1OtzrgTvMrFM6gxERyWVRE+4nwEnACjOrDB87zKyyoTeKiEggapPCQ8B0YCbVT5qJiEhEURNuR2CCu2tIHRGRJorapPAAOw+xIyIijRD1CPcw4DIzuw5YkTzB3YekPCoRkRwUNeHeFz5ERKSJoo5pltKOxkVEdkdRb+09v65p7n5/6sIREcldUZsUap4wKwZKgTcAJVwRkQiiNin8oGZZeNT73ZRHJCKSo6JeFlabB6lndAYREakuahtuzcTcBhhNMAaZiIhEELUNdztB37fJvgQuSm04IiK5K2rC7V3j9QZ3X5XqYEREclnUk2aL0x2IiEiuqzfhmtnL7NyUkMzd/ejUhiQikpsaOsJ9uI7ybsDPCE6eiYhIBPUmXHefmvzazDoC1xCcLJsJ3JS+0EREckuk63DNrMjMbgY+A7oAh7j7OHdfmtboRERySL0J18xam9k1wOcEd5UNcvdz3H1hLNGJiOSQhtpwFwF5BMOkzwO6mFmX5Bnc/aU0xSYiklMaSribCa5SuKSO6Q7sndKIRERyVEMnzXrFFIeISM5rTuc1IiLSCEq4IiIxUcIVEYmJEq6ISEyUcEVEYqKEKyISEyVcEZGYKOGKiMQkKxKumbU0s6lmttjM1pnZ/5jZjzMdl4hIKmVFwiW4420JMBRoB1wPPG5mvTIYk4hISkUd0yyt3H0DMDGp6GkzWwQcCpRnIiYRkVTLliPcasIeyfoACzIdi4hIqmTFEW4yM/sW8Agwzd3LakwbB4wD6NmzZwaiE8msXj9/psF5yif/JIZIGhYl1lSIUt+4YmlIVh3hmlkL4CFgK3BZzenufq+7D3D3AZ07d449PhGR5siaI1wzM2AqwRA+x7v7tgyHJCKSUlmTcIF7CIbxOcbdN2U6GBGRVMuKJgUzKwF+ChwELDez9eFjVIZDExFJmaw4wnX3xYBlOg4RkXTKiiNcEZHdgRKuiEhMlHBFRGKihCsiEhMlXBGRmCjhiojERAlXRCQmSrgiIjFRwhURiYkSrohITJRwRURiooQrIhITJVwRkZgo4YqIxEQJV0QkJlnRH67sflI1GGIqBgdMxXpSNXBjtgx2uCvZlbaZjnBFRGKihCsiEhMlXBGRmCjhiojERAlXRCQmSrgiIjFRwhURiYkSrohITJRwRURiooQrIhITJVwRkZgo4YqIxEQJV0QkJkq4IiIxUcIVEYmJEq6ISEyUcEVEYpI1CdfMOpjZbDPbYGaLzWxkpmMSEUmlbBpi5y5gK9AFOAh4xszed/cFmQ1LRCQ1suII18zaAqcD17v7end/HXgKOCezkYmIpE5WJFygD1Dp7p8klb0P9MtQPCIiKWfunukYMLPBwH+5e3FS2UXAKHc/KqlsHDAufLk/8GGccWZAJ2BVpoNIM9UxN6iOgRJ371zXxGxpw10PFNUoKwLWJRe4+73AvQBmNs/dB8QTXmaojrlBdcwNqahjtjQpfALkm9k+SWX9AZ0wE5GckRUJ1903ALOAm8ysrZl9HzgZeCizkYmIpE5WJNzQpUBr4J/Ao8AlDVwSdm8sUWWW6pgbVMfc0Ow6ZsVJMxGR3UE2HeGKiOQ0JVwRkZhkdcJtTP8KZvZvZrbczNaa2f1m1jLOWJsqah3NbKyZvWdmFWa21MxuM7NsuayvXk3pJ8PMXjIzz8U6mtneZva0ma0zs1VmdlucsTZVI/ZVM7NbzOzL8Pv4ipll/U1MZnaZmc0zsy1m9mAD8zYp32R1wqV6/wqjgHtq++DM7EfAz4GjgV7A3sCN8YXZLJHqCLQB/pXg4uvDCer673EF2UxR6wiAmY0ie64RjyrqvroH8BfgJaAY6A48HGOczRH1czwTOB8YDHQA3mTXuOJoGXALcH99MzUr37h7Vj6AtgQfbp+ksoeAybXMOwOYlPT6aGB5puuQyjrW8t4rgTmZrkOq6wi0I7gu+wjAgfxM1yGVdSS4U3JupmNOcx3HA48nve4HbM50HRpR11uAB+uZ3uR8k81HuI3pX6FfOC15vi5m1jGN8aVCc/qQGMKucWNIY+s4CbgHWJ7uwFKoMXU8Aig3s2fD5oRXzOyAWKJsnsbU8THgO2bWx8y+BYwFnoshxrg0Od9kc8ItANbWKFsLFEaYt+p5bfNmk8bUMcHMzgMGAL9OU1ypFLmOZjYA+D7wuxjiSqXGfI7dgbOB3wJdgWeAJ8OmhmzWmDp+BcwFPgY2ETQx/Ftao4tXk/NNNifcSP0r1DFv1fPa5s0mjakjAGZ2CjAZ+LG77wqdhUSqo5m1AO4GrnD37THFliqN+Rw3Aa+7+7PuvpXgn2ZH4LvpDbHZGlPHG4CBQA+gFUH75ktm1iatEcanyfkmmxNuY/pXWBBOS55vhbuvTmN8qdCoPiTM7DjgPuBEd//fGOJLhah1LCI4ap9pZsuBd8PypWFvctmsMZ/jBwRt07uaxtSxPzDT3Ze6+3Z3fxDYE+ib/jBj0fR8k+kG6gYarx8juM23LQoLLCgAAAn5SURBVMFPzbVAv1rmO46gza8vwQf7EhFOPGXDoxF1HAasBoZkOuZ01BEwgrP2VY+BBImpG7BHpuuQws9xX2AjcAyQR/BTe2GO1fEG4HWCqxlaEAwksAFon+k6NFC/fIIj8l8RnBBsRS0nbZuTbzJeyQY2QAfgifDD+gIYGZb3JDis75k075XACqACeABomen4U1lH4GVge1hW9Xg20/Gn+nNMek8vdpGrFBpbR+A04LNwX32ltqSVjY9G7KutCC4h+yqs49+B4zIdf4T6TQz3ueTHxFTmG/WlICISk2xuwxURySlKuCIiMVHCFRGJiRKuiEhMlHBFRGKihCsiEhMl3CxjZqPM7IUMrPf7Zvapma0Pbx9uzrJ6hsvJa8Yy1pvZ3s2JI9XMrNzMjol5nZeY2Ypwe3RM3i5m9qCZ3RI+P8rMljZxHU1+bzZJ3h7ZSgm3DuGXa1O4g1c9uqZ4Hb1qdrLt7o+4+7GpXE9ENwF3unuBuz/RnAW5+xfhciqbsYwCd/+8OXHs6sKetu4Ajg23x2ptl4CZnWtmr2c6jsZSwq3fieEOXvVYVnOGXWVEgghK2DW6e9yddCG4a2uX/1xq+57k0HcnMiXcRko6Kr3AzL4guI8aM/uvpCE3XkvuCd/MWpvZ7eGwJGvN7HUzaw28Fs6yJjyC/l7N/9xmdqSZvRu+710zOzJp2itmdrOZvREO1/KCmXWqJ/aLzOwzM/vazJ6qOmI3s4UEvdbPCePYabiQ8Ij//5nZBxYMsTLVzLqE/bquM7MXzWzPGtsoP3x9rpl9Hs63KBzRATP7jpm9GtZtlZnNTFqfm9l3wucPmtldZvZMuIy3zaw0ad5jzezjcDl3h8u8sJY6dA1/tXRIKjs4XPe3zKzUgqF9Vodlj5hZ+zq2ZbWfrzV/lofr+pOZrQzr/LOkaYdZMJRLhQXNBXfUsvw+BN0bQrB/VO1nie1SnwbW3zqM/xsz+4ig34r6ltXPzP4S7jcrzOzasLylmf3GzJaFj99U7TtV28PMxlvQGdEDtZWF855gZvPNbI2Z/c3MDkxadw8zmxXWY7WZ3Wlm3wV+D3wv3F/X1BLzh2Z2YtLrb4Wf6UENbbu0yvT9y9n6AMqBY2op70Vwj/V0gk48Wofl5xP0h9kS+A0wP+k9dxHcM9+NoMOSI8P5qpaVnzTvuQTd90Fw7/o3BJ1/5AMjwtcdw+mvEHR80gdoHb6uaySFYcAq4JBw3b8DXmuovjWmv0Vw1NUN+CfBPfIHh8t7CbihxjbKD7dRBbBvOO3bhH0HEHSEch3BP/5WwKCk9TnwnfD5g8DXwGHhMh8BHgundQqXf1o47QpgG3BhHfV4Cbgo6fV/AL8Pn38H+GFYn84E/xB/U9s2CmO6JWnaUcDS8HkL4D1gArAHwT+zz4EfhdPfBM4JnxcAR9QRa237R83tcksT1j+ZoL/aDgRdKH5Y9d5aYigk6BPhqvAzKgQOD6fdFO4Te4Xb62/AzUnxbAduDbdn6zrKDiHYlw4n+G6MDbdzy/D1+8AUgv0osY+Q9D1JijV5e1xN0GNZ1bSTgf/NeF7JdADZ+gg/9PXAmvDxRI0vwd71vLd9OE+7cOffBPSvZb7avlCJHYkg0b5T4z1vAueGz18BfpE07VLguTpimgrclvS6gCAx9Uqqb0MJd1TS6z8B9yS9vryWbVSVcNcApxP+c0p6z3TgXqB7LeurmVj+mDTteKAsfD4GeDNpmgFLqDvhXgi8VGPeWntgA04B/qfGNoiScA8HvqixrGuAB8LnrxH0EdupgX2wtv0jSsJtaP2fk9SZDMGwP3Ul3BHJ26DGtIXA8UmvfwSUJ8WzFWhVYxvVLLuHMEknlX0MDAW+B6yk9h67zqX+hNuVoH/aovD1fwNX17e943ioSaF+p7h7+/BR88z9kqonZpZnZpPNbKGZVRB8MSE4+upE8J95YRPW3xVYXKNsMcERZpXkoWg2EiTSBpfl7usJunvsVsf8tVmR9HxTLa93Wre7bwCGAxcDX4XNAvuFk68mSHrvmNkCMzu/nnXXVc+uJH0WHny76jvj/t8EP0W7EgxT5ARHe5jZXmb2mAWjzVYQDO5YZxNNPUqAruFP5DXhT95rCX4dAFxA8KukzIJmohOasI7mrL/aNmPnfSxZD+red2vun4vDsior3X1zjffULCsBrqoRa49wOT2Axd6EDuk9ON/yBnB62Cz0Y4JfRhmlhNt0yd2sjST4yXIMwVFtr7DcCH7GbwZK2VlDXbUtI9ghk/UEvmxkrDsty8zaEow00JRlNYq7P+/uPyRoTigj6EQdd1/u7he5e1fgp8DdUdona/iKYNgaIBiiO/l1LbGsAV4AziL43B4NkzQE/aA6cKC7FwGjCT7D2mwgGEm5SnHS8yXAoqR/1u3dvdDdjw9j+NTdRxD8FL8V+O/w80iVetdPsM16JM3fs4Fl1bbvws77Z8+wrEpt+3fNsiXAL2vE2sbdHw2n9bTaT65F6eZwGsFneCbBr6C07+sNUcJNjUJgC8ERYxuCgRABcPcdBMMu3xGeyMiz4ORYS4KfSzsI2thq82egj5mNNLN8MxtO0Onx002IcQZwnpkdFK57EvC2u5c3YVmRWXBi7aQwoWwhaKapDKedaWZVyfEbgi9RYy8lewY4wMxOCb+Y/0L15FebGQRNEaeHz6sUhvGtMbNuwP+rZxnzgePNrIOZFRMMYV/lHaAiPDnUOvzM9zezgQBmNtrMOof7RtUJnyZfQleLetcPPA5cY2Z7htv/8nqW9TRQbGb/Gp4kKzSzw8NpjwK/MLPOFpysnUDjh3y/D7jYzA63QFsz+4mZFYb1+AqYHJa3MrPvh+9bAXS3+seCe4KgjfgKguarjFPCTY3pBD+nvgQ+IjiRkOzfgf8lGDbma4KjmhbuvhH4JfBG+HPqiOQ3eTBkxwkEJyxWE/wEP8GbMJaZu/8VuJ6g7fUrgqOWsxu7nCZoQRD/MoK6DyVoa4bg7PjbZrYeeIpgPLNFjVl4uC3OBG4j2EZ9gXkEyb0uTwH7EAyLkjz66o0EX9C1BIl8Vj3LeIjghE45wRFz4goLD64/PhE4CFhE8CvnjwS/fiAYMWBBWO//BM6u5ad3k0VY/40E++uiMPaH6lnWOoITiScSNOt8CvwgnHwLwbb+gGD//ntY1phY5wEXAXcS/NP9jKB9Nrke3yHo8HwpQfMUBCc/FwDLzazW74O7byLY33tT/2cZG3VALjnFgsEolxKc4Hs50/FIZpnZBKCPu4/OdCygI1zJAWb2IzNrHzaVXEvQ7lrzV4bsZiy43voCgithsoISruSC7xGcSV9F8BP0lPDnpOymzOwigpNuz7r7aw3NHxc1KYiIxERHuCIiMVHCFRGJiRKuiEhMlHBFRGKihCsiEhMlXBGRmPx/+73kEESLc10AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 360x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "xlim = (0, 1.03)\n",
    "textstr = \"\"\"\\\n",
    "Average fraction correct: {:.4f}\n",
    "\"\"\".format(\n",
    "    (oneshot_results_df[\"num_correct\"] / oneshot_results_df[\"num_total\"]).mean()\n",
    ").strip()\n",
    "props = dict(facecolor=\"white\", alpha=0.3)\n",
    "\n",
    "fg, ax = plt.subplots(figsize=(5, 4))\n",
    "data = oneshot_results_df[\"fraction_correct\"]\n",
    "assert data.min() >= xlim[0] and data.max() <= xlim[1]\n",
    "ax.hist(data, bins=20)\n",
    "ax.set_ylabel(\"Number of puzzles\")\n",
    "ax.set_xlabel(\"Fraction of missing values filled correctly\")\n",
    "ax.set_xlim(*xlim)\n",
    "#     ax.set_ylim(0, 5)\n",
    "# ax.set_title(\"Test - Oneshot\")\n",
    "ax.text(0.04, 0.95, textstr, transform=ax.transAxes, fontsize=\"medium\", verticalalignment=\"top\", bbox=props)\n",
    "fg.tight_layout()\n",
    "fg.savefig(NOTEBOOK_PATH / f\"{UNIQUE_ID}-test-oneshot.svg\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Incremental"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "@torch.no_grad()\n",
    "def design_sudoku(net, x, edge_index, edge_attr, log_conf_lst=None):\n",
    "    mask = (x == 9)\n",
    "    if not mask.any():\n",
    "        return x\n",
    "    \n",
    "    index_array = torch.arange(x.size(0))\n",
    "    \n",
    "    output = net(x, edge_index, edge_attr)\n",
    "    output = output[mask]\n",
    "    index_array = index_array[mask]\n",
    "\n",
    "    max_pred, max_index = torch.softmax(output, dim=1).max(dim=1)\n",
    "\n",
    "    _, max_residue = max_pred.max(dim=0)\n",
    "    if log_conf_lst is not None:\n",
    "        log_conf_lst.append(torch.log(max_pred[max_residue]).data.cpu())\n",
    "\n",
    "    assert x[index_array[max_residue]] == 9\n",
    "    x[index_array[max_residue]] = max_index[max_residue]\n",
    "    return design_sudoku(net, x, edge_index, edge_attr, log_conf_lst)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/strokach/env/lib/python3.7/site-packages/ipykernel_launcher.py:3: TqdmDeprecationWarning: This function will be removed in tqdm==5.0.0\n",
      "Please use `tqdm.notebook.tqdm` instead of `tqdm.tqdm_notebook`\n",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "247fc4d0c1d641e78aa45979b6279648",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=30), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "0.9744897959183674\n"
     ]
    }
   ],
   "source": [
    "wrong_preds = []\n",
    "results = []\n",
    "for data in tqdm.tqdm_notebook(datasets[\"sudoku_test\"], total=len(datasets[\"sudoku_test\"])):\n",
    "    data = data.to(device)\n",
    "    x_in = data.x.clone()\n",
