notebooks/sklearn_year_with_train.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### If you haven't installed Hummingbird, do that first, by uncommenting the line below.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#! python -m pip install hummingbird_ml "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Import Hummingbird's convert function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from hummingbird.ml import convert"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Download and unzip the Year data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import zipfile\n",
    "import urllib.request as urllib\n",
    "url = 'https://archive.ics.uci.edu/ml/machine-learning-databases/00203/YearPredictionMSD.txt.zip'\n",
    "\n",
    "filehandle, _ = urllib.urlretrieve(url)\n",
    "zip_file_object = zipfile.ZipFile(filehandle, 'r')\n",
    "filename = zip_file_object.namelist()[0]\n",
    "bytes_data = zip_file_object.open(filename).read()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Convert the bytes to pandas and split it up"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from io import BytesIO\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "year = pd.read_csv(BytesIO(bytes_data), header = None)\n",
    "\n",
    "#train_size = 463715  # Note: this will extend the training time if we do the full dataset\n",
    "train_size = 200000\n",
    "X = year.iloc[:, 1:]\n",
    "y = year.iloc[:, 0]\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, shuffle=False, train_size=train_size, test_size=51630)\n",
    "\n",
    "# Store the test data as numpy by pulling the values out of the pandas dataframe\n",
    "data = X_test.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<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>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>10</th>\n",
       "      <th>...</th>\n",
       "      <th>81</th>\n",
       "      <th>82</th>\n",
       "      <th>83</th>\n",
       "      <th>84</th>\n",
       "      <th>85</th>\n",
       "      <th>86</th>\n",
       "      <th>87</th>\n",
       "      <th>88</th>\n",
       "      <th>89</th>\n",
       "      <th>90</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>200000</th>\n",
       "      <td>45.09540</td>\n",
       "      <td>-57.29138</td>\n",
       "      <td>20.05392</td>\n",
       "      <td>0.10414</td>\n",
       "      <td>12.00346</td>\n",
       "      <td>-14.31259</td>\n",
       "      <td>2.54547</td>\n",
       "      <td>-3.33986</td>\n",
       "      <td>-3.60175</td>\n",
       "      <td>-8.99414</td>\n",
       "      <td>...</td>\n",
       "      <td>7.15503</td>\n",
       "      <td>-81.48297</td>\n",
       "      <td>34.91722</td>\n",
       "      <td>15.11917</td>\n",
       "      <td>-6.09056</td>\n",
       "      <td>-78.52893</td>\n",
       "      <td>-46.87559</td>\n",
       "      <td>2.89571</td>\n",
       "      <td>64.25949</td>\n",
       "      <td>0.78578</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200001</th>\n",
       "      <td>45.11673</td>\n",
       "      <td>-18.39958</td>\n",
       "      <td>-1.61552</td>\n",
       "      <td>-3.67929</td>\n",
       "      <td>-13.24027</td>\n",
       "      <td>-6.84375</td>\n",
       "      <td>-10.33838</td>\n",
       "      <td>-11.12891</td>\n",
