Chapter5_3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Regularization of the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.datasets import imdb\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "import numpy as np\n",
    "\n",
    "(train_data, train_labels), _ = imdb.load_data(num_words=10000)\n",
    "def vectorize_sequences(sequences, dimension=10000):\n",
    "    results = np.zeros((len(sequences), dimension))\n",
    "    for i, sequence in enumerate(sequences):\n",
    "        results[i, sequence] = 1.\n",
    "    return results\n",
    "train_data = vectorize_sequences(train_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "30/30 [==============================] - 1s 34ms/step - loss: 0.5117 - accuracy: 0.7911 - val_loss: 0.3868 - val_accuracy: 0.8751\n",
      "Epoch 2/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.3110 - accuracy: 0.9026 - val_loss: 0.3056 - val_accuracy: 0.8864\n",
      "Epoch 3/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2274 - accuracy: 0.9251 - val_loss: 0.2757 - val_accuracy: 0.8953\n",
      "Epoch 4/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.1816 - accuracy: 0.9406 - val_loss: 0.2766 - val_accuracy: 0.8908\n",
      "Epoch 5/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.1486 - accuracy: 0.9529 - val_loss: 0.2921 - val_accuracy: 0.8860\n",
      "Epoch 6/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.1234 - accuracy: 0.9615 - val_loss: 0.2909 - val_accuracy: 0.8870\n",
      "Epoch 7/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.1033 - accuracy: 0.9680 - val_loss: 0.3057 - val_accuracy: 0.8868\n",
      "Epoch 8/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0859 - accuracy: 0.9757 - val_loss: 0.3250 - val_accuracy: 0.8843\n",
      "Epoch 9/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0713 - accuracy: 0.9804 - val_loss: 0.3603 - val_accuracy: 0.8799\n",
      "Epoch 10/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0592 - accuracy: 0.9841 - val_loss: 0.3939 - val_accuracy: 0.8745\n",
      "Epoch 11/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0513 - accuracy: 0.9859 - val_loss: 0.4013 - val_accuracy: 0.8788\n",
      "Epoch 12/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0403 - accuracy: 0.9901 - val_loss: 0.4378 - val_accuracy: 0.8721\n",
      "Epoch 13/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0326 - accuracy: 0.9928 - val_loss: 0.4906 - val_accuracy: 0.8664\n",
      "Epoch 14/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0258 - accuracy: 0.9950 - val_loss: 0.5596 - val_accuracy: 0.8595\n",
      "Epoch 15/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0210 - accuracy: 0.9966 - val_loss: 0.5257 - val_accuracy: 0.8711\n",
      "Epoch 16/20\n",
      "30/30 [==============================] - 0s 8ms/step - loss: 0.0185 - accuracy: 0.9969 - val_loss: 0.5583 - val_accuracy: 0.8689\n",
      "Epoch 17/20\n",
      "30/30 [==============================] - 0s 8ms/step - loss: 0.0137 - accuracy: 0.9983 - val_loss: 0.5982 - val_accuracy: 0.8680\n",
      "Epoch 18/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0090 - accuracy: 0.9993 - val_loss: 0.6416 - val_accuracy: 0.8674\n",
      "Epoch 19/20\n",
      "30/30 [==============================] - 0s 8ms/step - loss: 0.0085 - accuracy: 0.9991 - val_loss: 0.6796 - val_accuracy: 0.8672\n",
      "Epoch 20/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0078 - accuracy: 0.9986 - val_loss: 0.7180 - val_accuracy: 0.8650\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential([layers.Dense(16, activation=\"relu\"),\n",
    "                          layers.Dense(16, activation=\"relu\"),\n",
    "                          layers.Dense(1, activation=\"sigmoid\")])\n",
    "model.compile(optimizer=\"rmsprop\",\n",
    "              loss=\"binary_crossentropy\",\n",
    "              metrics=[\"accuracy\"])\n",
    "history_original = model.fit(train_data, train_labels,\n",
    "                             epochs=20, batch_size=512, validation_split=0.4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "30/30 [==============================] - 1s 30ms/step - loss: 0.6109 - accuracy: 0.7335 - val_loss: 0.5297 - val_accuracy: 0.8287\n",
      "Epoch 2/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.4654 - accuracy: 0.8717 - val_loss: 0.4306 - val_accuracy: 0.8624\n",
      "Epoch 3/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.3697 - accuracy: 0.8981 - val_loss: 0.3660 - val_accuracy: 0.8812\n",
      "Epoch 4/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.3029 - accuracy: 0.9149 - val_loss: 0.3245 - val_accuracy: 0.8880\n",
      "Epoch 5/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2548 - accuracy: 0.9253 - val_loss: 0.3016 - val_accuracy: 0.8893\n",
      "Epoch 6/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2195 - accuracy: 0.9335 - val_loss: 0.2972 - val_accuracy: 0.8810\n",
      "Epoch 7/20\n",
      "30/30 [==============================] - 0s 5ms/step - loss: 0.1921 - accuracy: 0.9425 - val_loss: 0.2768 - val_accuracy: 0.8941\n",
      "Epoch 8/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.1697 - accuracy: 0.9490 - val_loss: 0.2721 - val_accuracy: 0.8940\n",
      "Epoch 9/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.1508 - accuracy: 0.9556 - val_loss: 0.2820 - val_accuracy: 0.8892\n",
      "Epoch 10/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.1357 - accuracy: 0.9598 - val_loss: 0.2792 - val_accuracy: 0.8884\n",
      "Epoch 11/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.1218 - accuracy: 0.9642 - val_loss: 0.2808 - val_accuracy: 0.8911\n",
      "Epoch 12/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.1097 - accuracy: 0.9687 - val_loss: 0.2895 - val_accuracy: 0.8887\n",
      "Epoch 13/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0990 - accuracy: 0.9733 - val_loss: 0.2969 - val_accuracy: 0.8865\n",
      "Epoch 14/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0898 - accuracy: 0.9764 - val_loss: 0.3114 - val_accuracy: 0.8855\n",
      "Epoch 15/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0804 - accuracy: 0.9796 - val_loss: 0.3280 - val_accuracy: 0.8831\n",
      "Epoch 16/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0724 - accuracy: 0.9823 - val_loss: 0.3311 - val_accuracy: 0.8842\n",
      "Epoch 17/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0656 - accuracy: 0.9849 - val_loss: 0.3456 - val_accuracy: 0.8826\n",
      "Epoch 18/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0587 - accuracy: 0.9867 - val_loss: 0.3643 - val_accuracy: 0.8810\n",
      "Epoch 19/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0524 - accuracy: 0.9880 - val_loss: 0.3780 - val_accuracy: 0.8799\n",
      "Epoch 20/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0471 - accuracy: 0.9899 - val_loss: 0.3916 - val_accuracy: 0.8775\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential([layers.Dense(4, activation=\"relu\"),\n",
    "                          layers.Dense(4, activation=\"relu\"),\n",
    "                          layers.Dense(1, activation=\"sigmoid\")])\n",
    "model.compile(optimizer=\"rmsprop\",\n",
    "              loss=\"binary_crossentropy\",\n",
    "              metrics=[\"accuracy\"])\n",
    "history_smaller_model = model.fit(train_data, train_labels,\n",