    "    is_missing = x_in == 9\n",
    "    log_conf_lst = []\n",
    "    predicted = design_sudoku(net, x_in, data.edge_index, edge_attr, log_conf_lst)\n",
    "    num_correct = float((predicted[is_missing] == data.y[is_missing]).sum())\n",
    "    num_total = float(is_missing.sum())\n",
    "    results.append(\n",
    "        {\n",
    "            \"fraction_correct\": num_correct / num_total,\n",
    "            \"num_correct\": num_correct,\n",
    "            \"num_total\": num_total,\n",
    "            \"sum_log_prob\": np.sum(log_conf_lst),\n",
    "        }\n",
    "    )\n",
    "    if num_correct < num_total:\n",
    "        wrong_preds.append(\n",
    "            (\n",
    "                data.x.data.cpu().numpy(),\n",
    "                data.edge_index.data.cpu().numpy(),\n",
    "                data.y.data.cpu().numpy(),\n",
    "                predicted.data.cpu().numpy(),\n",
    "            )\n",
    "        )\n",
    "results_df = pd.DataFrame(results)\n",
    "print(results_df[\"num_correct\"].sum() / results_df[\"num_total\"].sum())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVwAAAEUCAYAAABnHY4tAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3deZwU1bn/8c8Do2wziMAIojAYEjQS0SggcUGCCmpEjYjKIoj7QhaXSOJKFLnqjcs1KvlpUBAFl1wEkRgVUcG44g0YURbZBEZxUIEZQITx+f1RNZ2eZmaoGbprhub7fr36Nd2nqus8p7r6mepTVafM3RERkcyrV9sBiIjsLpRwRURiooQrIhITJVwRkZgo4YqIxEQJV0QkJkq4IiIxiS3hmtkTZva5mW0ws0VmdlHStOPNbIGZbTKz18ysIK64RETiYnFd+GBmnYBP3X2LmR0EvA78AlgBLAEuAqYBtwHHunv3qpbXsmVLb9++fUZjFhGpjg8++GCtu+dXNj0nrkDcfX7yy/DRATgCmO/uzwKY2UhgrZkd5O4LKlte+/btmTNnTgYjFhGpHjNbUdX0WPtwzewhM9sELAA+B/4OdALmlc3j7hsJ9ng7xRmbiEimxZpw3f0KIA84FpgMbAFygfUps64P5yvHzC4xszlmNqeoqCjT4YqIpFXsZym4e6m7vwnsD1wOlABNU2ZrChRX8N6H3b2Lu3fJz6+0m0REpE6qzdPCcgj6cOcDh5YVmlmTpHIRkawRS8I1s33M7FwzyzWz+mbWBxgAzASeA35iZv3MrCFwM/BhVQfMRER2RXGdpeAE3Qd/IUjyK4DfuvtUADPrBzwAPAG8C5y7sxXOmDGDzZs37+xiRHYbjRo14oQTTqjtMLJaLAnX3YuA46qYPgM4KJ11bt68mb59+6ZzkSJZbdq0abUdQtbTpb0iIjGJ7cIHEZG6qP3vp0eab/kdv9jpurSHmwXWrFlDjx49yMvL45prrsl4fSeffDLjx4/PeD0i2Wa3TLg9e/Zk7733ZsuWLbUdSlo8/PDDtGzZkg0bNnD33XenddkjR45k8ODB5cpefPFFhg4dmtZ66jIz49NPP63WeyZOnEhBQQFNmjThjDPO4Ouvv6503rfeeotu3bqRl5dH586defPNNxPTRo8eTW5ubuLRqFEj6tWrx9q1axPzzJgxg8MPP5wmTZrQtm1bnnnmGQAWLVrE6aefTn5+Ps2bN6dPnz4sXLiwmq2XdNrtEu7y5cuZPXs2Zsbzzz+fkTq2bduWkeVWZsWKFRx88MGYWZ2IpzZV1Na42z9//nwuvfRSJkyYwJo1a2jcuDFXXHFFhfN+/fXXnHbaafzud79j3bp1XHfddfTt25dvvvkGgOuvv56SkpLEY8SIEfTs2ZOWLVsC8PHHHzNw4EBuv/121q9fz9y5czniiCMAWLduHaeddhoLFy5kzZo1dOvWjdNPPz2elSAV2u0S7uOPP0737t05//zzy/0sfuedd2jdujWlpaWJsueee47OnTsD8P3333PHHXfQoUMHWrRowdlnn53Ya1m+fDlmxtixY2nXrh29evUCoH///rRu3Zq99tqLHj16MH/+f67l+Oqrr+jbty9Nmzala9eu3HjjjRxzzDGJ6QsWLODEE0+kefPmHHjggYm9llRl7bjrrrvIzc1lxowZjBw5krPOOovBgwfTtGlTxo0bx3vvvcfPfvYzmjVrxr777svw4cP57rvvEsuZP39+or5WrVoxevRo/vGPfzB69GiefvppcnNzOfTQ4PqUnj178te//jWxXkaNGkVBQQH77LMPQ4YMYf369eXWy/jx42nXrh0tW7bk9ttvr/Sz2bx5M9dccw0FBQXstddeHHPMMYlT+55//nk6depEs2bN6NmzJ5988knife3bt+fOO++kc+fONGnShG3btlVYVlhYSL9+/cjPz+eAAw7g/vvvTyyjtLSU0aNH06FDB/Ly8jjiiCNYuXIlPXr0AODQQw8lNzeXp59+utL4yzz55JP07duXHj16kJuby2233cbkyZMpLt7u4kneeustWrVqRf/+/alfvz6DBw8mPz+fyZMnbzevuzNhwoRyvy5GjRrFpZdeysknn0xOTg4tWrSgQ4cOAHTr1o0LL7yQ5s2bs8cee3DVVVexcOFCvvrqqx22QTJjt0y4gwYNYtCgQbz00kusWbMGgO7du9OkSRNmzpyZmHfixIkMHDgQgPvvv58pU6bwxhtvUFhYyN57782VV15ZbtlvvPEGn3zyCS+99BIQ9HUuXryYL7/8ksMPP5xBgwYl5r3yyitp0qQJX3zxBePHjy+X/Ddu3MiJJ57IwIED+fLLL5k0aRJXXHFFuYRdZty4cQwaNIjrrruOkpKSxHmUU6dO5ayzzmLdunUMGjSI+vXrc++997J27VrefvttXn31VR566CEAiouLOeGEEzjppJMoLCzk008/5fjjj+ekk07i+uuv55xzzqGkpIR58+ZVWP+4ceN47bXXWLp0KSUlJQwfPrzcPG+++SYLFy7k1Vdf5dZbby2XLJNde+21fPDBB7z11lt8/fXX3HXXXdSrV49FixYxYMAA7rvvPoqKijjllFPo27dvuX8YkyZNYvr06axbt46cnJztyurVq0ffvn059NBDWb16Na+++ir33Xdf4rO65557mDRpEn//+9/ZsGEDjz76KI0bN2bWrFkAzJs3j5KSEs455xwAmjVrVu6nf7L58+cn/jkBdOjQgT333JNFixZtN6+7kzpEqrvz0UcfbTfv7NmzWbNmDf369UuUvfPOOwAccsgh7LvvvgwePLjS7otZs2bRunVrWrRoUeF0iUHZB76rPY444givyvPPP79d2ezZsz0nJ8eLiorc3f3AAw/0e+65JzH9hhtu8GHDhrm7+4YNG7xx48a+fPlyd3c/6KCDfMaMGYl5CwsLPScnx7du3erLli1zwJcsWVJpPN98840Dvm7dOt+2bZvn5OT4ggULytV99NFHu7v7U0895cccc0y5919yySU+cuTICpc9dOhQv+GGGxKvb7nlFj/22GMrjcXd/d577/UzzjjD3d0nTpzohx12WIXz3XLLLT5o0KByZccdd5w/8sgj7u7eq1cvf/DBBxPTFixYsN16WblyZWJ6165dfdKkSdvVU1pa6g0bNvS5c+duN+3WW2/1/v37l5u3TZs2/tprr7m7e0FBgY8dO7bce1LL3nnnHW/btm25eUaPHu3nn3++u7t37NjRp0yZUuE6AHzx4sUVTqtIr169fMyYMeXKkuNNtnbtWt9rr7184sSJ/t133/