       "      <td>16.56924</td>\n",
       "      <td>6.70243</td>\n",
       "      <td>...</td>\n",
       "      <td>8.26246</td>\n",
       "      <td>-138.26547</td>\n",
       "      <td>23.59451</td>\n",
       "      <td>60.99156</td>\n",
       "      <td>4.18891</td>\n",
       "      <td>105.75496</td>\n",
       "      <td>-126.39851</td>\n",
       "      <td>-3.80727</td>\n",
       "      <td>55.00532</td>\n",
       "      <td>-3.42354</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200002</th>\n",
       "      <td>46.85191</td>\n",
       "      <td>9.44824</td>\n",
       "      <td>0.31518</td>\n",
       "      <td>-16.85413</td>\n",
       "      <td>15.42389</td>\n",
       "      <td>-15.82587</td>\n",
       "      <td>-21.21385</td>\n",
       "      <td>-10.16067</td>\n",
       "      <td>14.45113</td>\n",
       "      <td>2.22865</td>\n",
       "      <td>...</td>\n",
       "      <td>13.24783</td>\n",
       "      <td>-91.25475</td>\n",
       "      <td>41.12300</td>\n",
       "      <td>55.22389</td>\n",
       "      <td>8.33048</td>\n",
       "      <td>7.57355</td>\n",
       "      <td>17.34516</td>\n",
       "      <td>0.21543</td>\n",
       "      <td>-58.34520</td>\n",
       "      <td>3.92760</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200003</th>\n",
       "      <td>45.79644</td>\n",
       "      <td>-36.86230</td>\n",
       "      <td>21.99320</td>\n",
       "      <td>-10.42360</td>\n",
       "      <td>-2.89410</td>\n",
       "      <td>-8.84010</td>\n",
       "      <td>-23.73864</td>\n",
       "      <td>-9.82956</td>\n",
       "      <td>13.08399</td>\n",
       "      <td>0.57577</td>\n",
       "      <td>...</td>\n",
       "      <td>8.13403</td>\n",
       "      <td>-194.21155</td>\n",
       "      <td>55.55883</td>\n",
       "      <td>21.65387</td>\n",
       "      <td>6.41164</td>\n",
       "      <td>-47.18867</td>\n",
       "      <td>-212.14270</td>\n",
       "      <td>-4.67550</td>\n",
       "      <td>-86.99988</td>\n",
       "      <td>-6.12034</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200004</th>\n",
       "      <td>40.92442</td>\n",
       "      <td>-43.26026</td>\n",
       "      <td>-18.72100</td>\n",
       "      <td>-11.72495</td>\n",
       "      <td>-19.69395</td>\n",
       "      <td>-10.54229</td>\n",
       "      <td>-9.91945</td>\n",
       "      <td>-14.85633</td>\n",
       "      <td>9.37409</td>\n",
       "      <td>-0.93093</td>\n",
       "      <td>...</td>\n",
       "      <td>17.63167</td>\n",
       "      <td>-203.56276</td>\n",
       "      <td>27.10714</td>\n",
       "      <td>36.90795</td>\n",
       "      <td>0.62431</td>\n",
       "      <td>-40.23377</td>\n",
       "      <td>-83.22141</td>\n",
       "      <td>4.44391</td>\n",
       "      <td>73.15568</td>\n",
       "      <td>3.25023</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 90 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              1         2         3         4         5         6         7   \\\n",
       "200000  45.09540 -57.29138  20.05392   0.10414  12.00346 -14.31259   2.54547   \n",
       "200001  45.11673 -18.39958  -1.61552  -3.67929 -13.24027  -6.84375 -10.33838   \n",
       "200002  46.85191   9.44824   0.31518 -16.85413  15.42389 -15.82587 -21.21385   \n",