    "                                  epochs=20, batch_size=512, validation_split=0.4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2272248df60>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "original_model = history_original.history[\"val_loss\"]\n",
    "smaller_model = history_smaller_model.history[\"val_loss\"]\n",
    "epochs = range(1,21)\n",
    "plt.plot(epochs, original_model, \"b--\", label=\"Original Model\")\n",
    "plt.plot(epochs, smaller_model, \"b-\", label=\"Smaller Model\")\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Smaller model tend to start overfitting after a while where the original model starts sooner."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "30/30 [==============================] - 3s 98ms/step - loss: 0.5536 - accuracy: 0.7571 - val_loss: 0.3469 - val_accuracy: 0.8642\n",
      "Epoch 2/20\n",
      "30/30 [==============================] - 3s 93ms/step - loss: 0.2456 - accuracy: 0.9043 - val_loss: 0.2806 - val_accuracy: 0.8885\n",
      "Epoch 3/20\n",
      "30/30 [==============================] - 3s 93ms/step - loss: 0.1784 - accuracy: 0.9311 - val_loss: 0.2732 - val_accuracy: 0.8933\n",
      "Epoch 4/20\n",
      "30/30 [==============================] - 3s 94ms/step - loss: 0.0868 - accuracy: 0.9717 - val_loss: 0.3454 - val_accuracy: 0.8931\n",
      "Epoch 5/20\n",
      "30/30 [==============================] - 3s 85ms/step - loss: 0.1187 - accuracy: 0.9751 - val_loss: 0.3369 - val_accuracy: 0.8881\n",
      "Epoch 6/20\n",
      "30/30 [==============================] - 2s 79ms/step - loss: 0.0057 - accuracy: 0.9993 - val_loss: 0.4934 - val_accuracy: 0.8860\n",
      "Epoch 7/20\n",
      "30/30 [==============================] - 2s 81ms/step - loss: 7.8860e-04 - accuracy: 0.9999 - val_loss: 0.6145 - val_accuracy: 0.8883\n",
      "Epoch 8/20\n",
      "30/30 [==============================] - 2s 81ms/step - loss: 1.1695e-04 - accuracy: 1.0000 - val_loss: 0.7275 - val_accuracy: 0.8866\n",
      "Epoch 9/20\n",
      "30/30 [==============================] - 3s 98ms/step - loss: 0.1015 - accuracy: 0.9933 - val_loss: 0.7276 - val_accuracy: 0.8835\n",
      "Epoch 10/20\n",
      "30/30 [==============================] - 3s 97ms/step - loss: 0.0013 - accuracy: 0.9999 - val_loss: 0.7269 - val_accuracy: 0.8870\n",
      "Epoch 11/20\n",
      "30/30 [==============================] - 3s 96ms/step - loss: 1.7129e-04 - accuracy: 0.9999 - val_loss: 0.7552 - val_accuracy: 0.8856\n",
      "Epoch 12/20\n",
      "30/30 [==============================] - 2s 83ms/step - loss: 1.5519e-05 - accuracy: 1.0000 - val_loss: 0.8023 - val_accuracy: 0.8861\n",
      "Epoch 13/20\n",
      "30/30 [==============================] - 2s 79ms/step - loss: 9.1358e-06 - accuracy: 1.0000 - val_loss: 0.8617 - val_accuracy: 0.8863\n",
      "Epoch 14/20\n",
      "30/30 [==============================] - 2s 80ms/step - loss: 4.4239e-06 - accuracy: 1.0000 - val_loss: 0.9370 - val_accuracy: 0.8859\n",
      "Epoch 15/20\n",
      "30/30 [==============================] - 2s 80ms/step - loss: 1.5878e-06 - accuracy: 1.0000 - val_loss: 1.0284 - val_accuracy: 0.8865\n",
      "Epoch 16/20\n",
      "30/30 [==============================] - 3s 95ms/step - loss: 5.3356e-07 - accuracy: 1.0000 - val_loss: 1.1092 - val_accuracy: 0.8859\n",
      "Epoch 17/20\n",
      "30/30 [==============================] - 2s 83ms/step - loss: 1.8555e-07 - accuracy: 1.0000 - val_loss: 1.1786 - val_accuracy: 0.8858\n",
      "Epoch 18/20\n",
      "30/30 [==============================] - 3s 88ms/step - loss: 7.8405e-08 - accuracy: 1.0000 - val_loss: 1.2327 - val_accuracy: 0.8851\n",
      "Epoch 19/20\n",
      "30/30 [==============================] - 3s 85ms/step - loss: 3.7210e-08 - accuracy: 1.0000 - val_loss: 1.2717 - val_accuracy: 0.8853\n",
      "Epoch 20/20\n",
      "30/30 [==============================] - 2s 82ms/step - loss: 2.1995e-08 - accuracy: 1.0000 - val_loss: 1.3010 - val_accuracy: 0.8855\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential([layers.Dense(512, activation=\"relu\"),\n",
    "                          layers.Dense(512, activation=\"relu\"),\n",
    "                          layers.Dense(1, activation=\"sigmoid\")])\n",
    "model.compile(optimizer=\"rmsprop\",\n",
    "              loss=\"binary_crossentropy\",\n",
    "              metrics=[\"accuracy\"])\n",
    "history_larger_model = model.fit(train_data, train_labels,\n",
    "                                 epochs=20, batch_size=512, validation_split=0.4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2272257f820>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiMAAAGdCAYAAADAAnMpAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAABWP0lEQVR4nO3deXhMZ/sH8O8kkURIgliSEIRaayuKUBS1U9uvotRSe1XVWlQtVX1ttVTVVktQtS+vqiJ2ag2JailKSBBiTWLLNs/vj/udRGSdSObMTL6f65rLOWfOTO5jkpw7z3I/OqWUAhEREZFGbLQOgIiIiHI2JiNERESkKSYjREREpCkmI0RERKQpJiNERESkKSYjREREpCkmI0RERKQpJiNERESkKTutA8gIvV6P27dvw9nZGTqdTutwiIiIKAOUUoiKioKnpydsbFJv/7CIZOT27dvw8vLSOgwiIiLKhNDQUBQrVizV5y0iGXF2dgYgF+Pi4qJxNERERJQRkZGR8PLySriPp8YikhFD14yLiwuTESIiIguT3hALDmAlIiIiTTEZISIiIk0xGSEiIiJNWcSYkYxQSiEuLg7x8fFah0JWIFeuXLC1tdU6DCKiHMEqkpGYmBiEhYXh2bNnWodCVkKn06FYsWLImzev1qEQEVk9i09G9Ho9goODYWtrC09PT9jb27MwGr0WpRTu3buHmzdvokyZMmwhISLKZhafjMTExECv18PLywtOTk5ah0NWolChQrh+/TpiY2OZjBARZTOrGcCaVplZImOxdY2IyHR4ByciIiJNMRmxUNevX4dOp0NQUFCGX+Pn54d8+fJpHkd2yMy16XQ6bNu2LVviISKijGMyoqHQ0FD06dMnYeBtiRIl8Pnnn+PBgwfpvtbLywthYWGoVKlShr+er68vLl++/DohZ8q7774LnU6HadOmJXuuVatW0Ol0mDRpksnjIiIi88BkRCPXrl1DzZo1cfnyZaxduxb//vsvFi1ahH379sHHxwcPHz5M9bUxMTGwtbWFu7s77OwyPgY5d+7cKFy4cFaEbzQvLy+sWLEiybHbt29j//798PDw0CQmIiIyD0xGNPLpp5/C3t4ee/bsQcOGDVG8eHG0bNkSe/fuxa1btzBu3LiEc0uWLIkpU6agV69ecHV1Rb9+/VLsHtm+fTvKlCmD3Llzo1GjRli5ciV0Oh0eP34MIHlXxqRJk1CtWjWsXr0aJUuWhKurK7p06YKoqKiEc3bt2oV33nkH+fLlg5ubG9q0aYOrV68afb1t2rTBgwcP8McffyQc8/PzQ7NmzZIlSI8ePUKPHj2QP39+ODk5oWXLlrhy5UqSc/z8/FC8eHE4OTmhQ4cOKbYm/frrr6hRowYcHR1RqlQpfP3114iLizM6diIiaxUbC2zaBLRuDTx6pF0cVpuMPH2a+uPFi4yf+/x5xs41xsOHD7F7924MGjQIuXPnTvKcu7s7unXrhvXr10MplXB85syZqFSpEs6cOYPx48cne8/r16/j//7v/9C+fXsEBQVhwIABSRKa1Fy9ehXbtm3Djh07sGPHDhw6dChJd8rTp08xfPhwnD59Gvv27YONjQ06dOgAvV5v1DXb29ujW7duSVpH/Pz80Lt372Tn9urVCwEBAdi+fTuOHz8OpRRatWqF2NhYAMDJkyfRu3dvDBo0CEFBQWjUqBGmTJmS5D12796Njz76CEOGDMGFCxewePFi+Pn54dtvvzUqbiIia3T9OjBuHFC8OPDBB8DOncDq1RoGpCxARESEAqAiIiKSPff8+XN14cIF9fz58yTHgdQfrVolfQ8np9TPbdgw6bkFC6Z8njFOnDihAKitW7em+Pzs2bMVAHX37l2llFIlSpRQ7du3T3JOcHCwAqACAwOVUkqNHj1aVapUKck548aNUwDUo0ePlFJKrVixQrm6uiY8P3HiROXk5KQiIyMTjo0aNUrVrl071djDw8MVAHX+/PkU40hJw4YN1eeff67OnTunnJ2d1ZMnT9ShQ4dU4cKFVUxMjKpataqaOHGiUkqpy5cvKwDqjz/+SHj9/fv3Ve7cudWGDRuUUkp9+OGHqkWLFkm+hq+vb5Jrq1+/vvrPf/6T5JzVq1crDw+PhP20PoPUvq+IiCxVbKxS//2vUi1bKqXTJd6/ihRR6ssvlQoOzvqvmdb9+2UWX/TMGqn/tYi8XOuiZs2aab7m0qVLePvtt5Mcq1WrVrpfq2TJknB2dk7Y9/DwQHh4eML+1atXMX78eJw4cQL3799PaBEJCQkxavAsAFSpUgVlypTBpk2bcODAAXTv3h25cuVKcs7FixdhZ2eH2rVrJxxzc3NDuXLlcPHixYRzOnTokOR1Pj4+2LVrV8L+mTNncPr06SQtIfHx8Xjx4gWePXvGAnlElGPcugUsXSqPmzcTjzdpAgwcCLz/PmBvr118gBVUYE3NkyepP/dqQc2X7r3JvFpL7fr1TIeU4I033oBOp8OFCxfQvn37ZM//888/yJ8/PwoWLJhwLE+ePGm+p1IqWaEu9VI3T2peTQZ0Ol2SLpi2bdvCy8sLP/30Ezw9PaHX61GpUiXExMSk+94p6d27N3788UdcuHABp06dSvE6UvLy9WXkuvR6Pb7++mt07Ngx2XOOjo5GRk1EZFn0emDPHmDRImDHDsCwhqybG/Dxx0D//kCZMtrG+DKrTUbSuXeb5NzUuLm5oWnTpliwYAGGDRuWZNzInTt3sGbNGvTo0cOoKqDly5fHzp07kxwLCAh4rTgfPHiAixcvYvHixahfvz4A4OjRo6/1nl27dsXIkSNRtWpVVKxYMdnzFStWRFxcHE6ePIm6desmxHH58mVUqFAh4ZwTJ04ked2r+9WrV8elS5fwxhtvvFa8RESW5O5dYPlyYMmSpH88N2gADBgAdOwImOPfY1abjJi7+fPno27dumjevDmmTJkCb29v/P333xg1ahSKFi1q9EDLAQMGYPbs2Rg9ejT69OmDoKAg+Pn5Ach8afP8+fPDzc0NS5YsgYeHB0JCQjBmzJhMvdfL7xkWFpasRcagTJkyaNeuHfr164fFixfD2dkZY8aMQdGiRdGuXTsAwJAhQ1C3bl3MmDED7du3x549e5J00QDAhAkT0KZNG3h5eeGDDz6AjY0N/vzzT5w/fz7ZYFciIkumFHDggLSCbN0KGCYN5ssH9OwprSAp/O1nVqx2No25K1OmDAICAlC6dGn4+vqidOnS6N+/Pxo1aoTjx4+jQIECRr2ft7c3Nm3ahC1btqBKlSpYuHBhwmwaBweHTMVoY2ODdevW4cyZM6hUqRKGDRuGmTNnZuq9XpYvX740u51WrFiBGjVqoE2bNvDx8YFSCjt37kxIYOrUqYOlS5fihx9+QLVq1bBnzx589dVXSd6jefPm2LFjB/z9/fH222+jTp06mD17NkqUKPHa8RMRmYP794FZs4By5WT8x8aNkojUqQP4+clYkblzzT8RAQCdykgHvMYiIyPh6uqKiIgIuLi4JHnuxYsXCA4Ohre3N8cCvOLbb7/FokWLEBoaqnUoFoffV0RkjpQCjh4FFi+W5MMwfM/ZGfjoI+mKqVpV2xhfltb9+2XsprEiCxYswNtvvw03Nzf88ccfmDlzJgYPHqx1WERE9JoePwZWrZIk5MKFxOPVq8uMmA8/BPLm1Sy818ZkxIpcuXIFU6ZMwcOHD1G8eHGMGDECY8eO1TosIiLKpIcPgWnTgPnzE4twOjkBXbtKK0g6VR8sBpMRKzJnzhzMmTNH6zCIiOg1PX8OzJsnicj/VvRA5cqSgHz0EeDqqml4WY7JCBERkZmIiwNWrAAmTQJu35ZjlSsDU6cCrVoBmZwcafaYjBAREWlMKZmW++WXwKVLcqxECWDyZKBbt+TFOq0NkxEiIiINHToEjB4NnDwp+25uwFdfAZ98AmSyMoPFYTJCRESkgXPngLFjgd9/l30nJ2D4cGDkSOsbE5IeJiNEREQmFBwMTJgArFkj3TN2dkC/fnLM3V3r6LTBZISIiMgE7t0DpkwBFi4EYmPlmK+vHMvpy2ixHDyZvZIlS2Lu3LkZPn/SpEmoVq1atsVDRGSMJ09kIGrp0jJdNzYWeO89ICAAWLeOiQjAZEQzvXr1Qvv27bUO47UcPHgQOp0O+fPnx4sXL5I8d+rUKeh0ukwv0kdEZOliYqRYWenSwMSJQFSUVEzdswfw9wdq1NA6QvPBZMRKxMfHQ6/XZ8t7xxraE1Ph7OyMrVu3Jjm2fPlyFC9ePFviISIyZ3o9sHYtUKEC8NlnQHi4JCTr1gGnTwNNm2odoflhMmKmZs+ejcqVKyNPnjzw8vLCoEGD8OTJk4Tn/fz8kC9fPuzYsQMVK1aEg4MDbty4gbCwMLRu3Rq5c+eGt7c3fvnll2TdHBEREejfvz8KFy4MFxcXNG7cGOfOnUt43tDNsXz5cpQqVQoODg5Iaz3Fnj17Yvny5Qn7z58/x7p169CzZ89k527evBlvvvkmHBwcULJkScyaNSvJ8+Hh4Wjbtm1C/GvWrEn2HunFT0SkBaWk1aNmTSnXfu0aUKQI8OOPwMWLMj7EhnfdFFndAFalgGfPtPnaTk5ZVx3PxsYG8+bNQ8mSJREcHIxBgwbhiy++wIIFCxLOefbsGaZOnYqlS5fCzc0NhQsXRvv27XH//n0cPHgQuXLlwvDhwxEeHp7wGqUUWrdujQIFCmDnzp1wdXXF4sWL0aRJE1y+fBkFChQAAPz777/YsGEDNm/eDNt0qu10794dM2fOREhICIoXL47NmzejZMmSqF69epLzzpw5g86dO2PSpEnw9fXFsWPHMGjQILi5uaFXr14ApPsqNDQU+/fvh729PYYMGZKp+ImITCkgQGqF7N8v+87OwBdfAEOHWvYCdiajLEBERIQCoCIiIpI99/z5c3XhwgX1/PlzpZRST54oJSmJ6R9PnmT8mnr27KnatWuX4fM3bNig3NzcEvZXrFihAKigoKCEYxcvXlQA1OnTpxOOXblyRQFQc+bMUUoptW/fPuXi4qJevHiR5P1Lly6tFi9erJRSauLEiSpXrlwqPDw8zZgOHDigAKhHjx6p9u3bq6+//loppVSjRo3U999/r7Zu3ape/hbr2rWratq0aZL3GDVqlKpYsaJSSqlLly4pAOrEiRPJrsnY+KtWrZpm7Ol59fuKiCglcXFKTZiglE4n9wF7e6WGDlUqnV+fOUZa9++XscHITB04cABNmzZF0aJF4ezsjB49euDBgwd4+vRpwjn29vaoUqVKwv6lS5dgZ2eXpEXijTfeQP78+RP2z5w5gydPnsDNzQ158+ZNeAQHB+Pq1asJ55UoUQKFChXKcLy9e/eGn58frl27huPHj6Nbt27Jzrl48SLq1auX5Fi9evVw5coVxMfH4+LFi7Czs0PNl5ahLF++PPLly2d0/ERE2S0sTGbFTJ4sf5J++KGUcp8zBzDi1yfBCrtpnJxkGpVWXzsr3LhxA61atcLAgQPxzTffoECBAjh69Cj69OmTZDBp7ty5k8xWUamM63j5uF6vh4eHBw4ePJjsvJdv+nny5DEq5latWmHAgAHo06cP2rZtCzc3txTjeHV2zcuxGbbTmoGT0fiJiLKTv7+snhseDuTJAyxeLGvIUOZYXTKi08k3hiULCAhAXFwcZs2aBZv/jXbasGFDuq8rX7484uLiEBgYiBr/mzP277//4rFh/WkA1atXx507d2BnZ4eSJUtmWcy2trbo3r07ZsyYgd8NtY1fUbFiRRw9ejTJsWPHjqFs2bKwtbVFhQoVEBcXh4CAANSqVQuAtPaYIn4iooyIi5MVdf/zH2kNqVIF2LABKFdO68gsG7tpNBQREYGgoKAkj5CQEJQuXRpxcXH44YcfcO3aNaxevRqLFi1K9/3Kly+P9957D/3798epU6cQGBiI/v37J2lBee+99+Dj44P27dtj9+7duH79Oo4dO4avvvoKAQEBr3U933zzDe7du4fmzZun+PyIESOwb98+fPPNN7h8+TJWrlyJ+fPnY+TIkQCAcuXKoUWLFujXrx9OnjyJM2fOoG/fvsidO3fCe2Rn/EREabl1C2jcGPj2W0lEBgwATpxgIpIVmIxo6ODBg3jrrbeSPCZMmIBq1aph9uzZmD59OipVqoQ1a9Zg6tSpGXrPVatWoUiRImjQoAE6dOiAfv36wdnZGY6OjgCkC2Tnzp1o0KABevfujbJly6JLly64fv06ihQp8lrXY29vj4IFC6bazVK9enVs2LAB69atQ6VKlTBhwgRMnjw5YSYNAKxYsQJeXl5o2LAhOnbsmDCF1yA74yciSs2uXUC1asCRIzJTZu1aYNEi4KW/leg16FRqAw3MSGRkJFxdXREREQEXF5ckz7148QLBwcHw9vZOuOFSops3b8LLywt79+5FkyZNtA7HYvD7iogAKd0+fjwwfbrsv/UWsH49UKaMtnFZirTu3y8zumXk8OHDaNu2LTw9PaHT6bBt27Y0z9+yZQuaNm2KQoUKwcXFBT4+Pti9e7exX5YyaP/+/di+fTuCg4Nx7NgxdOnSBSVLlkSDBg20Do2IyKKEhgLvvpuYiHz6KXDsGBOR7GB0MvL06VNUrVoV8+fPz9D5hw8fRtOmTbFz506cOXMGjRo1Qtu2bREYGGh0sJS+2NhYfPnll3jzzTfRoUMHFCpUKKEAGhERZcyOHdItc+wY4OICbNwo68ywoTR7vFY3jU6nw9atW41e8O3NN9+Er68vJkyYkKHz2U1DpsbvK6KcKTYWGDsWMKxUUaOGdMuULq1tXJYqo900Jp/aq9frERUVxbLdRERkVm7ckPVjTp6U/SFDgBkzAAcHbePKCUyejMyaNQtPnz5F586dUz0nOjoa0dHRCfuRkZGmCI2IiHKo//4X6NULePwYyJcPWL4c6NBB46ByEJNO7V27di0mTZqE9evXJ5mu+aqpU6fC1dU14eHl5ZXue1vApCCyIPx+IsoZYmJkMbv27SURqVULCAxkImJqJktG1q9fjz59+mDDhg1477330jx37NixiIiISHiEhoameq5hYOYzrZbqJasUExMDAOmuWExElis4GHjnHeD772V/+HCpI8LizqZnkm6atWvXonfv3li7di1at26d7vkODg5wyGAnna2tLfLly5ewzLyTk1Oaa5sQpUev1+PevXtwcnKCnZ3VrZhARAC2bAF69wYiIoD8+YGVK4G2bbWOKucy+jftkydP8O+//ybsBwcHIygoCAUKFEDx4sUxduxY3Lp1C6tWrQIgiUiPHj3w/fffo06dOrhz5w4AWeTN1dU1Sy7C3d0dABISEqLXZWNjg+LFizOxJbIy0dHAyJEyTRcAfHyAdeuA4sW1jSunM3pq78GDB9GoUaNkx3v27Ak/Pz/06tUL169fT1hV9d1338WhQ4dSPT8jMjo1KD4+PsmqtkSZZW9vn7BIIRFZh3//ldkyZ8/K/hdfAFOmACzDlH0yev+2+HLwRERE6dm8Gfj4YyAqCnBzA1atAlq10joq65dt5eCJiIgsyZIlwAcfSCLyzjtAUBATEXPDZISIiKzWnDnAgAGAUsAnnwAHDgDFimkdFb2KyQgREVkdpWQ8yPDhsv/FF8CPPwKcIGeemIwQEZFVUQr48ktg/HjZ//prYNo0gJPjzBdzRCIishp6vVRU/eEH2f/uO2DECE1DogxgMkJERFYhPl7GhyxbJvsLFwIDB2obE2UMkxEiIrJ4sbFAz57A2rWAjQ2wYgXQo4fWUVFGMRkhIiKLFh0txcz++18ZoPrLLzKVlywHkxEiIrJYz57JCrt79gAODsCmTUCbNlpHRcZiMkJERBYpKkoSj8OHAScnYPt2oEkTraOizGAyQkREFufRI6BFC+DUKcDFBdi5E6hXT+uoKLOYjBARkUUJDweaNQPOnQMKFJAumho1tI6KXgeTESIishi3bgHvvQf88w9QpAiwdy9QqZLWUdHrYjJCREQW4fp1GRNy7ZqsL7NvH1C2rNZRUVZgOXgiIjJ7ly8D9etLIlKqFHDkCBMRa8JkhIiIzNr580CDBsDNm0CFCpKIlCypdVSUlZiMEBGR2QoIAN59F7h7F6hWDTh0CPD01DoqympMRoiIyCwdPQo0bgw8fAjUrg3s3w8UKqR1VJQdmIwQEZHZ2bsXaN5cCps1aAD4+wP582sdFWUXJiNERGRWduyQyqrPnklC8vvvgLOz1lFRdmIyQkREZmPjRllrJjoaaN9eFr9zctI6KspuTEaIiMgsrFwJdOkCxMUBXbsCGzbI4ndk/ZiMEBGR5pYuBXr1AvR6oG9fYNUqIFcuraMiU2EyQkREmtq2DRgwQLaHDAGWLAFsbTUNiUyMyQgREWnm6FHgww8TW0TmzgV0Oq2jIlNjMkJERJr4+2+gbVvgxQv5d+FCJiI5FZMRIiIyuZs3gRYtgMePAR8fYN06wI5Lt+ZYTEaIiMikHj2SROTmTaB8eeDXXzl9N6djMkJERCbz/DnQrp100Xh6Art2AW5uWkdFWmMyQkREJhEfD3TrJqvuurpKIlKihNZRkTlgMkJERNlOKWDwYGDrVsDeXiqrVq6sdVRkLpiMEBFRtvv2W2DRIpkts2YN0LCh1hGROWEyQkRE2WrpUmD8eNmeNw/4v//TNh4yP0xGiIgo2/z6a2J11S+/lK4aolcxGSEiomxx/Djg6yvVVXv1AqZM0ToiMldMRoiIKMv98w/Qpo1M5W3VStabYXVVSg2TESIiylK3bwPNmwMPHwK1agEbNnAFXkobkxEiIsoyjx9LddWQEKBsWeC334A8ebSOiswdkxEiIsoSL14A7dsD588D7u5S1KxgQa2jIkvAZISIiF5bfDzQvTtw6BDg7Az8/jvg7a11VGQpmIwQEdFrUQoYOhTYtEnGhmzbBlSrpnFQZFGYjBAR0WuZPh2YP1+2V68GGjfWNh6yPExGiIgo0/z8gLFjZXvuXKkrQmQsJiNERJQpv/8O9O0r2198AXz+ubbxkOViMkJEREY7dUrWmDEMXJ06VeuIyJIxGSEiIqNcvgy0bg08eybFzZYtA2x4N6HXwG8fIiLKsDt3JAG5fx+oWTNxBg3R62AyQkREGRIZCbRsCVy/DpQuLdVV8+bVOiqyBkxGiIgoXdHRQMeOQFAQULgwsHu3/EuUFZiMEBFRuj77DNi3T1pCdu6UlhGirMJkhIiI0rRkCfDTT4BOJyvw1qihdURkbZiMEBFRqk6cAAYPlu1vv5UxI0RZjckIERGl6M4doFMnIDZWxouMGaN1RGStmIwQEVEyMTHABx8At28DFSpI2XedTuuoyFoxGSEiomRGjACOHgVcXGQVXmdnrSMia2Z0MnL48GG0bdsWnp6e0Ol02LZtW7qvOXToEGrUqAFHR0eUKlUKixYtykysRERkAitXJq7C+/PPQNmy2sZD1s/oZOTp06eoWrUq5hu+U9MRHByMVq1aoX79+ggMDMSXX36JIUOGYPPmzUYHS0RE2evMGWDAANmeOBFo21bbeChnsDP2BS1btkRLI4ZTL1q0CMWLF8fcuXMBABUqVEBAQAC+++47dOrUydgvT0RE2eTePaBDBylw1qYNMGGC1hFRTpHtY0aOHz+OZs2aJTnWvHlzBAQEIDY2NsXXREdHIzIyMsmDiIiyT1wc4OsLhIYCZcpI9wwXvyNTyfZvtTt37qBIkSJJjhUpUgRxcXG4f/9+iq+ZOnUqXF1dEx5eXl7ZHSYRUY42ejRw4IBUWN22DXB11ToiyklMkvfqXpkPppRK8bjB2LFjERERkfAIDQ3N9hiJiHKqX34BZs+WbT8/oGJFTcOhHMjoMSPGcnd3x507d5IcCw8Ph52dHdzc3FJ8jYODAxwcHLI7NCKiHO/cOaBvX9keO1aKnBGZWra3jPj4+MDf3z/JsT179qBmzZrIlStXdn95IiJKxcOHMmD1+XOgeXPgm2+0johyKqOTkSdPniAoKAhBQUEAZOpuUFAQQkJCAEgXS48ePRLOHzhwIG7cuIHhw4fj4sWLWL58OZYtW4aRI0dmzRUQEZHR4uOBDz8EgoOBUqWkq8bWVuuoKKcyupsmICAAjRo1StgfPnw4AKBnz57w8/NDWFhYQmICAN7e3ti5cyeGDRuGH3/8EZ6enpg3bx6n9RIRaeirr4A9ewAnJ2DrVqBAAa0jopxMpwyjSc1YZGQkXF1dERERARcXF63DISKyaJs2ybozALB2LdCli7bxkPXK6P2bs8iJiHKQv/8GevWS7REjmIiQeWAyQkSUQzx+LANWnz4FGjcGpk3TOiIiwWSEiCgH0OuB7t2BK1eA4sWBdesAu2wv7kCUMUxGiIhygMmTgR07AAcHYMsWoFAhrSMiSsRkhIjIym3fDnz9tWwvWQLUqKFtPESvYjJCRGTFLl0CPvpItgcPBl4qA0VkNpiMEBFZqchIoH17ICoKqF8/cf0ZInPDZISIyArp9TKF959/AE9PYMMGgCtwkLliMkJEZIWmTZPKqvb2wObNgLu71hERpY7JCBGRlfn9dyn3DgDz5wN16mgbD1F6mIwQEVmRq1eBrl0BpYD+/YF+/bSOiCh9TEaIiKzE06dSYfXxY2kNmTdP64iIMobJCBGRFdDrgT59gPPngSJFZDE8BwetoyLKGCYjREQWTing88+B9eulxPumTUDRolpHRZRxTEaIiCyYUsDo0TJQVacDli8H3nlH66iIjMNkhIjIgn39NTBzpmwvWiSL4RFZGiYjREQWavr0xDVn5s6V2TNElojJCBGRBZo3DxgzRranTpUxI0SWiskIEZGFWbIkMfmYMCExKSGyVExGiIgsyOrVwMCBsj1yJDBpkqbhEGUJJiNERBZi40ZZ/E4p4NNPgRkzZAYNkaVjMkJEZAF+/VXKvBuKm82bx0SErAeTESIiM7dnD/B//wfExUlCsngxYMPf3mRF+O1MRGTGDh0C2rcHYmKATp2AlSsBW1utoyLKWkxGiIjM1PHjQJs2wPPnQOvWwC+/SLl3ImvDZISIyAydPQu0bAk8eQK8956sN2Nvr3VURNmDyQgRkZn56y+gaVMgIgKoXx/Ytg1wdNQ6KqLsw2SEiMiMXLoENGkCPHwI1KoF7NgB5MmjdVRE2YvJCBGRmbh2TRKR8HCgWjVg1y7AxUXrqIiyH5MRIiIzEBoqicitW0DFijKdN39+raMiMg0mI0REGgsLAxo3Bq5fB8qUAfbuBQoV0joqItNhMkJEpKF792S2zL//AiVLAvv2AR4eWkdFZFpMRoiINPLoEdCsGXDhAlC0qCQiXl5aR0VkekxGiIg0EBkJtGgBBAUBRYpIIlKqlNZREWmDyQgRkYk9fSoVVU+dAtzcZIxIuXJaR0WkHSYjREQm9OIF0K4dcPQo4Ooqs2YqVdI6KiJtMRkhIjKRmBhZfXffPiBvXqkjUr261lERaY/JCBGRCURFyaq7v/0G5M4tlVXr1NE6KiLzwPUfiYiy2bVr0jXz11+Ag4OsNdOwodZREZkPtowQEWWjAweAt9+WRMTdHTh4UKbzElEiJiNERNlAKeDHH2X13YcPgZo1gYAAds0QpYTJCBFRFouJAQYMAAYPBuLjgW7dgMOHpbAZESXHMSNERFkoPFwGqh49Cuh0wPTpwMiRsk1EKWMyQkSURYKCZKBqSAjg4gKsXQu0aqV1VETmj900RERZYONGoG5dSUTKlgVOnmQiQpRRTEaIiF6DXg+MHw907gw8fw40by6JSPnyWkdGZDnYTUNElElRUUD37sB//yv7I0bIGBFbW23jIrI0TEaIiDLh2jXg/feBv/+WQmZLlgA9emgdFZFlYjJCRGSk/fuBDz6Q+iEeHsDWrUDt2lpHRWS5OGaEiCiDlALmz5cKqg8fSmXVgAAmIkSvi8kIEVEGxMQA/fsDn30mhcw++gg4dAjw9NQ6MiLLx24aIqJ0vFzIzMZGBqmOGMFCZkRZhckIEVEaAgOlkFloKODqKoXMWrbUOioi68JuGiKiVGzYANSrJ4mIoZAZExGirJepZGTBggXw9vaGo6MjatSogSNHjqR5/po1a1C1alU4OTnBw8MDH3/8MR48eJCpgImIspteD3z1FeDrK4XMWrSQRKRcOa0jI7JORicj69evx9ChQzFu3DgEBgaifv36aNmyJUJCQlI8/+jRo+jRowf69OmDv//+Gxs3bsTp06fRt2/f1w6eiCirRUUBHTsC334r+yNHAjt2APnyaRoWkVUzOhmZPXs2+vTpg759+6JChQqYO3cuvLy8sHDhwhTPP3HiBEqWLIkhQ4bA29sb77zzDgYMGICAgIDXDp6IKCtdugT4+EhFVQcHYNUqYOZMVlQlym5GJSMxMTE4c+YMmjVrluR4s2bNcOzYsRRfU7duXdy8eRM7d+6EUgp3797Fpk2b0Lp161S/TnR0NCIjI5M8iIiyS2wsMG0aULWqVFT18AAOH5ZS70SU/YxKRu7fv4/4+HgUKVIkyfEiRYrgzp07Kb6mbt26WLNmDXx9fWFvbw93d3fky5cPP/zwQ6pfZ+rUqXB1dU14eHl5GRMmEVGGBQZK0bKxY4HoaBkfEhAA1KqldWREOUemBrDqXplcr5RKdszgwoULGDJkCCZMmIAzZ85g165dCA4OxsCBA1N9/7FjxyIiIiLhERoampkwiYhS9eIF8OWXUkU1MBAoUEC6ZXbuZCEzIlMzqs5IwYIFYWtrm6wVJDw8PFlricHUqVNRr149jBo1CgBQpUoV5MmTB/Xr18eUKVPg4eGR7DUODg5wcHAwJjQiogw7ehTo0we4fFn2O3cG5s0DUvk1RkTZzKiWEXt7e9SoUQP+/v5Jjvv7+6Nu3bopvubZs2ewsUn6ZWz/NxpMKWXMlyciei1RUcDgwUD9+pKIGBa5W7+eiQiRloyuwDp8+HB0794dNWvWhI+PD5YsWYKQkJCEbpexY8fi1q1bWLVqFQCgbdu26NevHxYuXIjmzZsjLCwMQ4cORa1ateDJtlAiMpHffwcGDJACZgDQt6/MlOGUXSLtGZ2M+Pr64sGDB5g8eTLCwsJQqVIl7Ny5EyVKlAAAhIWFJak50qtXL0RFRWH+/PkYMWIE8uXLh8aNG2P69OlZdxVERKm4fx8YNgz4+WfZL1UKWLIEaNJE27iIKJFOWUBfSWRkJFxdXREREQEXFxetwyEiC6CUlHP/7DPg3j1Z4G7oUGDyZCBPHq2jI8oZMnr/5kJ5RGR1bt0CBg0Ctm+X/TffBJYtkym8RGR+uFAeEVkNpYCffgIqVpREJFcuYNIk4OxZJiJE5owtI0RkFf79F+jfHzhwQPZr15bWkDff1DYuIkofW0aIyKLFxQGzZgFVqkgi4uQEzJkD/PEHExEiS8GWESKyWOfPS/Gy06dlv0kTmSlTqpS2cRGRcdgyQkQWJzoamDABqF5dEhFXV+mS8fdnIkJkidgyQkQW5fhxaQ25eFH227cHfvyR68kQWTImI0Rk9p49AzZvBpYuBQ4flmOFC0sS0qkTkMo6nURkIZiMEJHZCgyUBGTNGiAiQo7Z2AA9esig1QIFtI2PiLIGkxEiMiuPHwNr10oScvZs4vGSJaV7plcvoFgxjYIjomzBZISINKcUcOSIJCAbNwIvXshxe3ugQwdZ1K5xY2kVISLrw2SEiDRz9y6wcqXMhLl8OfH4m28C/foBH30EuLlpFx8RmQaTESIyqbg4YPduSUB+/VX2ASBvXqBLF2kFqVWLg1KJchImI0RkEsHBwPLlwIoVspCdgY+PJCCdO0tCQkQ5D5MRIso20dHAtm0yFmTv3sTjbm4yI6ZPH5ZsJyImI0Rm6elTuZErJfuv/pvSsfT+fXXbxgawtZV/X95O6ZhOZ1y3yV9/SQKyejXw8KEc0+mApk2lFeT99wEHh4y/HxFZNyYjRGZm+XK5Yb+cOJiDVxOU1LYBGZhqUKwY0Ls38PHHMj2XiOhVTEaIzEhoKPD558YnIoZWi9T+ffWYUoBeL4+Mfi3D+YYBp2mxswPatZNumGbNJFEhIkoNkxEiM6EU8MknwJMnQN26wIEDiTfxtJKMrPi6hkQjPj7pv5k9Vrw4p+QSUcYxGSEyE+vXA7/9JoW+li6Vf01Bp5Okx9YWyJXLNF+TiOhlrGdIZAYePACGDJHtr74CKlTQNh4ism7XrwNHj2odRSImI0RmYPhw4N49oFIlYPRoraMhImt19y7w2WdA2bJA9+5ATIzWEQl20xBpbPduYNUq6S4xZfcMEeUcjx8D330HzJ0rpQMA4I03pFXWw0PLyASTESINPXkCDBgg20OGALVraxsPEVmXZ8+A+fOBadOAR4/kWK1awNSpsvikuWAyQqSh8eOBGzeAEiWAKVO0joaIrM2ZM4ldvxUrAt9+K9PuzW3tJyYjRBo5eRL4/nvZXryY67IQ0evT66UCcpUqsl+/PtC/P1CvHtCtm/nW/GEyQqSBmJjEKqvduwPNm2sdERFZMqWAnTuBceOAK1eAq1cBd3d5bvFibWPLCM6mIdLAjBny10vBgsDs2VpHQ0SW7MgRoEEDoE0b4Nw5qYB87pzWURmHyQiRiV28CHzzjWx//70kJERExgoKAlq3lkTk6FHA0REYNQq4ds3yWlvZTUNkQno90K+fdNO0agV8+KHWERGRJXr4EPDxAV68kHEgffvKgPiiRbWOLHOYjBCZ0KJFwB9/yGDVhQvNb0Q7EZmvR4+A/Pllu0ABWcsqLAyYPBkoU0bb2F4Xu2mITCQ0FBgzRranTpXF5IiI0vPgAfDFF9Lqcfp04vFZs4C1ay0/EQGYjBCZhFLAoEFAVJQ0rX7yidYREZG5i4qS8WWlSgEzZwLPnwPr1iU+b00tq+ymITKBDRuAHTtkVdylS813rj8Rae/FC+nG/c9/gPv35VjVqlKwrFUrbWPLLkxGiLLZgweyMBUgNQAqVtQ2HiIyX0oB774rRREB6YL55hvggw8AGyvuy2AyQpTNRoyQFXnffBMYO1braIjI3Oj18q+NjXS99OoF3L4NTJwI9OwpdUOsnRXnWUTa8/cHVq6UXzA//cQVeYkokVLSffvWWzIQ1aBPH+DyZfk3JyQiAJMRomzz9KmsCQEAgwfLwFUiIgA4eFDWi2nbFvjzT6nErJQ8lyuXFDDLSZiMEGWTCROA69dlCu+332odDRGZg9OngWbNgEaNgOPHgdy5ZVVdf3/rmh1jrBzSAERkWqdPA3PnyvaiRYCzs6bhEJEZmDAhcSmIXLmk5XTcOMDDQ9u4zAFbRoiyWGys9PXq9bJkd8uWWkdEROagUSMZpNqzJ3DpEjB/PhMRA7aMEGWxGTOA8+cBNzdgzhytoyEiLdy+DUyZIsnG+PFyrFEj4OpVoGRJTUMzS0xGiLLQP//IOhGArMhbqJC28RCRaT14AEyfDvzwgxQvy5NH6gzlyyfPMxFJGbtpiLLIyyvytmwJdO2qdUREZCpRUfKHiLe3lG5/8UJmy/z2W2IiQqljywhRFlmyBDh6VP4S4oq8RDmHv7+MD7t3T/arVpVS7i1b8vdARjEZIcoCN2/KqpqA/BIqUULbeIjIdEqVAiIjc07p9uzA/y6i1/Tyiry1awOffqp1RESUnR4+BH7+OXG/dGlg/37g778BX18mIpnB/zKi17RxI/Drr1I3YNkyrshLZK30elnWoWxZoHt3KVpmULeu/A6gzGE3DdFrePgwcUXesWNlMTwisj6nT0ur5+nTsl+pEv/wyEpsGSF6DSNGAOHhQIUKwJdfah0NEWW1+/elUmrt2pKIuLhI/aCzZ4FatbSOznqwZYQok/buBfz8ZLT80qWAg4PWERFRVtLrgfr1pX4QIF0zM2YA7u7axmWN2DJClAkvr8j76afSX0xE1sXGRmbJVakCHDkCrFrFRCS7MBkhyoSJE4HgYMDLS6byEpHlCw8HevcGNmxIPNazJ3DmDPDOO9rFlRMwGSEy0tmziWvOLFzIFXmJLF1cnCxaV64csGKFjAWLiZHnbGwAOw5oyHaZSkYWLFgAb29vODo6okaNGjhy5Eia50dHR2PcuHEoUaIEHBwcULp0aSxfvjxTARNpSa+XmiJ6vdQTaN1a64iI6HX88QdQs6bMinv8GHjrLWkZsbfXOrKcxeh8b/369Rg6dCgWLFiAevXqYfHixWjZsiUuXLiA4sWLp/iazp074+7du1i2bBneeOMNhIeHIy4u7rWDJzI1Pz/g5Ekgb15g9mytoyGizLp7V8aDrFol+/nzA99+K2PBOGXX9HRKKWXMC2rXro3q1atj4cKFCccqVKiA9u3bY+rUqcnO37VrF7p06YJr166hQIECmQoyMjISrq6uiIiIgIuLS6beg+h1PXwozbj37wPffSdNuURkmfbvB5o0ke2+fWXsF1fZznoZvX8b1U0TExODM2fOoFmzZkmON2vWDMeOHUvxNdu3b0fNmjUxY8YMFC1aFGXLlsXIkSPx/PlzY740kebGj5dEpGJFYMgQraMhImOFhSVuN24MTJgAnDghVVWZiGjLqG6a+/fvIz4+HkWKFElyvEiRIrhz506Kr7l27RqOHj0KR0dHbN26Fffv38egQYPw8OHDVMeNREdHIzo6OmE/MjLSmDCJstzZszJYFZCBbiz7TGQ5rl+XooTbt0vNkGLF5PjXX2saFr0kUwNYda+siayUSnbMQK/XQ6fTYc2aNahVqxZatWqF2bNnw8/PL9XWkalTp8LV1TXh4eXllZkwibKEXi+1RJQCunQBGjXSOiIiyohHj4BRo6R7de1a4NkzYPduraOilBiVjBQsWBC2trbJWkHCw8OTtZYYeHh4oGjRonB1dU04VqFCBSilcPPmzRRfM3bsWERERCQ8QkNDjQmTKEv5+UlTbt68MlaEiMxbdLRMvy9dWn5mY2JkfMiZM0CfPlpHRykxKhmxt7dHjRo14O/vn+S4v78/6qZSgrJevXq4ffs2njx5knDs8uXLsLGxQTFDW9krHBwc4OLikuRBpIVHj4DRo2V74kSgaFFt4yGitMXFAdWrA8OHy89vpUrAzp2Av79M2yXzZHQ3zfDhw7F06VIsX74cFy9exLBhwxASEoKBAwcCkFaNHj16JJzftWtXuLm54eOPP8aFCxdw+PBhjBo1Cr1790bu3Lmz7kqIssFXXyUOWv38c62jIaL02NkBbdoAHh6yZlRQENCypawhRebL6GTE19cXc+fOxeTJk1GtWjUcPnwYO3fuRIkSJQAAYWFhCAkJSTg/b9688Pf3x+PHj1GzZk1069YNbdu2xbx587LuKoiywdmzwKJFss1Bq0Tm6dIloEMH6Uo1GD8euHJFumRYM8QyGF1nRAusM0KmptcD9erJL7guXWTwGxGZj7t3ZTbMkiVAfDzQsCFw8KDWUdGrMnr/ZsV9ohSsXMlBq0Tm6NkzqX48fTpgGIrYtq3sk+ViMkL0ikePpEw0wEGrROZk40Zg6FDg9m3Zr1lT/lho2FDTsCgLcNVeoldw0CqReYqMlESkZEngl19knSgmItaBLSNEL+GgVSLzERgIPHgAvPee7PfqJcUHu3cHHBw0DY2yGFtGiP7HUGlVr2elVSIthYQAPXoANWoAvXsDhmLdtrayqB0TEevDZITofzholSzdw4eJgzotUUQEMGYMULYssHq1tIK88w7w9KnWkVF2YzJCBA5aJcvn7w8ULy7FvkaOBG7d0jqijHv8GJg2Tcq3T58u5dwbNgROnZKxIQULah0hZTfWGSECMHgw8OOPQIUKwLlzHCtCluW//wU6d5Y1WAwcHYHQUMu4kQcGSgl3AChfXhKStm1ZNdUaZPT+zZYRyvECA4GFC2Wbg1bJ0qxdC3TqJIlIx47Ar78CDRpISfSXE5GrV7WL8VXHjwOLFyfuv/UW8Mknsijl+fPA++8zEclp2DJCOZpeL33Sx48Dvr7AunVaR0SUcXq9dGccPSozTJYvl7VZABn0aVj+6+pVGYfRoIGMyWjWzPQ3+/h4YNs2YNYs+XlzcABu3ABSWfCdrARbRogyYOVK+cWYJ4/8kiSyJDY20hIybZq0Kti9VKzh5XVI//hDzj14EGjRQrpE1q2TFW6z25MnwA8/SDL0f/8nP2/29kC3bqb5+mQZ2DJCOdajR0C5csC9e8DMmTLoj8jcKQUcOSKtHMYIDQXmzJG1XAyzU0qVku/7jz+WMSZZbd8+SUAeP5b9AgWAQYNkCr27e9Z/PTI/bBkhSsf48ZKIVKjASqtkGZQCRo+Wrpk5c4x7rZeXrOkSEgJMnizjSa5dA8aOTTrw9XUZaoIAQOXKsl+mDLBggSRE33zDRISSYwVWypE4aJUsjV4vs74M37eZHfNRoIAk4iNGyBiT+HjA8AerUpKwfPgh4OmZ8fdUCti1S7o64+OBAwfkeOHCUrK9cmXpJiJKDbtpKMfhoFWyNHFxUol09WpJQpYskUqkWc3fXwa32tvLgNhRo6QrMzUvXgBr1kgCc+GCHLO1Ba5cAby9sz4+sjzspiFKxapVHLRKliMmRpYnWL1abvRr1mRPIgJI9eH69eVrLlsmXZidOknxsZfdvy/dLSVKSCwXLgDOzsCwYTJzh4kIGYvJCOUorLRKlkSvBzp0ADZvltaKzZulCyW7+PgAhw/L7Jv335fuly1bgNq1Za2me/fkvN27gQkTgPBwoFgxGQAeGiotJCVKZF98ZL2YjFCOMmECB62S5bCxAZo0AZycgB07gHbtTPN169aVqq5//QX07ClThsPDATc3eb5zZ6B1a2mluXZNZuS4upomNrJOHDNCOUZgIFCzpvy1uW8f0Lix1hERZUxoqMyG0UpICHDnDlCrlnYxkGXimBGil+j1UttAr5dBq0xEyFzdvSutERERice0TEQAWYCPiQhlJ07tpRzh5UGr332ndTREKQsNBd57D7h8WSqXbt6sdUREpsFkhKzeq4NWixXTNh6ilFy9KuNDbtyQlojp07WOiMh02E1DVs8waLV8eQ5aJfN04YJMqb1xQ6qVHj0KvPGG1lERmQ6TEbJqQUFShhqQSqv29pqGQ5TM2bOyzkxYGFCpkkyt1XqMCJGpMRkhq/XqoNUmTbSOiCip+HipdPrggcz0OniQ67ZQzsRkhKzWqlXAsWMctErmy9YW2LQJ6NhRppsb6ngQ5TRMRsgqPX7MQatkvu7fT9yuUEFmzbCEEuVkTEbIKo0fz0GrZJ7WrZO1W/bt0zoSIvPBZISszvXricusc9AqmQul5Puya1epIbJhg9YREZkP1hkhqzNnjgwMfO89Dlol8/D4MTBwILB+vewPGJA4y4uI2DJCVubBA2DpUtkePVrbWIgAqfz71luSiNjZAdOmSQuJDX/7EiVgywhZlfnzgWfP5Jc/W0VIa3/9JcXM4uNlnMjatUDt2lpHRWR+mIyQ1Xj6FPjhB9kePRrQ6bSNh+jNN4EuXSQZWbQIcHXVOiIi88RkhKzG8uXSTVOqFNCpk9bRUE61c6escFuwoCTEy5cDuXIxOSZKC3stySrExQGzZsn2yJHSN09kSi9eyDTy1q2B3r1l9gwgs7mYiBCljb+yySps2CCLjBUqBPTqpXU0lNP88490x5w7J/ulS0uCnCuXtnERWQomI2TxlAJmzJDtIUOA3Lm1jYdyDqWkG2bIEBk4XbAg4OcnrSNElHFMRsji7d4tf5HmyQMMGqR1NJRTRERIvRBD7ZAmTYDVqwEPD23jIrJEHDNCFs/QKtK/P1CggLaxUM6hFHDyZGLtkD17mIgQZRZbRsiinT4NHDggN4Rhw7SOhqydXi+DUXU6IF8+aRVRirVDiF4XW0bIok2fLv927Qp4eWkbC1m327eBpk2BJUsSj9WqxUSEKCswGSGLdfkysGWLbH/xhbaxkHXbsQOoUgXYvx/46ispsEdEWYfJCFmsWbOkibxNG6l0SZTVDLVD2raVgnpvvQUcPSqDpYko6zAZIYt05w6wcqVss1WEssM//wB16gDz5sn+sGGy6F25ctrGRWSNOICVLNL33wPR0YCPD/DOO1pHQ9bmwQMZDxIVJYX0/PyAVq20jorIejEZIYsTGSlLsANcEI+yh5ubLCtw+DBrhxCZArtpyOIsXiwFpypUkL58otf18KEMTN2/P/HYuHGsHUJkKmwZIYsSHQ3MnSvbo0YBNkyn6TU8fAjMni3jQqKipGZNo0bS2mZrq3V0RDkHkxGyKGvWSL0HT0+pLUKUGY8eAXPmyNijyEg5VrWqJLhEZHpMRshi6PWJpd+HDQMcHLSNhyzTokUy1siQhFSpAkyaBLRrx5Y2Iq0wGSGLsX07cOkS4Ooq69AQZYaLiyQilStLEtK+PZMQIq0xGSGLoFRi6fdBg+SGQpSeiAjpiilWDOjdW475+sr3T6tWTEKIzIVOKaW0DiI9kZGRcHV1RUREBFx4F8qRjhwBGjSQrpnr1wF3d60jInNmSELmzAEeP5YxRlevAo6OWkdGlLNk9P7NlhGyCIZWkZ49mYhQ6iIiZGbM7NmShABAxYrAxImAvb2moRFRGpiMkNn76y/gt99kuuXIkVpHQ+Zq40ZgwACZKQNIHZqJE4EPPmB3DJG5y9SP6IIFC+Dt7Q1HR0fUqFEDR44cydDr/vjjD9jZ2aFatWqZ+bKUQ82cKf926gSUKaNtLGS+SpWSRKR8eWDtWuD8eRkfwkSEyPwZ/WO6fv16DB06FOPGjUNgYCDq16+Pli1bIiQkJM3XRUREoEePHmjSpEmmg6WcJyQE+OUX2eaCeGQQFQVMmwaMH594rEYNKVr2119Aly4sWkZkSYwewFq7dm1Ur14dCw2LgwCoUKEC2rdvj6lTp6b6ui5duqBMmTKwtbXFtm3bEBQUlOGvyQGsOdewYVJxtVGjpKW6KeeIjpZCd4bHX38BP/4oi9nZ28vA1GLFtI6SiFKSLQNYY2JicObMGYwZMybJ8WbNmuHYsWOpvm7FihW4evUqfv75Z0yZMiXdrxMdHY3o6OiE/UhDdSLKUR4+BH76SbZHj9Y2Fsp6sbHA3buSYNy6lTTh+M9/EteEmTRJWkFeVaYMMGEC144hsgZGJSP3799HfHw8ihQpkuR4kSJFcOfOnRRfc+XKFYwZMwZHjhyBnV3GvtzUqVPx9ddfGxMaWaEffwSePpUy3c2aaR0NZZReD9y7lzS5uH0b+OQToHBhOWfKFEkkUmuX7d8/Mcnw9JQp3Z6eQNGi8m/bttIVk8FfKURk5jL1o6x7Zc12pVSyYwAQHx+Prl274uuvv0bZsmUz/P5jx47F8OHDE/YjIyPh5eWVmVDJQj1/LlM0ARkrksK3F5mZyEhgyRLpVrt1K/nzTZsmJiP58kkiYmcnSYenZ9LHy60dn3wCDB7M7wEia2ZUMlKwYEHY2tomawUJDw9P1loCAFFRUQgICEBgYCAGDx4MANDr9VBKwc7ODnv27EHjxo2Tvc7BwQEOXHgkR1uxArh/HyhZEujcWetoKCMWLADGjpVtnQ4oUiQxuShaVBIQg+7d5XMtWDD92S5s/SCyfkb9mNvb26NGjRrw9/dHhw4dEo77+/ujXbt2yc53cXHB+fPnkxxbsGAB9u/fj02bNsHb2zuTYWed6GhpUuYAOPMRFwd8951sjxjBm5G5unwZePIEqF5d9gcMALZsAQYOlBWV06p26upqmhiJyDIY/Wt++PDh6N69O2rWrAkfHx8sWbIEISEhGDhwIADpYrl16xZWrVoFGxsbVKpUKcnrCxcuDEdHx2THtRARAXToINNHjx1LbEImbW3aBAQHy1/NhvVEyHycOiUVcbduBWrXlp8dnQ7In1+eIyIyltHJiK+vLx48eIDJkycjLCwMlSpVws6dO1GiRAkAQFhYWLo1R8zF8+dy07t+HWjdWmoU5M2rdVQ5m1LAjBmy/dlngJOTtvGQUArYtUs+m4MHE48XKiSDjPlzQ0SvI8cvlHf5MlC3rtQsaNFClqnPlStLvwQZwd9fZs44OUmLlZub1hHRrl0yiNjQ42pnB3TrBowaBbz5praxEZF5y+j9O8cXSi5bVtY9yZ1bfun265f6dEPKfoYF8fr2ZSJiLqKiJBHJmxcYPhy4dg3w82MiQkRZJ8cnI4D0e2/YIOWjV64EvvpK64hypjNngH375HN4aWY3mdC9e7K43KJFicc6dpTpuiEhwKxZAGfZE1FWYzLyP23aJP4CXrAACAvTNp6cyDBW5MMPgf8NQSITuXZNanmUKAFMniwPQxFkW1vg889lgCoRUXbgpMmX9O0LPHsGNGnCEtOmdvWqzKIBZCwCmUZgoCSBGzZI5VQAqFlTyu9zSjURmQp/3bxiyJCk+7GxHNBqCt99JzfDli2BKlW0jiZnmDYtsUgZADRvLknIu++y2ikRmRa7adJw8KAMcP37b60jsW5370rFVYAL4mUnvV6m4Rq0aCFdMF27SgvJrl2yOjITESIyNSYjqVBK+s2vX5df2jdvah2R9frhBxmfULs20KCB1tFYn/h46YZ56y1g2LDE49Wqyff1mjWyTUSkFSYjqdDpgI0bgfLl5Rd2y5bA48daR2V9oqJkdV6AC+Jltbg4YNUqmYLr6wv8+aeUa3+5dcTdXbv4iIgMcnwyEh+f+nNubtJ07eEB/PUX0L498OKFyULLEX76SZK8smWBFJY3okyIjpb/17JlgZ49gUuXZJG6SZOkyF+ePFpHSESUVI5ORg4elEW+rl9P/ZwSJYDffwdcXIBDh4AePRJnHVi7e/ekMm12FYGLiQFmz5btUaNk/AK9vtmzgf79ZamDQoVkoOqNG1I/pEABraMjIkoux5aD1+ulD/3PP2WJ8927gbTW7tu/X8aOxMYC8+cDn36aJWGYJaWACROAb7+VbScnoHjx1B/FigEODsZ/HT8/4OOPpavg+vXMvQfJyrn37gGGRbDv3wfq1QM++UQqCrMlhIi0ktH7d45NRgDg1i2Zzvj339KMvWOH/BJPzfr1snbN8uXWe+NUChg3Dpg61bjXFSmSdsJSqFDS8SB6PVC5MnDhgvzlzlk0xouIkMR4zhwZF3LoUOJzSnH8DRFpj8lIBj18CLRtK8ugOzrKoNU2bbL0S1gMpYAvv5TkAJAS4AMGyADekJDUH8+fp//ejo5SRtyQnNjaAkuXSvdXSAjg6pqtl2ZVHjwAvv8emDdPEhIAKF0aOHECKFhQ29iIiF7GZMQIz54BnTvLgnm2tsCyZTLwLy16vfw137SprDJr6ZSSAliGheq+/z55AbjUXvfgQWJiEhqaPFkJC0t93MkXXyR+TUrb3bsyHmTBAumaAYAKFaQly9eXFVOJyPwwGTFSbKyUg1+1SvZnzgRGjkz9/AULZNxI3rzSPF69eraEZRJKSWI1c6bs//CDrFOSVWJipEvs1SRFKSlFni9f1n0ta7ZmDfDRR7JdrZos6NihA2CTo4ehE5E5YzKSCXq9/KU+a5bsjxolf7Wn1PceEwO0bg3s3QsULizdPKVLZ1to2UYpuebvvpN9ax+ca0muX5cZMY0ayX5cHNCtG9C9u3zvcUwIEZk7JiOvYeZMuUEDQK9eUrMhpSbwyEigYUMgKAh44w3gjz8kMbEUSknrj2F67Y8/AoMGaRtTTvfsGXDkCLB2LfDzz1Lj5upVwN5e68iIiIyX0fs3e5lTMGqUDATs10+mn96/LzNpnJySnufiIjVIfHyAf/+Vga/790vXjblTChgxQmZiAMDChcDAgdrGlFP99ZfM0vL3lxa2mJjE5ypUkO8/T0/t4iMiym7sbU7Fxx9L6WxHR5ny26wZ8OhR8vPc3aVGiZsbcPo00KVL9hUJyypKAcOHJyYiixYxETEVpaQK6ssl2devl0GoBw9KIuLlBfTuLbNj9uxhIkJE1o/JSBref19uBq6u0gXToAFw+3by88qWlZk4+fNLt4459+UrJYulzZ0r+4sXy/Rdyj537wK//CIJRokSQLlywL59ic+3agV07CiDoi9flmqpy5bJwoFERDkBu2nSUb8+cPiwVF/96y+gbl1JUMqWTXpe7doy4NAEQ1oyTSlg6FCpTwEAS5ZIVxRlvRs35P95716p8vsye3sZmGrg4wNs3mza+IiIzAlbRjKgShVpGXnjDbnJvPMOcOZM8vNeTkRCQmQQorlQSuqGzJsnLTdLlzIRySpxccDJk0BAQOKx2FgZGGxIRKpVk8HCu3ZJob3PP9ckVCIis8SWkQzy9paEpGVL4OxZ4N13gW3bgCZNkp975460oNy+LWNOOnQwdbRJKQV89pnMljEkIr17axuTJdPrZSXc/ful5ePAAamE2qGDjDMCZJr3iBFAzZryPVKokLYxExGZM7aMGKFwYbnxNG4sFTBbtZLy8a8qUkSeUwro1EnKze/cCcTHmz5mvV7qhhgSkWXLmIi8jtatpUhbxYpSGG7bNklE8ueXh4FOJ7VbunRhIkJElB62jBjJxUUSi48+AjZtkjLc9+/LCqkGOp0MRoyOloquO3bIw9tbBov275/0xpVdDInIokUS04oV6Ze5z8mUku61gIDEx4sXUvfD4PFjICoKyJ0bqFNHlgNo2lRWgLa11Sx0IiKLxqJnmRQfL38ZL1ok+5MmARMmJJ9Jc/mynLNihdzIdDrg2jWgZMnsjU+vlwJmixczEUnPjz9KshgQIInly2xspOXDUDvm5EmpN1OhAteCISJKDyuwmoBSwNdfywOQm/+8eSn/hfzsmdSTuHhR1mMxGDlSBsZ26wY4O2dNXHq9tNQsWSKJyMqVUkI8J7t7N7G149w5YMOGxGSiRw9g9WrZtrOTAcs1ayY+qlRhqwcRUWYwGTGh+fNlpopSsvrvqlWAg0P6r7txQ7pulJJEpHt3SSIqVcp8LHq9dAUtXSp/1a9cmbi4mtbi44ELF+T6DC1Ifn6SpNnZJX/kygVMnCi1OQCpzbF7d+rndu4MFC0q5168KINJDQnIzZtJYzl/PvH/ee9e4MoVSTwqV5ZBx0RE9PpYDt6EBg+WQYrdu8tf3A8fyo0wvZaO/Pll+ufChdKds2CBPOrXl6SkY8eMJTUGryYiq1ZJi4sWlJKVek+eBE6dSpz6+vSplM43LCp46ZJMd03N0KGJ28ePJ64snJLatROTkV27ZFVbA51OulZq1gRq1Eg6qPS99+RBRETaYDKSRXx9gQIFZHrn3r0y42bnzrRnUri4yM32889lmujChTI748gReRhTlEyvl3OXL5dEZPVqoGvXrLgy461aBYwZA4SFJX8ub14ZM2NIRjp3liQhLi7pIzZW/vXwSHxtnToyXfbVcw3nv7xIoZeXXL+hq6VatazrBiMioqzFbposdvq01CJ58ECqtO7ebdxg1Vu3pGVj3TppUTDcQHfvlm6O5s2Tj1/Q64G+fWWQqo0NsGaNTCnNLrGxUo325VaPuXNlVgkgrUKdOkmclSsDtWpJq0WtWpJ4cPwFEVHOwDEjGvrnH0kaQkKkm8XLSxY7e/VRtGji9qsrAiuVOK5CKZk6eu5c4vTg3r2l1SU+XhIRPz+5ya9ZI600We3qVZl1cuqUVJ998SLp85MnA+PHy/bjx5KsvPUWkCdP1sdCRESWgcmIxm7eBNq0kQQiI1xdU05SPD2BggWl62PzZplmCsj6Jp06SZJw6pQkIr/8It1DS5ZIkvLyIy5O/m3YUBYABGQa6+jRKZ8XHy8rF7drJ+f++SdQtWpivPnySUuHodWjTh2Jk4iIyIADWDVWrJiUjb92TcZO3Lol5eFffdy6JdN+IyLkcfFi2u/r7CwJw/PniWvf6HSy/cEH8vpx49J+D0My8uyZjDFJTb58iclIxYoyY6hmTUk+3nhDuoSIiIheF5ORbGRjIzftN95I/RylgMjIlJOUV4/Fxkr1z5fpdDIA9oMPZD9/fqBPH2kpMTzs7BK369dPfG2+fMC33yY/x7D/8uBROzvg+++z7L+GiIgoAbtpLIReL1OGX05S7t6VRdhq19Y6OiIiouTYTWNlbGxkTEbBgknHbhAREVk69voTERGRppiMEBERkaaYjBAREZGmmIwQERGRppiMEBERkaaYjBAREZGmmIwQERGRppiMEBERkaaYjBAREZGmmIwQERGRppiMEBERkaaYjBAREZGmmIwQERGRpixi1V6lFABZipiIiIgsg+G+bbiPp8YikpGoqCgAgJeXl8aREBERkbGioqLg6uqa6vM6lV66Ygb0ej1u374NZ2dn6HQ6rcPJNpGRkfDy8kJoaChcXFy0Didb5aRrBXLW9fJarVdOul5ea9ZQSiEqKgqenp6wsUl9ZIhFtIzY2NigWLFiWodhMi4uLlb/zW+Qk64VyFnXy2u1Xjnpenmtry+tFhEDDmAlIiIiTTEZISIiIk0xGTEjDg4OmDhxIhwcHLQOJdvlpGsFctb18lqtV066Xl6raVnEAFYiIiKyXmwZISIiIk0xGSEiIiJNMRkhIiIiTTEZISIiIk0xGTGRqVOn4u2334azszMKFy6M9u3b49KlS2m+5uDBg9DpdMke//zzj4mizpxJkyYli9nd3T3N1xw6dAg1atSAo6MjSpUqhUWLFpko2tdXsmTJFD+nTz/9NMXzLelzPXz4MNq2bQtPT0/odDps27YtyfNKKUyaNAmenp7InTs33n33Xfz999/pvu/mzZtRsWJFODg4oGLFiti6dWs2XUHGpXWtsbGxGD16NCpXrow8efLA09MTPXr0wO3bt9N8Tz8/vxQ/6xcvXmTz1aQvvc+2V69eyeKuU6dOuu9raZ8tgBQ/I51Oh5kzZ6b6nub62WbkXmOOP7dMRkzk0KFD+PTTT3HixAn4+/sjLi4OzZo1w9OnT9N97aVLlxAWFpbwKFOmjAkifj1vvvlmkpjPnz+f6rnBwcFo1aoV6tevj8DAQHz55ZcYMmQINm/ebMKIM+/06dNJrtXf3x8A8MEHH6T5Okv4XJ8+fYqqVati/vz5KT4/Y8YMzJ49G/Pnz8fp06fh7u6Opk2bJqwnlZLjx4/D19cX3bt3x7lz59C9e3d07twZJ0+ezK7LyJC0rvXZs2c4e/Ysxo8fj7Nnz2LLli24fPky3n///XTf18XFJcnnHBYWBkdHx+y4BKOk99kCQIsWLZLEvXPnzjTf0xI/WwDJPp/ly5dDp9OhU6dOab6vOX62GbnXmOXPrSJNhIeHKwDq0KFDqZ5z4MABBUA9evTIdIFlgYkTJ6qqVatm+PwvvvhClS9fPsmxAQMGqDp16mRxZKbx+eefq9KlSyu9Xp/i85b6uQJQW7duTdjX6/XK3d1dTZs2LeHYixcvlKurq1q0aFGq79O5c2fVokWLJMeaN2+uunTpkuUxZ9ar15qSU6dOKQDqxo0bqZ6zYsUK5erqmrXBZYOUrrdnz56qXbt2Rr2PtXy27dq1U40bN07zHEv5bF+915jrzy1bRjQSEREBAChQoEC657711lvw8PBAkyZNcODAgewOLUtcuXIFnp6e8Pb2RpcuXXDt2rVUzz1+/DiaNWuW5Fjz5s0REBCA2NjY7A41S8XExODnn39G7969013U0RI/15cFBwfjzp07ST47BwcHNGzYEMeOHUv1dal93mm9xhxFRERAp9MhX758aZ735MkTlChRAsWKFUObNm0QGBhomgCzwMGDB1G4cGGULVsW/fr1Q3h4eJrnW8Nne/fuXfz222/o06dPuudawmf76r3GXH9umYxoQCmF4cOH45133kGlSpVSPc/DwwNLlizB5s2bsWXLFpQrVw5NmjTB4cOHTRit8WrXro1Vq1Zh9+7d+Omnn3Dnzh3UrVsXDx48SPH8O3fuoEiRIkmOFSlSBHFxcbh//74pQs4y27Ztw+PHj9GrV69Uz7HUz/VVd+7cAYAUPzvDc6m9ztjXmJsXL15gzJgx6Nq1a5oLi5UvXx5+fn7Yvn071q5dC0dHR9SrVw9XrlwxYbSZ07JlS6xZswb79+/HrFmzcPr0aTRu3BjR0dGpvsYaPtuVK1fC2dkZHTt2TPM8S/hsU7rXmOvPrUWs2mttBg8ejD///BNHjx5N87xy5cqhXLlyCfs+Pj4IDQ3Fd999hwYNGmR3mJnWsmXLhO3KlSvDx8cHpUuXxsqVKzF8+PAUX/NqK4L6X2Hg9FoXzM2yZcvQsmVLeHp6pnqOpX6uqUnps0vvc8vMa8xFbGwsunTpAr1ejwULFqR5bp06dZIM+qxXrx6qV6+OH374AfPmzcvuUF+Lr69vwnalSpVQs2ZNlChRAr/99luaN2pL/mwBYPny5ejWrVu6Yz8s4bNN615jbj+3bBkxsc8++wzbt2/HgQMHUKxYMaNfX6dOHbPKvDMiT548qFy5cqpxu7u7J8uuw8PDYWdnBzc3N1OEmCVu3LiBvXv3om/fvka/1hI/V8MMqZQ+u1f/gnr1dca+xlzExsaic+fOCA4Ohr+/v9HLrdvY2ODtt9+2uM8akBa9EiVKpBm7JX+2AHDkyBFcunQpUz/D5vbZpnavMdefWyYjJqKUwuDBg7Flyxbs378f3t7emXqfwMBAeHh4ZHF02Ss6OhoXL15MNW4fH5+EGSgGe/bsQc2aNZErVy5ThJglVqxYgcKFC6N169ZGv9YSP1dvb2+4u7sn+exiYmJw6NAh1K1bN9XXpfZ5p/Uac2BIRK5cuYK9e/dmKlFWSiEoKMjiPmsAePDgAUJDQ9OM3VI/W4Nly5ahRo0aqFq1qtGvNZfPNr17jdn+3GbJMFhK1yeffKJcXV3VwYMHVVhYWMLj2bNnCeeMGTNGde/ePWF/zpw5auvWrery5cvqr7/+UmPGjFEA1ObNm7W4hAwbMWKEOnjwoLp27Zo6ceKEatOmjXJ2dlbXr19XSiW/zmvXriknJyc1bNgwdeHCBbVs2TKVK1cutWnTJq0uwWjx8fGqePHiavTo0cmes+TPNSoqSgUGBqrAwEAFQM2ePVsFBgYmzCCZNm2acnV1VVu2bFHnz59XH374ofLw8FCRkZEJ79G9e3c1ZsyYhP0//vhD2draqmnTpqmLFy+qadOmKTs7O3XixAmTX9/L0rrW2NhY9f7776tixYqpoKCgJD/D0dHRCe/x6rVOmjRJ7dq1S129elUFBgaqjz/+WNnZ2amTJ09qcYlJpHW9UVFRasSIEerYsWMqODhYHThwQPn4+KiiRYta3WdrEBERoZycnNTChQtTfA9L+Wwzcq8xx59bJiMmAiDFx4oVKxLO6dmzp2rYsGHC/vTp01Xp0qWVo6Ojyp8/v3rnnXfUb7/9ZvrgjeTr66s8PDxUrly5lKenp+rYsaP6+++/E55/9TqVUurgwYPqrbfeUvb29qpkyZKp/kIwV7t371YA1KVLl5I9Z8mfq2Ea8quPnj17KqVkmuDEiROVu7u7cnBwUA0aNFDnz59P8h4NGzZMON9g48aNqly5cipXrlyqfPnyZpGIpXWtwcHBqf4MHzhwIOE9Xr3WoUOHquLFiyt7e3tVqFAh1axZM3Xs2DHTX1wK0rreZ8+eqWbNmqlChQqpXLlyqeLFi6uePXuqkJCQJO9hDZ+tweLFi1Xu3LnV48ePU3wPS/lsM3KvMcefW93/giciIiLSBMeMEBERkaaYjBAREZGmmIwQERGRppiMEBERkaaYjBAREZGmmIwQERGRppiMEBERkaaYjBAREZGmmIwQERGRppiMEBERkaaYjBAREZGmmIwQERGRpv4fW3FiUN7qR+0AAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "original_model = history_original.history[\"val_loss\"]\n",
    "larger_model = history_larger_model.history[\"val_loss\"]\n",
    "epochs = range(1,21)\n",
    "plt.plot(epochs, original_model, \"b--\", label=\"Original Model\")\n",
    "plt.plot(epochs, larger_model, \"b-\", label=\"Larger Model\")\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Larger model tend to overfit as soon as the training start where the original model is overfitting after sometime."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "30/30 [==============================] - 4s 35ms/step - loss: 0.5903 - accuracy: 0.7805 - val_loss: 0.4642 - val_accuracy: 0.8672\n",
      "Epoch 2/20\n",
      "30/30 [==============================] - 0s 8ms/step - loss: 0.3920 - accuracy: 0.8975 - val_loss: 0.3901 - val_accuracy: 0.8860\n",
      "Epoch 3/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.3274 - accuracy: 0.9191 - val_loss: 0.3939 - val_accuracy: 0.8707\n",
      "Epoch 4/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2983 - accuracy: 0.9256 - val_loss: 0.3631 - val_accuracy: 0.8875\n",
      "Epoch 5/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.2727 - accuracy: 0.9361 - val_loss: 0.3633 - val_accuracy: 0.8861\n",
      "Epoch 6/20\n",
      "30/30 [==============================] - 0s 9ms/step - loss: 0.2654 - accuracy: 0.9373 - val_loss: 0.3694 - val_accuracy: 0.8842\n",
      "Epoch 7/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2539 - accuracy: 0.9411 - val_loss: 0.4048 - val_accuracy: 0.8734\n",
      "Epoch 8/20\n",
      "30/30 [==============================] - 0s 9ms/step - loss: 0.2446 - accuracy: 0.9452 - val_loss: 0.3819 - val_accuracy: 0.8794\n",
      "Epoch 9/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2366 - accuracy: 0.9485 - val_loss: 0.3768 - val_accuracy: 0.8816\n",
      "Epoch 10/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2338 - accuracy: 0.9483 - val_loss: 0.3817 - val_accuracy: 0.8816\n",
      "Epoch 11/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2245 - accuracy: 0.9533 - val_loss: 0.3844 - val_accuracy: 0.8809\n",
      "Epoch 12/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2291 - accuracy: 0.9497 - val_loss: 0.3895 - val_accuracy: 0.8808\n",
      "Epoch 13/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2186 - accuracy: 0.9549 - val_loss: 0.3954 - val_accuracy: 0.8789\n",
      "Epoch 14/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2222 - accuracy: 0.9513 - val_loss: 0.4023 - val_accuracy: 0.8762\n",
      "Epoch 15/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2122 - accuracy: 0.9566 - val_loss: 0.4173 - val_accuracy: 0.8761\n",
      "Epoch 16/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.2108 - accuracy: 0.9565 - val_loss: 0.4141 - val_accuracy: 0.8749\n",
      "Epoch 17/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.2068 - accuracy: 0.9593 - val_loss: 0.4082 - val_accuracy: 0.8780\n",
      "Epoch 18/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.2055 - accuracy: 0.9608 - val_loss: 0.4172 - val_accuracy: 0.8750\n",