m4cePczPySSy7Zbt4LLrjAhw4dWq5sjz328IKCAl+4cKEXFxf7mWee6QMHDtzuvStXrvQ2bdr4xIkTK427ou/M7qBgxAuRHlEAc7yKvLVbnRY2fvx4evfunej/GjhwIOPHj+eqq65KvD7qqKMYM2YMkydP5vDDD6egILj5xIoVK/jlL39JvXr/+VFQv379xB4yQNu2bRPPS0tLueGGG3j22WcpKipKvG/t2rVs3ryZbdu2lZs/+fmKFSt49913adasWaJs27ZtnHfeeZHbmrw8CA6gXH311cyZM4dNmzaxbdu2RF/fypUrEz9Dq6uwsDCxjgAKCgrYtm1bufXSunXrxPPGjRtTUlKy3XLWrl3Lt99+W2EcqXXUq1ePtm3bsnr16kRZantTy1asWEFhYWG5dVpaWsqxxx4L7Nw6SJWbm8uGDRvKlW3YsIG8vO0GwKNFixZMnTqVa6+9liuvvJI+ffpwwgknsP/++5ebb/PmzTz77LNMnTq1XHmjRo0YNmwYHTt2BII+39SrxYqKiujduzdXXHEFAwYMSEcTpYZ2m4S7efNmnnnmGUpLSxMJYMuWLaxbt4558+Zx6KGHcvDBB1NQUMCLL75YrjsBgi/vo48+ytFHH73dspcvXw5Q7qDVxIkTmTp1KjNmzKB9+/asX7+evffeG3cnPz+fnJwcVq1alfiirFy5slxdxx13HK+88kqN25t6AO3yyy/npz/9KZMmTSIvL4/77ruPv/3tb4n6Jk2aFGk5qdq0acOKFf8Zc/mzzz4jJyeHVq1asWrVqsjxtmzZkoYNG7JkyZJyP8fL6vj3v/+deO3urFy5kv3226/KOJPL2rZtywEHHMDixYsrrL9t27YsWbKEn/zkJ5FjrkynTp3Kdb8sXbqULVu2JD7rVMcddxzvv/8+EPxj7dChw3an902ePJnmzZvTs2fPcuWdO3eu8jP65ptv6N27N6eddho33HBDDVsk6bLb9OFOmTKF+vXr8/HHHzN37lzmzp3LJ598wrHHHsvjjz+emG/gwIHcf//9zJo1i/79+yfKL7vsMm644YZEcikqKtpubyNZcXExDRo0oEWLFmzatInrr78+Ma1+/fqceeaZjBw5kk2bNrFgwYJyMZx66qksWrSICRMmsHXrVrZu3cr7779fad9nFMXFxTRt2pTc3FwWLFjAmDFjytX3xRdfcN9997FlyxaKi4t59913AWjVqhXLly/n+++/r3C5AwYM4N5772XZsmWUlJQk+nzL+lGjqlevHhdccAFXX301hYWFlJaW8vbbb7NlyxbOPvtspk+fzquvvsrWrVu5++67adCgAUcddVTk5Xfr1o2mTZty5513snnzZkpLS/noo48Sie6iiy7ipptuYvHixbg7H374YeLgUqtWrVi6dGnkugYNGsS0adOYPXs2Gzdu5Oabb+bMM8+scA8X4F//+hdbt25lw4YNXHvttey///706dOn3Dzjx49nyJAh2yXXYcOG8dhjj7F06VI2bdrEnXfeyamnngoEe9V9+vTh6KOP5o477ogcv2TObpNwx48fz7Bhw2jXrh2tW7dOPIYPH86TTz6ZOHVowIABvP766/Tq1SvR9QDwm9/8htNOO43evXuTl5dH9+7dE0mpIkOGDKGgoID99tuPgw8+mO7dy9+i7YEHHmD9+vW0bt2a8847jwEDBtCgQQMA8vLyePnll3nqqado06YNrVu3ZsSIETt13vCf/vQnJk6cSF5eHhdffHHi4E9Zfa+88grTpk2jdevW/OhHP+K1114DSPzTadGiBYcffvh2y73gggs477zz6NGjBwcccAANGzbkz3/+c41jPOSQQ+jatSvNmzdnxIgRfP/99xx44IE88cQT/OpXv6Jly5ZMmzaNadOmseeee0Zedv369Zk2bRpz587lgAMOoGXLllx00UWJMyquvvpqzj77bHr37k3Tpk258MILE2dIjBw5kqFDh9KsWbPE2SK5ubnMnj27wro6derEX/7yFwYNGsQ+++xDcXFx4gAlBP+8L7vsssTru+66i5YtW9K2bVs+//xznnvuuXLLW716NTNnzmTIkCHb1XXBBRcwZMgQjjzySAoKCmjQoEHi7IvnnnuO999/n8cee6zcubyfffZZ5PUm6RXbTSTTrUuXLl7VPc2mTZu2Sw1eM2LEiMQZCyK1YVf7zqRLOi/tNbMP3L1LZdN3mz3cumbBggV8+OGHuDvvvfceY8eO5Ze//GVthyUiGbTbHDSra4qLixkwYACFhYXss88+XHPNNboKSCTLKeHWkq5du1b7+nwR2bWpS0FEJCZKuCIiMVHCFRGJSdb24TZq1Ej3aBKphkaNGtV2CFkvaxOu7j4qInWNuhRERGKihCsiEhMlXBGRmCjhiojERAlXRCQmSrgiIjGJJeGaWQMzG2tmK8ys2Mz+ZWYnh9Pam5mbWUnS46Y44hIRiVNc5+HmACuB44DPgFOAZ8zskKR5mrn7tpjiERGJXSx7uO6+0d1Huvtyd//e3V8AlgFHxFG/iEhdUCt9uGbWCugIzE8qXmFmq8zsMTNrWcn7LjGzOWY2p6ioKJZYRUTSJfaEa2Z7AE8C4919AbAW6AoUEOzx5oXTt+PuD7t7F3fvkp+fH1fIIiJpEetYCmZWD5gAfAcMB3D3EqDs5mRrzGw48LmZNXX3DXHGJyKSSbElXAvu7zwWaAWc4u5bK5m17K6WVsl0EZFdUpx7uGOAHwMnuPvmskIzOxJYBywG9gbuB1539/UxxiYiknFxnYdbAFwKHAZ8kXS+7SDgB8A/gGLgI2ALMCCOuERE4hTLHq67r6DqLoJJccQhIlKbdGmviEhMlHBFRGKihCsiEhMlXBGRmCjhiojERAlXRCQmSrgiIjFRwhURiYkSrohITJRwRURiooQrIhKTSAnXzH5uZgeEz/c1s/Fm9qiZtc5seCIi2SPqHu5DQGn4/G5gD4Jxax/ORFAiItko6mhh+7n7Z2aWA/QhuB3Od0BhxiITEckyURPuhvDGjz8BPnb3EjPbk2BPV0REIoiacP8MvA/sCfw2LDsaWJCJoEREslGkhOvud5rZc0Cpuy8Ji1cDF2UsMhGRLFOd08KWAW3M7Jzw9WpgafpDEhHJTlFPCzsEWAQ8QnDnXYDjgEczFJeISNaJuoc7BrjZ3Q8Cym5v/gZwTEaiEhHJQlETbifgifC5A7j7RqBRJoISEclGURPucuCI5AIz6wZ8mu6ARESyVdTTwm4CppvZX4A9zewPwGXAxRmLTEQky0Taw3X3F4CTgXyCvtsC4Ex3fzmDsYmIZJWoe7i4+/8BV2QwFhGRrFZpwjWzW6MswN1vTl84IiLZq6o93LaxRSEishuoNOG6+7B0VWJmDQiGeDwBaE5wdsP17v5iOP144EGgHfAucL67r0hX/SIidUHUK81mmtl24yaY2fSI9eQAKwmuTtuL4KyHZ8ysvZm1BCaHZc2BOcDTEZcrIrLLiHrQ7CigtZkdBvzG3csGIz82ypvDiyRGJhW9YGbLCM7tbQHMd/dnAcxsJLDWzA5yd41GJiJZI+qFD98BRwLtgRlm1jwst5pUGo6t2xGYT3AV27yyaWFyXhKWp77vEjObY2ZzioqKalK1iEitiTxamLsXA32Bd4A5ZtaZ8DLf6jCzPYAngfHhHmwusD5ltvVAXgUxPOzuXdy9S35+fnWrFhGpVVG7FAzA3R34g5nNA2YADatTmZnVAyYQ7DEPD4tLgKYpszYFiquzbBGRui5qwr0w+YW7P2VmC4HTolZkZkYwtGMr4BR3Lxt1bD4wNGm+JkCHsFxEJGtE7VI4KOxCSHD3fwFbqlHXGODHQF9335xU/hzwEzPrZ2YNgZuBD3XATESyTdSEexPwipn1Tym/PsqbzawAuBQ4DPjCzErCxyB3LwL6AbcD3xAcnDs3YlwiIruMqF0K3wK9gSlm1tndbwrLI52lEF7EUOm87j4DOChiLCIiu6Soe7ju7vOAbsCxZjbVzHKpwVkKIiK7q6gJt+wshSKCy3M/B94D9shQXCIiWSdqwh1X9sTdt7n7ZcD/EJyTKyIiEUQdgPxXFZT9P3f/efpDEhHJTpEOmpnZ45VNc/ch6QtHRCR7RT1LYUnK69bAWQSX6IqISASREq67/zG1zMzGArekPSIRkSwVefCaCswlGN9WREQiiNqH2yulqDHB1WAfpz0iEZEsFbUPd2zK640Ee7gD0huOiEj2itqHe0CmAxERyXZR93Axs2bAL4A2QCHwd3f/JlOBiYhkm6g3kewFLAd+DXQFfgUsC++2KyIiEUTdw30AuMTdnykrCIdqfBCN8iUiEknU08LaAP+bUvYcwQUQIiISQdSE+zhwZUrZ5WG5iIhEELVL4XDgcjO7DlgN7AfsA7xrZrPKZnL3HukPUUQkO0RNuI+EDxERqaGo5+GOz3QgIiLZbmfGUhARkWpQwhURiYkSrohITCpNuGb2TtJzjXsrIrKTqtrD7WhmDcPn18QRjIhINqvqLIWpwCIzWw40Sj7fNpnOvRURiabShOvuw8zsGKA9wYA1qWPiiohINVR5Hq67vwm8aWZ77uy5uGY2HDgfOASY5O7nh+XtgWUEg5qXudPdb9uZ+kRE6pqoFz48amY/B84juKx3NfCEu8+sRl2FwCigD9CogunN3H1bNZYnIrJLiToe7kXA08AXwGTgc2CimV0ctSJ3n+zuU4CvahKoiMiuLupYCtcBJ7r7vLICM3uaYMjGdI2xsMLMHHgF+J27r03TckVE6oSoFz60YPs79C4EmqchhrUEB+UKgCOAPODJimY0s0vMbI6ZzSkqKkpD1SIi8YmacN8E7jGzxgBm1gT4b+CtnQ3A3UvcfY67b3P3NcBwoLeZNa1g3ofdvYu7d8nPz9/ZqkVEYhU14V4GdAbWm9kaYB1wKHBpBmLy8K9lYNkiIrUm6lkKnwPHmdn+hHftdfdV1anIzHLC+uoD9cOr2LYRdCOsAxYDewP3A6+7+/rqLF9EpK6r1uA17r7K3d+rbrIN3QhsBn4PDA6f3wj8APgHUAx8BGwBBtRg+SIidVrUsxR2mruPBEZWMnlSXHGIiNQWDc8oIhKTHSZcM6tnZr3MbM84AhIRyVY7TLju/j0w1d2/iyEeEZGsFbVLYZaZdc9oJCIiWS7qQbMVwItmNhVYyX/OlcXdb85EYCIi2SZqwm0ETAmf75+hWEREslrUCx+GZToQEZFsF/k8XDP7MXAW0Mrdh5vZgUADd/8wY9GJiGSRqOPh9gdmEQw+PiQszgPuyVBcIiJZJ+pZCrcSjId7GVAals0jGMBGREQiiJpw9yFIsPCfMxQ86bmIiOxA1IT7AcH9zJKdC7yX3nBERLJX1INmvwZeNrMLgSZm9hLQEeidschERLJM1NPCFpjZQcCpwAsEFz+84O4lmQxORCSbRD4tzN03mdk/gWUEA5Ar2YqIVEPU08LamdlsYDkwHVhuZm+aWUEmgxMRySZRD5qNJzhw1szd9yG4Fc77YbmIiEQQtUvhCKC3u2+F4E67ZjYC+CpjkYmIZJmoe7jvAN1SyroAb6c3HBGR7FXpHq6Z3Zr0cgnwdzObTnCGQlvgFGBiZsMTEckeVXUptE15PTn8uw/BnXWfAxpmIigRkWxUacLVkIwiIulVneEZGwM/BHKTy939rXQHJSKSjSIlXDMbAjwAfAdsTprkQLsMxCUiknWi7uHeBfRz91cyGYyISDaLelrYd8DrGYxDRCTrRU24NwH3mFnLTAYjIpLNoibcRcBpwBozKw0f35tZ6Y7eWMbMhpvZHDPbYmbjUqYdb2YLzGyTmb2mMRpEJBtFTbgTgMcJbqnTMXz8KPwbVSEwCng0uTDca55MsBfdHJgDPF2N5YqI7BKiHjRrAdzs7jW+pY67TwYwsy7A/kmTzgTmu/uz4fSRwFozO8jdF9S0PhGRuibqHu5jbH+LnXTpxH/ul4a7byS4lLhT6oxmdknYLTGnqKgoQ+GIiGRG1ITbDfirmS00s1nJjzTEkAusTylbT3Ab9nLc/WF37+LuXfLz89NQtYhIfKJ2KTwSPjKhBGiaUtYUKM5QfSIitSLqPc0yOdD4fGBo2QszawJ0CMtFRLJG1Et7L6hsmrs/Wtm0lGXkhPXVB+qbWUNgG8GoY/9tZv0Ibt9zM/ChDpiJSLaJ2qWQesCsNcFe6D9JOc2rCjcCtyS9Hgz80d1Hhsn2AeAJ4F3g3IjLFBHZZUTtUvh5alm41/vjqBW5+0hgZCXTZgAHRV2WiMiuKOpZChUZB1yYpjhERLJe1D7c1MTcmKBLYF3aIxIRyVJR+3C3EYx9m2w1cHF6wxERyV5RE+4BKa83uvvadAcjIpLNoh40W5HpQEREsl2VCdfMXmP7roRk7u7HpzckEZHstKM93CcqKd8P+DXBwTMREYmgyoTr7mOTX5tZC+APBAfLngZuzVxoIiLZJdJ5uGbW1MxuAz4FWgGHu/sl7r4qo9GJiGSRKhOumTUysz8ASwmuKjvG3c9z9yWxRCcikkV21Ie7jGCwmbsIbn3TysxaJc/g7jMzFJuISFbZUcL9luAshcsrme7AD9IakYhIltrRQbP2McUhIpL1dmbwGhERqQYlXBGRmCjhiojERAlXRCQmSrgiIjFRwhURiYkSrohITJRwRURiooQrIhITJVwRkZgo4YqIxEQJV0QkJkq4IiIxUcIVEYlJnUm4Zva6mX1rZiXhY2FtxyQikk51JuGGhrt7bvg4sLaDERFJp7qWcEVEslZdS7j/ZWZrzeyfZtYzdaKZXWJmc8xsTlFRUS2EJyJSc3Up4Y4guD/afsDDwDQz65A8g7s/7O5d3L1Lfn5+bcQoIlJjdSbhuvu77l7s7lvcfTzwT+CU2o5LRCRd6kzCrYADVttBiIikS51IuGbWzMz6mFlDM8sxs0FAD+Cl2o5NRCRdqrxNeoz2AEYBBwGlwALgDHfXubgikjXqRMJ19yKga23HISKSSXWiS0FEZHeghCsiEhMlXBGRmCjhiojERAlXRCQmSrgiIjFRwhURiYkSrohITJRwRURiooQrIhITJVwRkZgo4YqIxEQJV0QkJkq4IiIxUcIVEYmJEq6ISEyUcEVEYqKEKyISEyVcEZGYKOGKiMRECVdEJCZKuCIiMakTt0mX2tP+99Mjzbf8jl9kOBKR7Kc9XBGRmCjhiojERAlXRCQmdSbhmllzM3vOzDaa2QozG1jbMYmIpFNdOmj2IPAd0Ao4DJhuZvPcfX7thiUikh51Yg/XzJoA/YCb3L3E3d8EngfOq93IRETSp04kXKAjUOrui5LK5gGdaikeEZG0qytdCrnA+pSy9UBecoGZXQJcEr7cYmYfxRBbbWoJrK3tIADszowtus60MYPUxixgd0ZqY0FVE+tKwi0BmqaUNQWKkwvc/WHgYQAzm+PuXeIJr3aojdlBbcwO6WhjXelSWATkmNmPksoOBXTATESyRp1IuO6+EZgM3GpmTczsaOB0YELtRiYikj51IuGGrgAaAV8Ck4DLd3BK2MOxRFW71MbsoDZmh51uo7l7OgIREZEdqEt7uCIiWU0JV0QkJnU64VZnfAUzu8rMvjCz9Wb2qJk1iDPWmoraRjMbamYfmNkGM1tlZneZWV05ra9KNRknw8xmmplnYxvN7Adm9oKZFZvZWjO7K85Ya6oa26qZ2SgzWx1+H183szp/EZOZDTezOWa2xczG7WDeGuWbOp1wKT++wiBgTEUfnJn1AX4PHA+0B34A/DG+MHdKpDYCjYHfEpxgfiRBW6+NK8idFLWNAJjZIOrOOQ4Avt4AAAvvSURBVOJRRd1W9wReAWYCrYH9gSdijHNnRP0c+wMXAMcCzYG32TXOOCoERgGPVjXTTuUbd6+TD6AJwYfbMalsAnBHBfNOBEYnvT4e+KK225DONlbw3quBabXdhnS3EdiL4Lzs7oADObXdhnS2keBKydm1HXOG2zgCeCbpdSfg29puQzXaOgoYV8X0GueburyHW53xFTqF05Lna2VmLTIYXzrszBgSPdg1LgypbhtHA2OALzIdWBpVp43dgeVm9mLYnfC6mR0SS5Q7pzptfAr4oZl1NLM9gKHAP2KIMS41zjd1OeFGGl+hknnLnlc0b11SnTYmmNkwoAvwpwzFlU6R22hmXYCjgT/HEFc6Vedz3B84F7gfaANMB6aGXQ11WXXa+DkwG1gIbCboYrgqo9HFq8b5pi4n3EjjK1Qyb9nziuatS6rTRgDM7AzgDuBkd98VBguJ1EYzqwc8BPzG3bfFFFu6VOdz3Ay86e4vuvt3BP80WwA/zmyIO606bbwF6Aq0BRoS9G/ONLPGGY0wPjXON3U54VZnfIX54bTk+da4+1cZjC8dqjWGhJmdBDwC9HX3f8cQXzpEbWNTgr32p83sC+D9sHyVmR2b+TB3SnU+xw8J+qZ3NdVp46HA0+6+yt23ufs4YG/g4MyHGYua55va7qDeQef1UwSX+TYh+Km5HuhUwXwnEfT5HUzwwc4kwoGnuvCoRht7AV8BPWo75ky0ETCCo/Zlj64EiWk/YM/abkMaP8cDgU3ACUB9gp/aS7KsjbcAbxKczVCP4EYCG4Fmtd2GHbQvh2CP/L8IDgg2pIKDtjuTb2q9kTtYAc2BKeGH9RkwMCxvR7Bb3y5p3quBNcAG4DGgQW3Hn842Aq8B28KysseLtR1/uj/HpPe0Zxc5S6G6bQTOBD4Nt9XXK0padfFRjW21IcEpZJ+Hbfw/4KTajj9C+0aG21zyY2Q6843GUhARiUld7sMVEckqSrgiIjFRwhURiYkSrohITJRwRURiooQrIhITJdw6xswGmdnLtVDv0Wa22MxKwsuHd2ZZ7cLl1N+JZZSY2Q92Jo50M7PlZnZCzHVebmZrwvXRInm9mNk4MxsVPu9pZqtqWEeN31uXJK+PukoJtxLhl2tzuIGXPdqkuY72qYNsu/uT7t47nfVEdCvwgLvnuvuUnVmQu38WLqd0J5aR6+5LdyaOXV040tY9QO9wfXyl9RIws/PN7M3ajqO6lHCr1jfcwMsehakz7Cp3JIiggF1juMfdSSuCq7Z2+c+lou9JFn13IlPCraakvdILzewzguuoMbNnk265MSt5JHwza2Rmd4e3JVlvZm+aWSNgVjjLunAP+mep/7nN7Cgzez983/tmdlTStNfN7DYz+2d4u5aXzaxlFbFfbGafmtnXZvZ82R67mS0hGLV+WhjHdrcLCff4f2dmH1pwi5WxZtYqHNe12MxmmNneKesoJ3x9vpktDedbFt7RATP7oZm9EbZtrZk9nVSfm9kPw+fjzOxBM5seLuNdM+uQNG9vM1sYLuehcJkXVdCGNuGvluZJZT8N697DzDpYcGufr8KyJ82sWSXrstzP19Sf5WFd/2tmRWGbf500rZsFt3LZYEF3wT0VLL8jwfCGEGwfZdtZYr1UZQf1Nwrj/8bMPiYYt6KqZXUys1fC7WaNmV0fljcws/vMrDB83Fe27ZStDzMbYcFgRI9VVBbOe6qZzTWzdWb2lpl1Tqq7rZlNDtvxlZk9YGY/Bv4C/CzcXtdVEPNHZtY36fUe4Wd62I7WXUbV9vXLdfUBLAdOqKC8PcE11o8TDOLRKCy/gGA8zAbAfcDcpPc8SHDN/H4EA5YcFc5XtqycpHnPJxi+D4Jr178hGPwjBxgQvm4RTn+dYOCTjkCj8HVld1LoBawFDg/r/jMwa0ftTZn+DsFe137AlwTXyP80XN5M4JaUdZQTrqMNwIHhtH0Jxw4gGAjlBoJ//A2BY5Lqc+CH4fNxwNdAt3CZTwJPhdNahss/M5z2G2ArcFEl7ZgJXJz0+r+Bv4TPfwicGLYnn+Af4n0VraMwplFJ03oCq8Ln9YAPgJuBPQn+mS0F+oTT3wbOC5/nAt0ribWi7SN1vYyqQf13EIxX25xgCMWPyt5bQQx5BGMiXBN+RnnAkeG0W8NtYp9wfb0F3JYUzzbgznB9Nqqk7HCCbelIgu/G0HA9NwhfzwPuJdiOEtsISd+TpFiT18d1BCOWlU07Hfh3reeV2g6grj7CD70EWBc+pqR8CX5QxXubhfPsFW78m4FDK5ivoi9UYkMiSLTvpbznbeD88PnrwI1J064A/lFJTGOBu5Je5xIkpvZJ7d1Rwh2U9Pp/gTFJr39VwToqS7jrgH6E/5yS3vM48DCwfwX1pSaWvyZNOwVYED4fArydNM2AlVSecC8CZqbMW+EIbMAZwL9S1kGUhHsk8FnKsv4APBY+n0UwRmzLHWyDFW0fURLujupfStJgMgS3/aks4Q5IXgcp05YApyS97gMsT4rnO6BhyjpKLRtDmKSTyhYCxwE/A4qoeMSu86k64bYhGJ+2afj6b8B1Va3vOB7qUqjaGe7eLHykHrlfWfbEzOqb2R1mtsTMNhB8MSHY+2pJ8J95SQ3qbwOsSClbQbCHWSb5VjSbCBLpDpfl7iUEwz3uV8n8FVmT9HxzBa+3q9vdNwLnAJcBn4fdAgeFk68jSHrvmdl8M7ugirora2cbkj4LD75dVR1x/xvBT9E2BLcpcoK9PcxsHzN7yoK7zW4guLljpV00VSgA2oQ/kdeFP3mvJ/h1AHAhwa+SBRZ0E51agzp2pv5y64ztt7Fkbal8203dPleEZWWK3P3blPeklhUA16TE2jZcTltghddgQHoPjrf8E+gXdgudTPDLqFYp4dZc8jBrAwl+spxAsFfbPiw3gp/x3wId2N6OhmorJNggk7UDVlcz1u2WZWZNCO40UJNlVYu7v+TuJxJ0JywgGEQdd//C3S929zbApcBDUfonU3xOcNsaILhFd/LrCmJZB7wMnE3wuU0KkzQE46A60NndmwKDCT7DimwkuJNymdZJz1cCy5L+WTdz9zx3PyWMYbG7DyD4KX4n8Lfw80iXKusnWGdtk+Zvt4NlVbTtwvbbZ7uwrExF23dq2Urg9pRYG7v7pHBaO6v44FqUYQ7HE3yG/Ql+BWV8W98RJdz0yAO2EOwxNia4ESIA7v49wW2X7wkPZNS34OBYA4KfS98T9LFV5O9ARzMbaGY5ZnYOwaDHL9QgxonAMDM7LKx7NPCuuy+vwbIis+DA2mlhQtlC0E1TGk7rb2ZlyfEbgi9RdU8lmw4cYmZnhF/MKymf/CoykaArol/4vExeGN86M9sP+F0Vy5gLnGJmzc2sNcEt7Mu8B2wIDw41Cj/zn5hZVwAzG2xm+eG2UXbAp8an0FWgyvqBZ4A/mNne4fr/VRXLegFobWa/DQ+S5ZnZkeG0ScCNZpZvwcHam6n+Ld8fAS4zsyMt0MTMfmFmeWE7PgfuCMsbmtnR4fvWAPtb1feCm0LQR/wbgu6rWqeEmx6PE/ycWg18THAgIdm1wL8JbhvzNcFeTT133wTcDvwz/DnVPflNHtyy41SCAxZfEfwEP9VrcC8zd38VuImg7/Vzgr2Wc6u7nBqoRxB/IUHbjyPoa4bg6Pi7ZlYCPE9wP7Nl1Vl4uC76A3cRrKODgTkEyb0yzwM/IrgtSvLdV/9I8AVdT5DIJ1exjAkEB3SWE+wxJ86w8OD8477AYcAygl85fyX49QPBHQPmh+3+H+DcCn5611iE+v9IsL0uC2OfUMWyigkOJPYl6NZZDPw8nDyKYF1/SLB9/19YVp1Y5wAXAw8Q/NP9lKB/NrkdPyQY8HwVQfcUBAc/5wNfmFmF3wd330ywvR9A1Z9lbDQAuWQVC25GuYrgAN9rtR2P1C4zuxno6O6DazsW0B6uZAEz62NmzcKukusJ+l1Tf2XIbsaC860vJDgTpk5QwpVs8DOCI+lrCX6CnhH+nJTdlJldTHDQ7UV3n7Wj+eOiLgURkZhoD1dEJCZKuCIiMVHCFRGJiRKuiEhMlHBFRGKihCsiEpP/DyHINYJ8Mb6VAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 360x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "xlim = (0, 1.03)\n",
    "textstr = \"\"\"\\\n",
    "Average fraction correct: {:.4f}\n",
    "\"\"\".format(\n",
    "    (results_df[\"num_correct\"] / results_df[\"num_total\"]).mean()\n",
    ").strip()\n",
    "props = dict(facecolor=\"white\", alpha=0.3)\n",
    "\n",
    "fg, ax = plt.subplots(figsize=(5, 4))\n",
    "data = results_df[\"fraction_correct\"]\n",
    "assert data.min() >= xlim[0] and data.max() <= xlim[1]\n",
    "ax.hist(data, bins=20)\n",
    "ax.set_ylabel(\"Number of puzzles\")\n",
    "ax.set_xlabel(\"Fraction of missing values filled correctly\")\n",
    "ax.set_xlim(*xlim)\n",
    "#     ax.set_ylim(0, 5)\n",
    "# ax.set_title(\"Test - Incremental\")\n",
    "ax.text(\n",
    "    0.04,\n",
    "    0.95,\n",
    "    textstr,\n",
    "    transform=ax.transAxes,\n",
    "    fontsize=\"medium\",\n",
    "    verticalalignment=\"top\",\n",
    "    bbox=props,\n",
    ")\n",
    "fg.tight_layout()\n",