       "200003  45.79644 -36.86230  21.99320 -10.42360  -2.89410  -8.84010 -23.73864   \n",
       "200004  40.92442 -43.26026 -18.72100 -11.72495 -19.69395 -10.54229  -9.91945   \n",
       "\n",
       "              8         9        10  ...        81         82        83  \\\n",
       "200000  -3.33986  -3.60175 -8.99414  ...   7.15503  -81.48297  34.91722   \n",
       "200001 -11.12891  16.56924  6.70243  ...   8.26246 -138.26547  23.59451   \n",
       "200002 -10.16067  14.45113  2.22865  ...  13.24783  -91.25475  41.12300   \n",
       "200003  -9.82956  13.08399  0.57577  ...   8.13403 -194.21155  55.55883   \n",
       "200004 -14.85633   9.37409 -0.93093  ...  17.63167 -203.56276  27.10714   \n",
       "\n",
       "              84       85         86         87       88        89       90  \n",
       "200000  15.11917 -6.09056  -78.52893  -46.87559  2.89571  64.25949  0.78578  \n",
       "200001  60.99156  4.18891  105.75496 -126.39851 -3.80727  55.00532 -3.42354  \n",
       "200002  55.22389  8.33048    7.57355   17.34516  0.21543 -58.34520  3.92760  \n",
       "200003  21.65387  6.41164  -47.18867 -212.14270 -4.67550 -86.99988 -6.12034  \n",
       "200004  36.90795  0.62431  -40.23377  -83.22141  4.44391  73.15568  3.25023  \n",
       "\n",
       "[5 rows x 90 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# peak at the data if desired\n",
    "X_test.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Train the model  (Note, this may take a bit of time for larger values of _num_est_. )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=8,\n",
       "                      max_features='auto', max_leaf_nodes=None,\n",
       "                      min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "                      min_samples_leaf=1, min_samples_split=2,\n",
       "                      min_weight_fraction_leaf=0.0, n_estimators=100,\n",
       "                      n_jobs=None, oob_score=False, random_state=None,\n",
       "                      verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestRegressor\n",
    "num_est=100\n",
    "\n",
    "skl_model = RandomForestRegressor(n_estimators=num_est, max_depth=8)\n",
    "skl_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### scikit-learn (CPU only)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "382 ms ± 4.7 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "skl_time = %timeit -o skl_model.predict(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Convert scikit-learn model to PyTorch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = convert(skl_model, 'torch')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Time PyTorch - CPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "317 ms ± 8.94 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "pred_cpu_hb = %timeit -o model.predict(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Switch PyTorch from CPU to GPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%capture \n",
    "model.to('cuda')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Time PyTorch - GPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "12.5 ms ± 88.3 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n"
     ]
    }
   ],
   "source": [
    "pred_gpu_hb = %timeit -o model.predict(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Note: Continuing in this notebook requires TVM built with LLVM support. Install instructions [here](https://tvm.apache.org/docs/install/index.html)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Convert scikit-learn model to TVM (CPU)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_tvm = convert(skl_model, 'tvm', data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Time TVM - CPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "155 ms ± 1.58 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    }
   ],
   "source": [
    "pred_cpu_tvm = %timeit -o model_tvm.predict(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Convert scikit-learn model to TVM (GPU)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_tvm = convert(skl_model, 'tvm', data, 'cuda')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Time TVM - GPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20.5 ms ± 166 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    }
   ],
   "source": [
    "pred_gpu_tvm = %timeit -o model_tvm.predict(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Plot the results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot(title, skl_time, pred_cpu_hb, pred_gpu_hb, pred_cpu_tvm, pred_gpu_tvm):\n",
    "    import matplotlib.pyplot as plt\n",
    "    import numpy as np\n",
    "    from matplotlib.pyplot import cm\n",
    "\n",
    "    fig = plt.figure()\n",
    "\n",
    "    x = ['skl','pyt-cpu','pyt-gpu','tvm-cpu','tvm-gpu']\n",
    "    height = [skl_time.average,pred_cpu_hb.average,pred_gpu_hb.average,pred_cpu_tvm.average,pred_gpu_tvm.average]\n",
    "    width = 1.0\n",
    "    plt.ylabel('time in seconds')\n",
    "    plt.xlabel(title)\n",
    "\n",
    "    rects = plt.bar(x, height, width, color=cm.rainbow(np.linspace(0,1,5)))\n",
    "    def autolabel(rects):\n",
    "\n",
    "        for rect in rects:\n",
    "            height = rect.get_height()\n",
    "            plt.text(rect.get_x() + rect.get_width()/2., 1.05*height,\n",
    "                    '%.4f' % (height),\n",
    "                    ha='center', va='bottom')\n",
    "\n",
    "    autolabel(rects)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEQCAYAAACqduMIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/d3fzzAAAACXBIWXMAAAsTAAALEwEAmpwYAAAp1klEQVR4nO3de5xWZb338c8XJiAhS2V26QBxGDyAsEcc1NIOmgppDh0w0TJNjTR4PJf6ZGiYr8B8cm83mJoHzF2SyhbHRNQ85JM7g0EJBTMQKGY2TykiZirE8Hv+WGuGm5k1cAOz5ubwfb9e94u1rsNa1yV4/+61rmtdSxGBmZlZS51K3QAzM9sxOUCYmVkmBwgzM8vkAGFmZpkcIMzMLJMDhJmZZXKA2E6SRkp6RdISSZdn5J8r6UVJ8yX9VtKgNP19ku5K816WdEWa3lvSU5IWSVoo6YKCY10tqSE91nxJJ3RcT81sdyM/B7HtJHUG/gQcB9QDc4FTI2JRQZk9I+KtdLsG+FZEjJR0GlATEWMk7QEsAj4NrAX2jYjnJX0AmAd8PiIWSboaeDsiru+4XprZ7spXENvnMGBJRCyNiHXAdGBUYYGm4JDqDjRF5AC6SyoD3g+sA96KiJUR8Xxa9+/Ay0BFvt0wM2vNAWL7VAArCvbryfgylzRO0qvAdcD5afL9wD+AlcBfgOsj4o0W9foChwC/L0geL2mBpDsk7dVeHTEza8kBogNExNSIGABcBlyZJh8GNAL7Af2ASyT1b6ojqQcwA7iw4CrkJ8AAoIoksPyfDumAme2WHCC2TwPQu2C/V5rWlunA59Pt04DZEfHPiPgb8CxQDckANklw+HlE/FdT5Yj4a0Q0RsQG4KckQcbMLBcOENtnLjBQUj9JXYAxQG1hAUkDC3ZPBBan238BjknLdAeOAP4oScDtwMsR8eMWx9q3YPcLwEvt2Bczs03kGiC2NAW0oNyXJIWk6oK0K9J6r0gakWc7t1VErAfGA4+SDCbfGxELJU1MZyxBMmawUNJ84GLgjDR9KtBD0kKSQHNnRCwAjgROB47JmM56XTotdgFwNHBRR/TTzHZPuU1zLWYKaFruA8DDQBdgfETUpc8K3ENyC2U/4NfA/hHRmEtjzcyslbIcj908BRRAUtMU0EUtyl0DTAa+XZA2CpgeEWuBZZKWpMf7XVsn69mzZ/Tt27f9Wm9mthuYN2/e6xFRnpWXZ4DImgJ6eGEBScOA3hHxsKRvt