    "fg.savefig(NOTEBOOK_PATH / f\"{UNIQUE_ID}-test-incremental.svg\", bbox_inches=\"tight\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fraction_correct_oneshot = (oneshot_results_df[\"num_correct\"] / oneshot_results_df[\"num_total\"]).mean()\n",
    "fraction_correct_incremental = (results_df[\"num_correct\"] / results_df[\"num_total\"]).mean()\n",
    "\n",
    "fg, ax = plt.subplots(figsize=(5, 4))\n",
    "\n",
    "data = oneshot_results_df[\"fraction_correct\"]\n",
    "assert data.min() >= xlim[0] and data.max() <= xlim[1]\n",
    "ax.hist(data, bins=30, range=xlim, alpha=1, label=f\"oneshot (accuracy: {fraction_correct_oneshot:.2%})\", color=cm.get_cmap(\"Set1\")(1))\n",
    "\n",
    "data = results_df[\"fraction_correct\"]\n",
    "assert data.min() >= xlim[0] and data.max() <= xlim[1]\n",
    "ax.hist(data, bins=30, range=xlim, alpha=0.6, label=f\"incremental (accuracy: {fraction_correct_incremental:.2%})\", color=cm.get_cmap(\"Set1\")(0))\n",
    "\n",
    "ax.set_ylabel(\"Number of puzzles\")\n",
    "ax.set_xlabel(\"Fraction of missing values filled correctly\")\n",
    "ax.set_xlim(*xlim)\n",
    "ax.legend(loc=\"upper left\")\n",
    "fg.tight_layout()\n",
    "fg.savefig(NOTEBOOK_PATH / f\"{UNIQUE_ID}-test-oneshot-incremental.svg\", bbox_inches=\"tight\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Validation plots"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Oneshot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/strokach/env/lib/python3.7/site-packages/ipykernel_launcher.py:13: TqdmDeprecationWarning: This function will be removed in tqdm==5.0.0\n",
      "Please use `tqdm.notebook.tqdm` instead of `tqdm.tqdm_notebook`\n",
      "  del sys.path[0]\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "527cf1b85d594c619064f4f12784b7cf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=1000), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "results = []\n",
    "\n",
    "for i, data in tqdm.tqdm_notebook(\n",
    "    enumerate(\n",
    "        DataLoader(\n",
    "            datasets[\"sudoku_valid_0\"],\n",
    "            shuffle=False,\n",
    "            num_workers=1,\n",
    "            batch_size=1,\n",
    "            drop_last=False,\n",
    "        )\n",
    "    ),\n",
    "    total=1000,\n",
    "):\n",
    "    if i >= 1000:\n",
    "        break\n",
    "    data = data.to(device)\n",
    "    is_missing = data.x == 9\n",
    "    output = net(data.x, data.edge_index, data.edge_attr)\n",
    "    output = torch.softmax(output, dim=1)\n",
    "    _, predicted = output.max(dim=1)\n",
    "    num_correct = float((predicted[is_missing] == data.y[is_missing]).sum())\n",
    "    num_total = float(is_missing.sum())\n",
    "    results.append(\n",
    "        {\n",
    "            \"fraction_correct\": num_correct / num_total,\n",
    "            \"num_correct\": num_correct,\n",
    "            \"num_total\": num_total,\n",
    "        }\n",
    "    )\n",
    "val_oneshot_results_df = pd.DataFrame(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "xlim = (0.0, 1.03)\n",
    "textstr = \"\"\"\\\n",
    "Fraction correct overall: {:.4f}\n",
    "\"\"\".format(\n",
    "    (val_oneshot_results_df[\"num_correct\"] / val_oneshot_results_df[\"num_total\"]).mean()\n",
    ").strip()\n",
    "props = dict(facecolor=\"white\", alpha=0.3)\n",
    "\n",
    "with plt.rc_context(rc={\"font.size\": 12}):\n",
    "    fg, ax = plt.subplots(figsize=(6, 4))\n",
    "    data = val_oneshot_results_df[\"fraction_correct\"]\n",
    "    assert data.min() >= xlim[0] and data.max() <= xlim[1]\n",
    "    ax.hist(data)\n",
    "    ax.set_ylabel(\"Number of puzzles\")\n",
    "    ax.set_xlabel(\"Fraction of missing values filled correctly\")\n",
    "    ax.set_xlim(*xlim)\n",
    "    #     ax.set_ylim(0, 5)\n",
    "    ax.set_title(\"Validation - Oneshot\")\n",
    "    ax.text(0.04, 0.95, textstr, transform=ax.transAxes, fontsize=\"medium\", verticalalignment=\"top\", bbox=props)\n",
    "fg.savefig(NOTEBOOK_PATH / f\"{UNIQUE_ID}-validation-oneshot.svg\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Incremental"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "@torch.no_grad()\n",
    "def design_sudoku(net, x, edge_index, edge_attr):\n",
    "    mask = (x == 9)\n",
    "    if not mask.any():\n",
    "        return x\n",
    "    output = net(x, edge_index, edge_attr)\n",