6j7XIu6WdNHxwJjAfr06UNdXV07Nd3MbPcg6c9t5ZVskFpSJ+DHwCXbeoyIuDUiqiOiurw8MwCamdk2yvMKYktTQD8AHAw8nUzc4SNAbTq4u7XTR83MrJ3leQWx2SmgEbEmInpGRN+I6EtyS6kmIurScmMkdZXUDxgIzMmxrWZm1kJuVxARsV5S0xTQzsAdTVNAgbqIqN1M3YWS7iUZ0F4PjPMMJjOzjrXLrOZaXV0dHqQ2M9s6kuZFRHVWnp+kNjOzTA4QZmaWyQHCzMwyOUCYmVmmPJ+D2Kl8X6VuQce7ateYn2BmOfEVhJmZZXKAMDOzTA4QZmaWyQHCzMwyOUCYmVkmBwgzM8vkAGFmZpkcIMzMLJMDhJmZZXKAMDOzTA4QZmaWyQHCzMwyOUCYmVmmXAOEpJGSXpG0RNLlGfnnSnpR0nxJv5U0KE3vK+ndNH2+pJvzbKeZmbWW23LfkjoDU4HjgHpgrqTaiFhUUOwXEXFzWr4G+DEwMs17NSKq8mqfmZltXp5XEIcBSyJiaUSsA6YDowoLRMRbBbvdAb+hwMxsB5FngKgAVhTs16dpm5A0TtKrwHXA+QVZ/SS9IOk3kj6RdQJJYyXVSap77bXX2rPtZma7vZIPUkfE1IgYAFwGXJkmrwT6RMQhwMXALyTtmVH31oiojojq8vLyjmv0bm727NkccMABVFZWMmnSpFb5N998M0OGDKGqqoqjjjqKRYuSu4qrVq3i6KOPpkePHowfP36TOuvWrWPs2LHsv//+HHjggcyYMQOAadOmUV5eTlVVFVVVVdx22235d9DMgHxfOdoA9C7Y75WmtWU68BOAiFgLrE2356VXGPsDdfk01YrV2NjIuHHjePzxx+nVqxfDhw+npqaGQYMGNZc57bTTOPfccwGora3l4osvZvbs2XTr1o1rrrmGl156iZdeemmT41577bX8y7/8C3/605/YsGEDb7zxRnPeKaecwpQpUzqmg2bWLM8riLnAQEn9JHUBxgC1hQUkDSzYPRFYnKaXp4PcSOoPDASW5thWK9KcOXOorKykf//+dOnShTFjxvDggw9uUmbPPTde7P3jH/9ASl743b17d4466ii6devW6rh33HEHV1xxBQCdOnWiZ8+eOfbCzIqRW4CIiPXAeOBR4GXg3ohYKGliOmMJYLykhZLmk9xKOiNN/ySwIE2/Hzg3It7ASq6hoYHevTdeGPbq1YuGhtYXhlOnTmXAgAF85zvf4cYbb9zsMd98800Avve97zFs2DBOPvlk/vrXvzbnz5gxg6FDhzJ69GhWrFjRxlHMrL3lOgYREbMiYv+IGBAR16ZpEyKiNt2+ICIGR0RVRBwdEQvT9BkF6cMi4qE822ntb9y4cbz66qtMnjyZH/zgB5stu379eurr6/n4xz/O888/z8c+9jEuvfRSAE466SSWL1/OggULOO644zjjjDM2eywzaz8lH6S2nUtFRcUmv+Lr6+upqGg1Oa3ZmDFjmDlz5maPuc8++7DHHnvwxS9+EYCTTz6Z559/vjmva9euAJxzzjnMmzdvO3tgZsVygLCtMnz4cBYvXsyyZctYt24d06dPp6amZpMyixcvbt5++OGHGThwYMvDbEISJ510Ek8//TQATzzxRPOg98qVK5vL1dbWctBBB7VTT8xsS/KcxWS7oLKyMqZMmcKIESNobGzkrLPOYvDgwUyYMIHq6mpqamqYMmUKv/71r3nf+97HXnvtxV133dVcv2/fvrz11lusW7eOmTNn8thjjzFo0CAmT57M6aefzoUXXkh5eTl33nknADfeeCO1tbWUlZWx9957M23atBL13Gz3o4hd4+Hl6urqqKvb9lmw31c7NmYncdWu8VdvZttB0ryIqM7K8y0mMzPL5ABhZmaZHCDMzCyTA4SZmWVygDAzs0wOEGZmlskBwszMMjlAmJlZJj9JvRvTrNdL3YQOFSd4CXGzreErCDMzy+QAYWZmmRwgzMwskwOEmZllcoAwM7NMuQYISSMlvSJpiaTLM/LPlfSipPmSfitpUEHeFWm9VySNyLOdZmbWWm4BQlJnYCrwWWAQcGphAEj9IiKGREQVcB3w47TuIGAMMBgYCdyUHs/MzDpInlcQhwFLImJpRKwDpgOjCgtExFsFu92BplfYjAKmR8TaiFgGLEmPZ2ZmHSTPB+UqgBUF+/XA4S0LSRoHXAx0AY4pqPtci7oVGXXHAmMB+vTp0y6NNjOzRMkHqSNiakQMAC4DrtzKurdGRHVEVJeXl+fTQDOz3VSeAaIB6F2w3ytNa8t04PPbWNfMzNpZngFiLjBQUj9JXUgGnWsLC0gaWLB7IrA43a4FxkjqKqkfMBCYk2NbzcyshdzGICJivaTxwKNAZ+COiFgoaSJQFxG1wHhJxwL/BFYDZ6R1F0q6F1gErAfGRURjXm01M7PWcl3NNSJmAbNapE0o2L5gM3WvBa7Nr3VmZrY5JR+kNjOzHZMDhJmZZXKAMDOzTA4QZmaWyQHCzMwyOUCYmVkmBwgzM8vkAGFmZpkcIMzMLJMDhJmZZdpigJDUXVKndHt/STWS3pd/08zMrJSKuYJ4BugmqQJ4DDgdmJZno8zMrPSKCRCKiHeALwI3RcTJJO+KNjOzXVhRAULSx4CvAA+naZ3za5KZme0IigkQFwJXAA+k72noDzyVa6vMzKzktvg+iIj4DfCbgv2lwPl5NsrMzEqvzQAh6SEg2sqPiJpcWmRmZjuEzV1BXJ/++UXgI8B/pvunAn/Ns1FmZlZ6bQaI9NYSkv5PRFQXZD0kqa6Yg0saCfw7yaD2bRExqUX+xcA5JO+dfg04KyL+nOY1Ai+mRf/iKxYzs45VzCB193RgGgBJ/YDuW6okqTMwFfgsMAg4VdKgFsVeAKojYihwP3BdQd67EVGVfhwczMw62BYHqYGLgKclLQUEfBT4ZhH1DgOWpIPaSJoOjAIWNRWIiMLZUM8BXy2y3WZmlrNiZjHNljQQODBN+mNErC3i2BXAioL9euDwzZQ/G3ikYL9beitrPTApIma2rCBpLDAWoE+fPkU0yczMilXMFQTAoUDftPy/SiIiftZejZD0VaAa+FRB8kcjoiG9vfWkpBcj4tXCehFxK3ArQHV1dZszrszMbOttMUBIuhsYAMwHGtPkALYUIBqA3gX7vdK0lsc/Fvgu8KnCK5OIaEj/XCrpaeAQ4NWW9c3MLB/FXEFUA4MiYmt/oc8FBqaD2g3AGOC0wgKSDgFuAUZGxN8K0vcC3omItZJ6Akey6QC2mZnlrJgA8RLJcxArt+bAEbFe0njgUZJprnekS3VMBOoiohb4EdADuE8SbJzOehBwi6QNJDOtJkXEoswTmZlZLooJED2BRZLmAIW3gLY49TQiZgGzWqRNKNg+to16/w0MKaJtZmaWk2ICxNV5N8LMzHY8RS3WJ+nDwPA0aU7heIGZme2ainnl6JeBOcDJwJeB30sanXfDzMystIq5xfRdYHjTVYOkcuDXJEtjmJnZLqqYtZg6tbiltKrIemZmthMr5gpitqRHgXvS/VPYdEkMMzPbBRUzSP1tSV8EjkqTbo2IB/JtlpmZlVoxS230A2ZFxH+l+++X1DcilufdODMzK51ixhLuAzYU7DemaWZmtgsrJkCURcS6pp10u0t+TTIzsx1BMQHiNUnNy2pIGgW8nl+TzMxsR1DMLKZzgZ9LmkqyzHc98LVcW2VmZiVXzCymV4EjJPVI99/OvVVmZlZyxSy18WFJtwP3RcTbkgZJOrsD2mZmZiVUzBjENJJ3OuyX7v8JuDCn9piZ2Q6imADRMyLuJZ3qGhHr2fjqUTMz20UVEyD+IWkfkgFqJB0BrMm1VWZmVnLFzGK6GKgFBkh6FigHvNy3mdkubotXEBHxPPAp4OPAN4HBEbGgmINLGinpFUlLJF2ekX+xpEWSFkh6QtJHC/LOkLQ4/ZxRfJfMzKw9FDOL6WTg/RGxEPg88EtJw4qo1xmYCnwWGAScKmlQi2IvANURMZTk/RLXpXX3Bq4CDgcOA66StFexnTIzs+1XzBjE9yLi75KOAj4D3A78pIh6hwFLImJpujzHdGBUYYGIeCoi3kl3nwN6pdsjgMcj4o2IWA08Dows4pxmZtZOigkQTTOWTgR+GhEPU9xaTBXAioL9+jStLWez8T0TW1vXzMzaWTGD1A2SbgGOAyZL6ko7v1FO0leBapKxjq2pNxYYC9CnT5/2bJKZ2W6vmC/6L5M8KDciIt4E9ga+XUS9BqB3wX6vNG0Tko4lee91TUSs3Zq6EXFrRFRHRHV5eXkRTTIzs2IVM4vpnYj4r4hYnO6vjIjHijj2XGCgpH6SugBjSKbLNpN0CHALSXAofO/1o8DxkvZKB6ePT9PMzKyDFHOLaZtExHpJ40m+2DsDd0TEQkkTgbqIqAV+BPQA7pME8JeIqImINyRdQxJkACZGxBt5tdXMzFrLLUAARMQsYFaLtAkF28dupu4dwB35tc7MzDanXQebzcxs11HMg3JfTJ9mXiPpLUl/l/RWRzTOzMxKp5hbTNcBJ0XEy3k3xszMdhzF3GL6q4ODmdnup5griDpJvwRmAk3PKRAR/5VXo8zMrPSKuYLYE3iH5FmEk9LP5/JslJmV3uzZsznggAOorKxk0qRJrfKfeeYZhg0bRllZGffff/8meZ07d6aqqoqqqipqamqa088880z69evXnDd//vxN6s2dOzfzeFYaW7yCiIivd0RDzGzH0djYyLhx43j88cfp1asXw4cPp6amhkGDNi7I3KdPH6ZNm8b111/fqv773//+Vl/+TX70ox8xenTrV8o0NjZy2WWXcfzxx7dbP2z7tBkgJH0nIq6T9B+kb5MrFBHn59oyMyuZOXPmUFlZSf/+/QEYM2YMDz744CYBom/fvgB06tQ+s+X/4z/+gy996UvMnTt3y4WtQ2zub7ZpYLoOmJfxMbNdVENDA717b1wOrVevXjQ0tFoOrU3vvfce1dXVHHHEEcycOXOTvO9+97sMHTqUiy66iLVr1zaf74EHHuC8885rl/Zb+2jzCiIiHkr/vKvjmmNmu4I///nPVFRUsHTpUo455hiGDBnCgAED+OEPf8hHPvIR1q1bx9ixY5k8eTITJkzgwgsvZPLkye12NWLtI9elNsxs51RRUcGKFRtfyVJfX09FRfGvZGkq279/fz796U/zwgsvMGDAAPbdd18Aunbtyte//vXm8Yu6ujrGjBkDwOuvv86sWbMoKyvj85//fDv1yLaFw7WZtTJ8+HAWL17MsmXLWLduHdOnT99kNtLmrF69uvnW0euvv86zzz7bPHaxcuVKACKCmTNncvDBBwOwbNkyli9fzvLlyxk9ejQ33XSTg8MOwFcQZtZKWVkZU6ZMYcSIETQ2NnLWWWcxePBgJkyYQHV1NTU1NcydO5cvfOELrF69moceeoirrrqKhQsX8vLLL/PNb36TTp06sWHDBi6//PLmAPGVr3yF1157jYigqqqKm2++ucQ9tc1RRKsJSpsWkPYneQf1hyPiYElDSd7f8IOOaGCxqquro66ubpvrf1/t2JidxNUPv17qJnSoOKFnqZtgtsORNC8iqrPyirnF9FPgCuCfABGxgOTlP2ZmtgsrJkDsERFzWqStz6MxZma24ygmQLwuaQDpw3KSRgMrc22VmZmVXDGD1OOAW4EDJTUAy4Cv5toqMzMruS1eQUTE0vTVoOXAgRFxVEQsL+bgkkZKekXSEkmXZ+R/UtLzktanVyaFeY2S5qef2iL7Y2Zm7WSLVxCSPgR8DegLlEnJdJ8trcUkqTMwFTgOqAfmSqqNiEUFxf4CnAlcmnGIdyOiakvtM7PN+NVuuNbm5+4sdQt2GcXcYpoFPAe8CGzYimMfBiyJiKUAkqYDo4DmANF0JSJpa45rZmYdoJgA0S0iLt6GY1cAKwr264HDt6J+N0l1JDOmJkXEzG1og5mZbaNiAsTdkr4B/IpN3yj3Rm6tSnw0Ihok9QeelPRiRLxaWEDSWGAsJGvTm5lZ+ylmmus64EfA79i41Hcxjyw3AL0L9nulaUWJiIb0z6XA08AhGWVujYjqiKguLy8v9tBmZlaEYq4gLgEqI2Jr12WYCwyU1I8kMIwBTiumoqS9gHciYq2knsCRwHVbeX4zM9sOxVxBLCF5J/VWiYj1wHjgUZKXD90bEQslTZRUAyBpuKR64GTgFkkL0+oHAXWS/gA8RTIGsaj1WczMLC/FXEH8A5gv6Sk2HYPY4itHI2IWySyowrQJBdtzSW49taz338CQItpmZmY5KSZAzEw/Zma2G9ligPArR83Mdk9tBghJ90bElyW9SLpQX6GIGJpry8zMrKQ2dwVxQfrn5zqiIWZmtmNpcxZTRDQt6f2tiPhz4Qf4Vsc0z8zMSqWYaa7HZaR9tr0bYmZmO5bNjUGcR3Kl0F/SgoKsDwDP5t0wMzMrrc2NQfwCeAT4IVD4Loe/d8A6TGZmVmJtBoiIWAOsAU7tuOaYmdmOopgxCDMz2w05QJiZWSYHCDMzy+QAYWZmmRwgzMwskwOEmZllcoAwM7NMDhBmZpbJAcLMzDLlGiAkjZT0iqQlki7PyP+kpOclrZc0ukXeGZIWp58z8mynmZm1lluAkNQZmEqy8usg4FRJg1oU+wtwJsm6T4V19wauAg4HDgOukrRXXm01M7PW8ryCOAxYEhFLI2IdMB0YVVggIpZHxAJgQ4u6I4DHI+KNiFgNPA6MzLGtZmbWQp4BogJYUbBfn6blXdfMzNrBTj1ILWmspDpJda+99lqpm2NmtkvJM0A0AL0L9nulae1WNyJujYjqiKguLy/f5oaamVlreQaIucBASf0kdQHGALVF1n0UOF7SXung9PFpmpmZdZDcAkRErAfGk3yxvwzcGxELJU2UVAMgabikeuBk4BZJC9O6bwDXkASZucBEv8XOzKxjbe6Vo9stImYBs1qkTSjYnkty+yir7h3AHXm2z8zM2rZTD1KbmVl+HCDMzCyTA4SZmWVygDAzs0wOEGZmlskBwszMMjlAmJlZJgcIMzPL5ABhZmaZHCDMzCyTA4SZmWVygDAzs0wOEGZmlskBwszMMjlAmJlZJgcIMzPL5ABhZmaZHCDMzCyTA4SZmWXKNUBIGinpFUlLJF2ekd9V0i/T/N9L6pum95X0rqT56efmPNtpZmatleV1YEmdganAcUA9MFdSbUQsKih2NrA6IioljQEmA6ekea9GRFVe7TMzs83L8wriMGBJRCyNiHXAdGBUizKjgLvS7fuBz0hSjm0yM7Mi5RkgKoAVBfv1aVpmmYhYD6wB9knz+kl6QdJvJH0i6wSSxkqqk1T32muvtW/rzcx2czvqIPVKoE9EHAJcDPxC0p4tC0XErRFRHRHV5eXlHd5IM7NdWZ4BogHoXbDfK03LLCOpDPggsCoi1kbEKoCImAe8CuyfY1vNzKyFPAPEXGCgpH6SugBjgNoWZWqBM9Lt0cCTERGSytNBbiT1BwYCS3Nsq5mZtZDbLKaIWC9pPPAo0Bm4IyIWSpoI1EVELXA7cLekJcAbJEEE4JPAREn/BDYA50bEG3m11czMWsstQABExCxgVou0CQXb7wEnZ9SbAczIs21mZrZ5O+ogtZlZh5s9ezYHHHAAlZWVTJo0qVX+2rVrOeWUU6isrOTwww9n+fLlADz++OMceuihDBkyhEMPPZQnn3yyuc68efMYMmQIlZWVnH/++UQEAFdffTUVFRVUVVVRVVXFrFmzWp2v1BwgzMyAxsZGxo0bxyOPPMKiRYu45557WLRo0SZlbr/9dvbaay+WLFnCRRddxGWXXQZAz549eeihh3jxxRe56667OP3005vrnHfeefz0pz9l8eLFLF68mNmzZzfnXXTRRcyfP5/58+dzwgkndExHt4IDhJkZMGfOHCorK+nfvz9dunRhzJgxPPjgg5uUefDBBznjjGRezejRo3niiSeICA455BD2228/AAYPHsy7777L2rVrWblyJW+99RZHHHEEkvja177GzJkzO7pr28wBwswMaGhooHfvjTPze/XqRUNDQ5tlysrK+OAHP8iqVas2KTNjxgyGDRtG165daWhooFevXm0ec8qUKQwdOpSzzjqL1atX59Gt7eIAYWbWThYuXMhll13GLbfcssWy5513Hq+++irz589n33335ZJLLumAFm4dBwizImzr4OWqVas4+uij6dGjB+PHj28u/84773DiiSdy4IEHMnjwYC6/fONix9OmTaO8vLx58PK2227LvX8GFRUVrFixcXWg+vp6Kioq2iyzfv161qxZwz777NNc/gtf+AI/+9nPGDBgQHP5+vr6zGN++MMfpnPnznTq1IlvfOMbzJkzJ9f+bQsHCLMt2J7By27dunHNNddw/fXXtzrupZdeyh//+EdeeOEFnn32WR555JHmvFNOOaV58PKcc87Jt4MGwPDhw1m8eDHLli1j3bp1TJ8+nZqamk3K1NTUcNddyfqi999/P8cccwySePPNNznxxBOZNGkSRx55ZHP5fffdlz333JPnnnuOiOBnP/sZo0Yla5auXLmyudwDDzzAwQcf3AG93DoOEGZbsD2Dl927d+eoo46iW7dum5TfY489OProowHo0qULw4YN2+SXpnW8srIypkyZwogRIzjooIP48pe/zODBg5kwYQK1tckiEGeffTarVq2isrKSH//4x81Xk1OmTGHJkiVMnDix+crvb3/7GwA33XQT55xzDpWVlQwYMIDPfvazAHznO99hyJAhDB06lKeeeoobbrihNB3fjFwflDPbFWQNXv7+979vs0zh4GXPnj23ePw333yThx56iAsuuKA5bcaMGTzzzDPsv//+3HDDDZuc3/JzwgkntJpuOnHixObtbt26cd9997Wqd+WVV3LllVdmHrO6upqXXnqpVfrdd9+9na3Nn68gzEpo/fr1nHrqqZx//vn0798fgJNOOonly5ezYMECjjvuuOYrE7OO5gBhtgXbO3i5OWPHjmXgwIFceOGFzWn77LMPXbt2BeCcc85h3rx57dCL3Yi0+31y4gBhtgXbM3i5OVdeeSVr1qzh3/7t3zZJLxy8rK2t5aCDDmqfjphtJY9BmG1B4eBlY2MjZ511VvPgZXV1NTU1NZx99tmcfvrpVFZWsvfeezN9+vTm+n379uWtt95i3bp1zJw5k8cee4w999yTa6+9lgMPPJBhw4YBMH78eM455xxuvPFGamtrKSsrY++992batGkl6rnt7tS0cNTOrrq6Ourq6ra5/vd3wzdhX/3w66VuQoeKE7Y8YLzL+dXXS92CjnfStFK3oONtx/e4pHkRUZ2V5ysI221M5OFSN6HDTdhyEbM2eQzCzMwyOUCYmVkmBwgzM8uUa4CQNFLSK5KWSLo8I7+rpF+m+b+X1Lcg74o0/RVJI/Jsp5mZtZZbgJDUGZgKfBYYBJwqaVCLYmcDqyOiErgBmJzWHQSMAQYDI4Gb0uOZmVkHyfMK4jBgSUQsjYh1wHRgVIsyo4C70u37gc8oebpoFDA9ItZGxDJgSXo8MzPrIHlOc60AVhTs1wOHt1UmItZLWgPsk6Y/16JuRYu6SBoLjE1335b0Svs0vUP1BErzQMKJJTkrlKjPV3X0CTflPu8eStPn7Vtu46NtZezUz0FExK3AraVux/aQVNfWQyq7Kvd59+A+7/zyvMXUABSuUdwrTcssI6kM+CCwqsi6ZmaWozwDxFxgoKR+krqQDDrXtihTCzStZTwaeDKStT9qgTHpLKd+wEBgx3sfn5nZLiy3W0zpmMJ44FGgM3BHRCyUNBGoi4ha4HbgbklLgDdIgghpuXuBRcB6YFxENObV1hLbqW+RbSP3effgPu/kdpnF+szMrH35SWozM8vkAGFmZpkcIEpI0nJJrV5SIOntUrSnvUn636VuQ0famfsr6UOSvlXqdnSk3bHPW8sBwvK0035hbqOdub8fAna3L8sPsfv1eas4QHQQSd0lPSzpD5JeknRKQd77JT0i6RulbOOWSOor6Y+Sfi7pZUn3SzpB0syCMsdJekDSJOD9kuZL+nnGsXpIulPSi5IWSPpSmv62pBskLZT0hKTyNP1pSdXpdk9Jy3ey/pZLejzt122S/pz2I+sce6R1mq8wJVVLejrH7k4CBqTtv09S83P2kqZJGi3pTEkz034slzRe0sWSXpD0nKS9sw4s6Wvp3/EfJN1dcMybJdVJ+pOkz6XpZ0qaUlD3V5I+vRP2+XtKFhr9raR7JF2apj8t6d/Tc74k6bA0/eqmMun+SypYvLRUHCA6zkjgfyLiXyPiYGB2mt4DeAi4JyJ+WrLWFe8A4KaIOAh4i2RBxQObvsiBr5NMab4ceDciqiLiKxnH+R6wJiKGRMRQ4Mk0vTvJNOjBwG8o+WoR7dbfq0ie8xlMsu5Yn82coxS/ai8HXo2IKuAXwJcBlDzD9Blofh3fwcAXgeHAtcA7EXEI8Dvgay0PKmkwcCVwTET8K3BBQXZfkjXWTgRultSt3Xu1eXn1eTjwJeBfSRYrbflk9R7pOb8F3NGuPWpnDhAd50XgOEmTJX0iItak6Q8Cd0bEz0rYtq2xIiKeTbf/EzgSuBv4qqQPAR8DHiniOMeSrPYLQESsTjc3AL8sOP5R7dDm7dFe/T2KZMFKImI2sLogr+U5St3nR4CjJXUl+YJ7JiLeTfOeioi/R8RrwBqSHzeQ/Pvum3GsY4D7IuJ1gIh4oyDv3ojYEBGLgaXAge3flaK1Z5+PBB6MiPci4u8F5ZvcAxARzwB7pv+Odkg79VpMO5OI+JOkYcAJwA8kPZFmPQuMlPSL2DkeSmnZxgDuJPmf4D2SL4P1LStJGgc03UI7YRvOt56NP2g68pdmR/Q36xxQoj5HxHvp7awRwCmkgS21tmB7Q8H+BqBMUm82fiHevKVTZewX9hk6qN8d2GfYgfq9Jb6C6CCS9iO5NP1P4EfAsDRrAsmvyalt1d3B9JH0sXT7NOC3EfE/wP+Q3Eq4s6DsPyW9DyAipqa3X6rS8o8D45oKStor3exEsuxK8/HT7eXAoel2U35HaK/+PsvGWxjHA3sV1Gt1jnR7ORv7/KX27VYrfwc+ULD/S5LbZ59g4+3QLYqIFQX9vpnk1uHJkvYBaHHP/mRJnSQNAPoDr5D0uSpN702+y/zn1edngZMkdZPUA/hciyqnAEg6iuQ26xqSfg9L04cB/batS+3LAaLjDAHmSJpPcj/6BwV5F5AMcF5XioZtpVeAcZJeJvmS+0ma/nOSWyUvF5S9FViQNWhL0v+90sG4PwBHp+n/AA6T9BLJ7YmJafr1wHmSXiBZUrmjtFd/vw8cn/brZOD/kXxBbe4c3wf+XVIdkOtSMxGxCng2/fv4EfAY8Cng1+n7XLb1uAtJ7tv/Jv17/nFB9l9I1lh7BDg3It4j+XJdRrLMzo3A89t67iLallef55KsJ7eApG8vktyaavJe+u/4ZpKXpgHMAPaWtBAYD/xpW8/fnrzUhhUtnVXxq3SQvWXeFOCFiLh9O8/xdkT02J5jtJf27G96b7sxXaPsY8BPIqJqc+fYlUmaRtLv+0vdljxI6hERbyuZkfYMMDYink9vY10aEXWlbWFxPAZh203SPJJf/peUui0dYRv72we4V1InYB0bxyds13SrklcndwPuiojcroTy5CsIMzPL5DEIMzPL5ABhZmaZHCDMzCyTA4SVjKTvKlmbaEG6Ns3haXrhukv9JC2WNELSpyX9ajvOd62kFWqxWq6SV9v+UtISSb8vXANH0hVp+iuSRmzFubopWWNpSEHatyXdsq3tb+M8jel/u4VK1jq6JB0I31ydvpJOa892pMe9MJ21Y7sIBwgriXSq5+eAYelaTMcCK1qU6UXywNIlEfFoO5z2IbIfvDobWB0RlcANwOT0/INIXoM7mGQtrZskdS7mROmc/gvTOpJUAZxLsv7PNpGUNeuwaf2nwcBxJMtEbGn9qr4kD+S1twsBB4hdiAOElcq+wOsRsRYgIl5PnzguzH8M+G76/vLtFhHPRcTKjKxRwF3p9v3AZyQpTZ8eEWsjYhmwhK14sjddc2klyYJuNwBXkyzNMEPS3PRzJICkwyT9Tskqof8t6YA0/UxJtZKeBJ5o41RN5/sbMBYYnwalvpL+r6Tn08/H06KTgE+kVx4XtVVO0r6SntHGlUc/kaYfn7b1eSWroPaQdD6wH/CUpKeK/W9kO7iI8MefDv+QrGI7n+SJ0ZuATxXkPQ28AXyrRZ1Pkzxctb3nfrvF/ktAr4L9V0me1p4CfLUg/XZg9Faeaz+gnmTBN0hWDT0q3e4DvJxu7wmUpdvHAjPS7TPT+nsX05c07U3gwyS/5rulaQNJVslt9d9xM+UuIQnQAJ1JlqXoSfLgV/c0/TJgQrq9HOhZ6n9b/rTfxw/KWUlE8pTpoSTr3hwN/FLS5RExLS3ya5IVU6dFxDulauf2ioj/SX/9N42dHAsMSi5QgGQ1zx7AB4G7JA0kWbztfQWHeTw2XQW1WO8DpkiqIlmqY/+tLDcXuEPJ+lIzI2K+pE8Bg0iWqADoQrLste2CfIvJSiYiGiPi6Yi4imT9mcIF6a4j+YK6r417761I6p3eDpkv6dytaEoD0Ds9RhnJl/WqwvRUrzSt8JxfKDhny3X/m2xIP5D8P3dEbFzcrSIi3gauIbnKOBg4iU1X8/xHsR2R1J/kS/5vwEXAX0neS1BN8mWeJbNcJMtRfzLt8zRJXwNEErCa2j8oIs5u47i2k3OAsJKQdED6a7lJFfDnFsUuJHmBzu0q+Mndlmi9qmaxaoEz0u3RJC/2iTR9TDrLqR/J7Zc5Lc75QME5i1lf5zHgfzXtpL/aIQlKTcHnzK1oezMlLzG6GZiStv+DwMqI2ACcTnKbCFqvYppZTtJHgb9G8iKr20hWG30OOFJSZVqmu6T92ziu7eQcIKxUepDcUlkkaQHJbYurCwukX3JnkAxYN610+xlJ9QWfj1EkSddJqgf2SOs2ne92YB9JS4CLSWcaRbIS6b0kK4vOBsZFxPauqno+UK1kau8ikplNpP37oZJVPrfm1m/Ta04XktyWe4xkFVhIxnbOULKK6oFsvBJZADSm02Iv2ky5TwN/SNt0CvDvkbw050zgnvTv7XdsfNHPrcBsD1LvOrwWk5mZZfIVhJmZZXKAMDOzTA4QZmaWyQHCzMwyOUCYmVkmBwgzM8vkAGFmZpn+PyP5mbwM5QpCAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "chartname = \"SKL - \" + str(num_est) + \" - Year Dataset\"\n",
    "\n",
    "plot(chartname, skl_time, pred_cpu_hb, pred_gpu_hb, pred_cpu_tvm, pred_gpu_tvm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}