    "    output = torch.softmax(output, dim=1)\n",
    "    max_pred, max_index = output.max(dim=1)\n",
    "    max_pred[~mask] = -1\n",
    "    _, max_residue = max_pred.max(dim=0)\n",
    "    x[max_residue] = max_index[max_residue]\n",
    "    return design_sudoku(net, x, edge_index, edge_attr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/strokach/env/lib/python3.7/site-packages/ipykernel_launcher.py:13: TqdmDeprecationWarning: This function will be removed in tqdm==5.0.0\n",
      "Please use `tqdm.notebook.tqdm` instead of `tqdm.tqdm_notebook`\n",
      "  del sys.path[0]\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "099df2a48af74d5db2223eb5c51274f1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=1000), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "results = []\n",
    "\n",
    "for i, data in tqdm.tqdm_notebook(\n",
    "    enumerate(\n",
    "        DataLoader(\n",
    "            datasets[\"sudoku_valid_0\"],\n",
    "            shuffle=False,\n",
    "            num_workers=1,\n",
    "            batch_size=1,\n",
    "            drop_last=False,\n",
    "        )\n",
    "    ),\n",
    "    total=1000,\n",
    "):\n",
    "    if i >= 1000:\n",
    "        break\n",
    "    data = data.to(device)\n",
    "    x_in = data.x.clone()\n",
    "    is_missing = x_in == 9\n",
    "    predicted = design_sudoku(net, x_in, data.edge_index, data.edge_attr)\n",
    "    num_correct = float((predicted[is_missing] == data.y[is_missing]).sum())\n",
    "    num_total = float(is_missing.sum())\n",
    "    results.append({\"fraction_correct\": num_correct / num_total, \"num_correct\": num_correct, \"num_total\": num_total})\n",
    "val_results_df = pd.DataFrame(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "xlim = (0.0, 1.03)\n",
    "textstr = \"\"\"\\\n",
    "Fraction correct overall: {:.4f}\n",
    "\"\"\".format(\n",
    "    (val_results_df[\"num_correct\"] / val_results_df[\"num_total\"]).mean()\n",
    ").strip()\n",
    "props = dict(facecolor=\"white\", alpha=0.3)\n",
    "\n",
    "with plt.rc_context(rc={\"font.size\": 12}):\n",
    "    fg, ax = plt.subplots(figsize=(6, 4))\n",
    "    data = val_results_df[\"fraction_correct\"]\n",
    "    assert data.min() >= xlim[0] and data.max() <= xlim[1]\n",
    "    ax.hist(data, bins=20)\n",
    "    ax.set_ylabel(\"Number of puzzles\")\n",
    "    ax.set_xlabel(\"Fraction of missing values filled correctly\")\n",
    "    ax.set_xlim(*xlim)\n",
    "#     ax.set_ylim(0, 5)\n",
    "    ax.set_title(\"Validation - Incremental\")\n",
    "    ax.text(0.04, 0.95, textstr, transform=ax.transAxes, fontsize=\"medium\", verticalalignment=\"top\", bbox=props)\n",
    "fg.savefig(NOTEBOOK_PATH / f\"{UNIQUE_ID}-validation-incremental.svg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fraction_correct_oneshot = (val_oneshot_results_df[\"num_correct\"] / val_oneshot_results_df[\"num_total\"]).mean()\n",
    "fraction_correct_incremental = (val_results_df[\"num_correct\"] / val_results_df[\"num_total\"]).mean()\n",
    "\n",
    "fg, ax = plt.subplots(figsize=(5, 4))\n",
    "\n",
    "data = val_oneshot_results_df[\"fraction_correct\"]\n",
    "assert data.min() >= xlim[0] and data.max() <= xlim[1]\n",
    "ax.hist(data, bins=30, range=xlim, alpha=1, density=False, label=f\"oneshot (accuracy: {fraction_correct_oneshot:.2%})\", color=cm.get_cmap(\"Set1\")(1))\n",
    "\n",
    "data = val_results_df[\"fraction_correct\"]\n",
    "assert data.min() >= xlim[0] and data.max() <= xlim[1]\n",
    "ax.hist(data, bins=30, range=xlim, alpha=0.6, density=False, label=f\"incremental (accuracy: {fraction_correct_incremental:.2%})\", color=cm.get_cmap(\"Set1\")(0))\n",
    "\n",
    "ax.set_ylabel(\"Number of puzzles\")\n",
    "ax.set_xlabel(\"Fraction of missing values filled correctly\")\n",
    "ax.set_xlim(*xlim)\n",
    "ax.legend(loc=\"upper left\")\n",
    "fg.tight_layout()\n",
    "fg.savefig(NOTEBOOK_PATH / f\"{UNIQUE_ID}-validation-oneshot-incremental.svg\", bbox_inches=\"tight\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [
    "1fXJz1l9HvhW",
    "u9q4iTifHvhl"
   ],
   "name": "Copy of 2019-03-30-sudoku-4xEdgeConv-09862+.ipynb",
   "provenance": [],
   "toc_visible": true,
   "version": "0.3.2"
  },
  "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}