      "Epoch 19/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.2092 - accuracy: 0.9569 - val_loss: 0.4268 - val_accuracy: 0.8740\n",
      "Epoch 20/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.1983 - accuracy: 0.9641 - val_loss: 0.4297 - val_accuracy: 0.8742\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.keras import regularizers\n",
    "model = keras.Sequential([layers.Dense(16, kernel_regularizer=regularizers.l2(0.002), activation=\"relu\"),\n",
    "                          layers.Dense(16, kernel_regularizer=regularizers.l2(0.002), activation=\"relu\"),\n",
    "                          layers.Dense(1, activation=\"sigmoid\")])\n",
    "model.compile(optimizer=\"rmsprop\",\n",
    "              loss=\"binary_crossentropy\",\n",
    "              metrics=[\"accuracy\"])\n",
    "history_l2_reg = model.fit(train_data, train_labels,\n",
    "                           epochs=20, batch_size=512, validation_split=0.4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x22722764790>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "original_model = history_original.history[\"val_loss\"]\n",
    "l2_model = history_l2_reg.history[\"val_loss\"]\n",
    "epochs = range(1,21)\n",
    "plt.plot(epochs, original_model, \"b--\", label=\"Original Model\")\n",
    "plt.plot(epochs, l2_model, \"b-\", label=\"L2 Model\")\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "L2 is resistant to overfitting than the original model as it adds values to the loss and the neural net trying to reduce it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "30/30 [==============================] - 2s 16ms/step - loss: 0.6134 - accuracy: 0.6561 - val_loss: 0.4739 - val_accuracy: 0.8426\n",
      "Epoch 2/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.4739 - accuracy: 0.7959 - val_loss: 0.3758 - val_accuracy: 0.8703\n",
      "Epoch 3/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.3917 - accuracy: 0.8476 - val_loss: 0.3246 - val_accuracy: 0.8766\n",
      "Epoch 4/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.3299 - accuracy: 0.8769 - val_loss: 0.2946 - val_accuracy: 0.8853\n",
      "Epoch 5/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.2906 - accuracy: 0.8973 - val_loss: 0.2806 - val_accuracy: 0.8903\n",
      "Epoch 6/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2543 - accuracy: 0.9138 - val_loss: 0.2778 - val_accuracy: 0.8881\n",
      "Epoch 7/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.2227 - accuracy: 0.9241 - val_loss: 0.2833 - val_accuracy: 0.8851\n",
      "Epoch 8/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.2030 - accuracy: 0.9339 - val_loss: 0.2920 - val_accuracy: 0.8868\n",
      "Epoch 9/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.1777 - accuracy: 0.9438 - val_loss: 0.3128 - val_accuracy: 0.8861\n",
      "Epoch 10/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.1538 - accuracy: 0.9508 - val_loss: 0.3596 - val_accuracy: 0.8777\n",
      "Epoch 11/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.1409 - accuracy: 0.9553 - val_loss: 0.3565 - val_accuracy: 0.8883\n",
      "Epoch 12/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.1266 - accuracy: 0.9601 - val_loss: 0.3630 - val_accuracy: 0.8872\n",
      "Epoch 13/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.1118 - accuracy: 0.9667 - val_loss: 0.4189 - val_accuracy: 0.8882\n",
      "Epoch 14/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.1029 - accuracy: 0.9687 - val_loss: 0.4409 - val_accuracy: 0.8870\n",
      "Epoch 15/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0983 - accuracy: 0.9724 - val_loss: 0.4436 - val_accuracy: 0.8863\n",
      "Epoch 16/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0928 - accuracy: 0.9723 - val_loss: 0.4933 - val_accuracy: 0.8870\n",
      "Epoch 17/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0854 - accuracy: 0.9739 - val_loss: 0.5159 - val_accuracy: 0.8857\n",
      "Epoch 18/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0795 - accuracy: 0.9768 - val_loss: 0.5584 - val_accuracy: 0.8855\n",
      "Epoch 19/20\n",
      "30/30 [==============================] - 0s 7ms/step - loss: 0.0793 - accuracy: 0.9769 - val_loss: 0.5728 - val_accuracy: 0.8836\n",
      "Epoch 20/20\n",
      "30/30 [==============================] - 0s 6ms/step - loss: 0.0758 - accuracy: 0.9780 - val_loss: 0.6049 - val_accuracy: 0.8855\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential([layers.Dense(16, activation=\"relu\"),\n",
    "                          layers.Dropout(0.5),\n",
    "                          layers.Dense(16, activation=\"relu\"),\n",
    "                          layers.Dropout(0.5),\n",
    "                          layers.Dense(1, activation=\"sigmoid\")])\n",
    "model.compile(optimizer=\"rmsprop\",\n",
    "              loss=\"binary_crossentropy\",\n",
    "              metrics=[\"accuracy\"])\n",
    "history_dropout = model.fit(train_data, train_labels,\n",
    "                            epochs=20, batch_size=512, validation_split=0.4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2271d7af6a0>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiMAAAGdCAYAAADAAnMpAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAABeg0lEQVR4nO3deXhMZxsG8HuyCxI7Qey7KG2oraWlYm2r2tpaqtRHF2sXVFtLWxRFUVRr60prq6oiqtS+RCi1E0tJGoIEiWzzfn88neyJTDKTd5b7d11z5czMOWeeMTLnybs8r0EppUBERESkiYvuAIiIiMi5MRkhIiIirZiMEBERkVZMRoiIiEgrJiNERESkFZMRIiIi0orJCBEREWnFZISIiIi0ctMdQG4YjUZcvXoVRYsWhcFg0B0OERER5YJSCrdv30b58uXh4pJ9+4ddJCNXr16Fv7+/7jCIiIgoDy5fvoyKFStm+7xdJCNFixYFIG/Gx8dHczRERESUGzExMfD390+5jmfHLpIRU9eMj48PkxEiIiI7c78hFhzASkRERFoxGSEiIiKtmIwQERGRVnYxZiQ3lFJISkpCcnKy7lDIAbi7u8PV1VV3GERETsEhkpGEhASEh4cjNjZWdyjkIAwGAypWrIgiRYroDoWIyOHZfTJiNBoRFhYGV1dXlC9fHh4eHiyMRvmilMK1a9fwzz//oGbNmmwhISKyMrtPRhISEmA0GuHv7w9vb2/d4ZCDKF26NC5cuIDExEQmI0REVuYwA1hzKjNLZC62rhERFRxewYmIiEgrJiN26sKFCzAYDDh8+HCuj1m6dCmKFSumPQ5ryMt7MxgMWLt2rVXiISKi3GMyotHly5cxYMCAlIG3lStXxrBhwxAVFXXfY/39/REeHo6AgIBcv16PHj1w+vTp/IScJ4899hgMBgOmTJmS6blOnTrBYDBg/PjxBR4XERHZBiYjmpw/fx6NGzfG6dOn8cMPP+Ds2bNYsGABfv/9dzRv3hw3btzI9tiEhAS4urqiXLlycHPL/RjkQoUKoUyZMpYI32z+/v5YsmRJuseuXr2KrVu3ws/PT0tMRERkG5iMaPL666/Dw8MDmzdvRuvWrVGpUiV07NgRW7ZswZUrVzB27NiUfatUqYKPPvoI/fr1g6+vLwYOHJhl98i6detQs2ZNFCpUCI8//jiWLVsGg8GAW7duAcjclTF+/Hg0atQI33zzDapUqQJfX1/07NkTt2/fTtln48aNeOSRR1CsWDGULFkSXbp0wblz58x+v126dEFUVBR27dqV8tjSpUsRFBSUKUG6efMm+vbti+LFi8Pb2xsdO3bEmTNn0u2zdOlSVKpUCd7e3njmmWeybE365ZdfEBgYCC8vL1SrVg0TJkxAUlKS2bETETmyuDjdEThwMnL3bva3e/dyv2/GDym7/cxx48YNbNq0Ca+99hoKFSqU7rly5crhhRdewIoVK6CUSnl82rRpCAgIQEhICN5///1M57xw4QKee+45dO3aFYcPH8agQYPSJTTZOXfuHNauXYv169dj/fr12L59e7rulLt372LkyJE4cOAAfv/9d7i4uOCZZ56B0Wg06z17eHjghRdeSNc6snTpUvTv3z/Tvv369cPBgwexbt067NmzB0opdOrUCYmJiQCAffv2oX///njttddw+PBhPP744/joo4/SnWPTpk148cUXMXToUBw/fhxffPEFli5dio8//tisuImIHNm5c0CNGsA332gORNmB6OhoBUBFR0dnei4uLk4dP35cxcXFpXscyP7WqVP6c3h7Z79v69bp9y1VKuv9zLF3714FQK1ZsybL52fMmKEAqH///VcppVTlypVV165d0+0TFhamAKjQ0FCllFKjRo1SAQEB6fYZO3asAqBu3ryplFJqyZIlytfXN+X5cePGKW9vbxUTE5Py2Ntvv62aNm2abeyRkZEKgDp69GiWcWSldevWatiwYerIkSOqaNGi6s6dO2r79u2qTJkyKiEhQTVs2FCNGzdOKaXU6dOnFQC1a9eulOOvX7+uChUqpH788UellFK9evVSHTp0SPcaPXr0SPfeHn30UTVp0qR0+3zzzTfKz88v5X5On0F2/6+IiBzJ2bNK1aunVNOmSiUmWv78OV2/03LYlhF7pv5rEUlb66Jx48Y5HnPq1Ck0adIk3WMPP/zwfV+rSpUqKFq0aMp9Pz8/REZGptw/d+4cevfujWrVqsHHxwdVq1YFAFy6dOn+bySDBx54ADVr1sTKlSuxePFi9OnTB+7u7un2OXHiBNzc3NC0adOUx0qWLInatWvjxIkTKfs0b9483XEZ74eEhGDixIkoUqRIym3gwIFcNoCIKI3q1YG9e4GffwbMGIJocXZfgTU7d+5k/1zGgppprr2ZZKylduFCnkNKUaNGDRgMBhw/fhxdu3bN9PzJkydRvHhxlCpVKuWxwoUL53hOpVSmQl0qTTdPdjImAwaDIV0XzJNPPgl/f398+eWXKF++PIxGIwICApCQkHDfc2elf//++Pzzz3H8+HHs378/y/eRlbTvLzfvy2g0YsKECejWrVum57y8vMyMmojIsZw/D1SrJttFi8pNJ4dtGSlcOPtbxmtRTvtmGNKR7X7mKFmyJNq1a4d58+YhLsOglIiICHz33Xfo0aOHWVVA69SpgwMHDqR77ODBg+YFlkFUVBROnDiB9957D23btkXdunVx8+bNfJ2zd+/eOHr0KAICAlCvXr1Mz9erVw9JSUnYt29fujhOnz6NunXrpuyzd+/edMdlvP/QQw/h1KlTqFGjRqYbq/USkTNbtQqoVQv49FPdkaTit7Imc+fORXx8PNq3b48///wTly9fxsaNG9GuXTtUqFDB7IGWgwYNwsmTJzFq1CicPn0aP/74I5YuXQog76XNixcvjpIlS2LhwoU4e/Ystm7dipEjR+bpXGnPGR4ejt9//z3L52vWrImnn34aAwcOxM6dO3HkyBG8+OKLqFChAp5++mkAwNChQ7Fx40ZMnToVp0+fxty5c7Fx48Z05/nggw/w9ddfY/z48fj7779x4sQJrFixAu+9916+4icismdHjgB9+wLJycCVK7qjScVkRJOaNWvi4MGDqF69Onr06IHq1avjf//7Hx5//HHs2bMHJUqUMOt8VatWxcqVK7F69Wo88MADmD9/fspsGk9PzzzF6OLiguXLlyMkJAQBAQEYMWIEpk2blqdzpVWsWLEcu52WLFmCwMBAdOnSBc2bN4dSChs2bEjpUmrWrBm++uorzJkzB40aNcLmzZszJRnt27fH+vXrERwcjCZNmqBZs2aYMWMGKleunO/4iYjs0fXrwNNPA7GxQLt2wNSpuiNKZVC56YDXLCYmBr6+voiOjoaPj0+65+7du4ewsDBUrVqVYwEy+Pjjj7FgwQJcvnxZdyh2h/+viMiRJCYCQUHAtm0yaHX/fsDMv3nzJKfrd1oOO4DVGc2bNw9NmjRByZIlsWvXLkybNg1vvPGG7rCIiEiz4cMlESlSBFi3rmASEXMwGXEgZ86cwUcffYQbN26gUqVKePPNNzFmzBjdYRERkUbbtwPz5gEGA/Ddd0AWcwe0y9OYkXnz5qU0XwcGBmLHjh3Z7tuvXz8YDIZMt/r16+c5aMrazJkzcfXqVdy7dw+nT5/G+++/b9baNURE5HhatQLmzAEmTQKeekp3NFkzOxlZsWIFhg8fjrFjxyI0NBSPPvooOnbsmG0RrM8++wzh4eEpt8uXL6NEiRJ4/vnn8x08ERER5cxgAN54Axg9Wnck2TM7GZkxYwYGDBiAV155BXXr1sWsWbPg7++P+fPnZ7m/r68vypUrl3I7ePAgbt68iZdffjnfwRMREVFmsbHAiBHAf+uk2jyzkpGEhASEhIQgKCgo3eNBQUHYvXt3rs6xaNEiPPHEEzlOsYyPj0dMTEy6GxEREd2fUkD//sCsWUCXLnLf1pmVjFy/fh3JyckoW7ZsusfLli2LiIiI+x4fHh6O3377Da+88kqO+02ePBm+vr4pN39/f3PCJCIiclqffAKsWCFrzUyaJN00ti5PA1izWgMlN1U+ly5dimLFimW5HktaY8aMQXR0dMqNdTKIiIjub/164N13ZXvOHBm8ag/MmmpRqlQpuLq6ZmoFiYyMzNRakpFSKmWlVg8Pjxz39fT0zHPVUCIiImd04gTQu7d0ywwaBAwerDui3DOrZcTDwwOBgYEIDg5O93hwcDBatGiR47Hbt2/H2bNnMWDAAPOjJJuxbds2GAwG3DJjVFSVKlUwa9Ysq8VEROTsbt6UUu+3bwOPPgrMnq07IvOY3U0zcuRIfPXVV1i8eDFOnDiBESNG4NKlSxj8Xwo2ZswY9O3bN9NxixYtQtOmTREQEJD/qB1Av379su2uunHjBoYMGYLatWvD29sblSpVwtChQxEdHX3fcxoMhpTPIq3XXnsNBoMB/fr1s0D0RERkS/79F4iPBypVAlauBO7TAWFzzK6I1aNHD0RFRWHixIkIDw9HQEAANmzYkDI7Jjw8PFPNkejoaKxatQqfffaZZaJ2cFevXsXVq1cxffp01KtXDxcvXsTgwYNx9epVrFy5Msdj/f39sXz5csycOROFChUCIOus/PDDD6hUqVJBhE9ERAWsTh3g4EEgMhIoU0Z3NObL0wDW1157DRcuXEB8fDxCQkLQKs0ImaVLl2Lbtm3p9vf19UVsbCwGDhyYr2CdRUBAAFatWoUnn3wS1atXR5s2bfDxxx/jl19+QVJSUo7HPvTQQ6hUqRJWr16d8tjq1avh7++PBx98MN2+8fHxGDp0KMqUKQMvLy888sgjOHDgQLp9NmzYgFq1aqFQoUJ4/PHHceHChUyvuXv3brRq1QqFChWCv78/hg4dirt37+b9H4CIiHIlbYN56dKAvRY3z1MyYsuUAu7e1XOz5lxu04qHuSnv/vLLL2PJkiUp9xcvXoz+/ftn2u+dd97BqlWrsGzZMhw6dAg1atRA+/btcePGDQDA5cuX0a1bN3Tq1AmHDx/GK6+8gtEZSvgdPXoU7du3R7du3fDXX39hxYoV2LlzJxfoIyKysgMHgMqVgWXLdEeSfw6XjMTGyqqEOm6xsdZ5T1FRUfjwww8xaNCgXO3fp08f7Ny5ExcuXMDFixexa9cuvPjii+n2uXv3LubPn49p06ahY8eOqFevHr788ksUKlQIixYtAgDMnz8f1apVw8yZM1G7dm288MILmcacTJs2Db1798bw4cNRs2ZNtGjRArNnz8bXX3+Ne/fuWeT9ExFReuHhQNeu0jKycqV9FDbLCVdRs3ExMTHo3Lkz6tWrh3HjxuXqmFKlSqFz585YtmwZlFLo3LkzSpUqlW6fc+fOITExES1btkx5zN3dHQ8//DBOnDgBADhx4gSaNWuWroZM8+bN050nJCQEZ8+exXfffZfymFIKRqMRYWFhqFu3rtnvmYiIshcfDzz7LHD1KlC3rqzEaw+FzXLicMmItzdw546+17ak27dvo0OHDihSpAjWrFkDd3f3XB/bv3//lK6Szz//PNPz6r80OqcCdioXqbbRaMSgQYMwdOjQTM9xwCwRkWUpBbz6KrBnD1CsGPDzz4CPj+6o8s/hkhGDAShcWHcU+RcTE4P27dvD09MT69atg5eXl1nHd+jQAQkJCQCA9u3bZ3q+Ro0a8PDwwM6dO9G7d28AQGJiIg4ePIjhw4cDAOrVq4e1a9emO27v3r3p7j/00EP4+++/UaNGDbPiIyIi882ZAyxZAri4SMn3mjV1R2QZDpeM2JPo6GgcPnw43WMlSpRA8eLFERQUhNjYWHz77bfpFgssXbo0XF1d73tuV1fXlO6WrPYvXLgwXn31Vbz99tsoUaIEKlWqhKlTpyI2NjalMN3gwYPx6aefYuTIkRg0aBBCQkKwdOnSdOcZNWoUmjVrhtdffx0DBw5E4cKFceLECQQHB2POnDl5+FchIqKshIYCI0fK9rRpQIY1a+0akxGNtm3blmm67UsvvYR+/fph3759AJCpxSEsLAxVqlTJ1fl97tN2N2XKFBiNRvTp0we3b99G48aNsWnTJhQvXhyAdLOsWrUKI0aMwLx58/Dwww9j0qRJ6WbmPPDAA9i+fTvGjh2LRx99FEopVK9eHT169MhVjERElDsNGwLvvQdcuACMGKE7GssyqNwMDNAsJiYGvr6+KdNb07p37x7CwsJQtWpVs7syiLLD/1dEZKuUsp8Bqzldv9NyuKm9REREjsJoBGbNAuLiUh+zl0TEHExGiIiIbNQHH0iXTFCQJCaOiskIERGRDVqxAvj4Y9n+3/9kBo2jcuC3RkREZJ9CQoCXX5btt98G+vTRG4+1MRkhIiKyIRERUuo9Lg7o1AmYPFl3RNbnMMmIHUwKIjvC/09EpEN8PPDMM8A//wB16gDffw/korSU3bP7ZMRUIj3WWqvUkVMyVa/NTYE5IiJLOXtWbsWLA+vWAb6+uiMqGHZf9MzV1RXFihVDZGQkAMDb2zvTeitE5jAajbh27Rq8vb3h5mb3vyJEZEfq1wcOHAAuX3acUu+54RDftOXKlQOAlISEKL9cXFxQqVIlJrZEVCDi4wFPT9muUkVuzsQhkhGDwQA/Pz+UKVMGiYmJusMhB+Dh4QEXR55HR0Q249QpoF074PPPgSef1B2NHg6RjJi4urqyj5+IiOzGrVvAU09Jt8y0aUDnzo5dTyQ7TviWiYiI9EtKAnr2BE6fBvz9gZ9+cs5EBGAyQkREpMWoUcCmTYC3N/Dzz0DZsroj0ofJCBERUQFbuhSYMSN1+8EHdUajH5MRIiKiAnTkCDBokGx/8AHw/PN647EFDjWAlYiIyNbVqwcMHAiEhwPjxumOxjYwGSEiIipA7u7A3LlAYqLzDljNiP8MREREVqYUsGSJJCAm/61mQmAyQkREZHWTJgH9+0tNEa7DmRmTESIiIiv6+Wfgvfdku1s3gKtMZMZkhIiIyEqOHgVeeEG233hDBq5SZkxGiIiIrOD6demWuXsXaNsWmDlTd0S2i8kIERGRhSUmAs89B1y4AFSvDqxYAbhx/mq2mIwQERFZ2N9/AyEhQNGiMmakZEndEdk25mlEREQW1qgRsHcvcOUKUL++7mhsH5MRIiIiC0lOBlxdZbt+fSYiucVuGiIiIgs4fx5o0ADYsUN3JPaHyQgREVE+3b4tM2dOnADGjGFhM3MxGSEiIsoHoxF48UUZtOrnJzNnWNjMPExGiIiI8mHGDGDdOsDTE1izBqhQQXdE9ofJCBERUR6FhADvvivbs2cDTZvqjcdeMRkhIiLKgzt3gF69pMBZt24s9Z4fnNpLRESUB0YjEBgIxMUBX37JcSL5wWSEiIgoD3x8gO+/ByIjgRIldEdj39hNQ0REZIaYmNSpuwYDULas3ngcAZMRIiKiXEpKAjp1kjEiUVG6o3Ec7KYhIiLKpUmTgF27ZAG86GgugGcpbBkhIiLKhV27gAkTZHv+fKBaNb3xOBImI0RERPdx6xbwwgup1VZfeEF3RI6FyQgREVEOlAIGDwYuXpTWkM8/1x2R42EyQkRElINly2S9GVdXmcrr46M7IsfDAaxEREQ5qFkTqFwZ+N//WO7dWpiMEBER5aBlS+DIEaBIEd2ROC520xAREWUhbR0RX1/ppiHrYDJCRESUwZYt0jXz5Ze6I3EOTEaIiIjSuH4d6NsXuHsXOHhQdzTOgckIERHRf5QC+vcHwsOBOnWAmTN1R+QcmIwQERH9Z/584JdfAA8PYPlywNtbd0TOgckIERERgGPHgDfflO2pU4GGDfXG40yYjBARkdOLiwN69QLu3QM6dgSGDtUdkXNhnREiInJ67u7As88CN24AS5YABoPuiJwLW0aIiMjpubkB48cDp04BZcvqjsb5MBkhIiKnFRUFxMen3meVVT2YjBARkVMyGmWcSLNmwOnTuqNxbhwzQkRETmnGDCA4WKbvGo26o3FubBkhIiKnExICvPuubH/2mRQ4c2a6kzEmI0RE5FTu3JHumcREmUEzYIDuiPRat05WJk67MGBBy1MyMm/ePFStWhVeXl4IDAzEjh07ctw/Pj4eY8eOReXKleHp6Ynq1atj8eLFeQqYiIgoP4YNA86cASpWBBYudO5pvBs3As8/D+zdC8yapS8Os8eMrFixAsOHD8e8efPQsmVLfPHFF+jYsSOOHz+OSpUqZXlM9+7d8e+//2LRokWoUaMGIiMjkZSUlO/giYiIzPHTT8DixZKAfPstUKKE7oj0+f13oGtXICEBeO45YNw4fbEYlFLKnAOaNm2Khx56CPPnz095rG7duujatSsmT56caf+NGzeiZ8+eOH/+PErk8VOPiYmBr68voqOj4ePjk6dzEBERhYUBL7wAtG0LfPih7mj02b5dKs3GxQFPPy1Jmru75V8nt9dvs7ppEhISEBISgqCgoHSPBwUFYffu3Vkes27dOjRu3BhTp05FhQoVUKtWLbz11luIi4vL9nXi4+MRExOT7kZERJRfVasCf/6ptxVAt927gc6dJRHp1AlYscI6iYg5zOqmuX79OpKTk1E2Q3m6smXLIiIiIstjzp8/j507d8LLywtr1qzB9evX8dprr+HGjRvZjhuZPHkyJkyYYE5oRERE2bpwAahSRbbdnLioxYED0iJy9y7wxBPAqlWAp6fuqPI4gNWQYbSPUirTYyZGoxEGgwHfffcdHn74YXTq1AkzZszA0qVLs20dGTNmDKKjo1Nuly9fzkuYRERECA4GatQA3npL/xRWnUJDgaAgICYGaN0a+PlnwMtLd1TCrPywVKlScHV1zdQKEhkZmam1xMTPzw8VKlSAr69vymN169aFUgr//PMPatasmekYT09PeNpCqkZERHbt0iWZxpucDNy6Bbg4aUGLo0eBdu3k36BFC2D9ein2ZivM+lg8PDwQGBiI4ODgdI8HBwejRYsWWR7TsmVLXL16FXfu3El57PTp03BxcUHFihXzEDIREdH93bsndUSiooDAQGDuXN0R6XHypHTJREUBTZoAGzbY3ho8ZueII0eOxFdffYXFixfjxIkTGDFiBC5duoTBgwcDkC6Wvn37puzfu3dvlCxZEi+//DKOHz+OP//8E2+//Tb69++PQoUKWe6dEBERpTFsGHDwoEzfXbXKdrokCtKZM0CbNkBkJPDgg8CmTUCajgqbYfYwnh49eiAqKgoTJ05EeHg4AgICsGHDBlSuXBkAEB4ejkuXLqXsX6RIEQQHB2PIkCFo3LgxSpYsie7du+Ojjz6y3LsgIiJKY/Hi1IJmP/wA/HeJciphYZKIhIcDDRoAmzcDxYvrjiprZtcZ0YF1RoiIKLf++UcGrMbHAx99BIwdqzuignfpEtCqFXDxIlC3LrBtG1CmTMHHkdvrtxNPcCIiIkdUsSKwZImsuTJmjO5oCt6VK9IicvEiULOmVFrVkYiYgy0jREREDiIiAnjsMeDUqdQCbzrnililAisREZGtWrZMLsbO6to1mTVz6hRQqRKwdaveRMQcTEaIiMjurV8P9OsnU3ivX9cdTcG7cUPqiPz9N1C+vCQipoqz9oDJCBER2bVz54A+fWT7mWeAUqX0xlPQbt2SyqpHjgBly0oiUr267qjMw2SEiIjsVmysFDa7dQto1gyYMUN3RAXr9m1ZayYkRJKw338HatfWHZX5mIwQEZFdUgoYPFhaBMqUAX76CfDw0B1Vwbl7V1bd3btX6ods2QLUr687qrxhMkJERHZpwQLgm28AV1dgxQr7GaxpCXFxwFNPATt3SkXV4GCgYUPdUeUdkxEiIrI7SUmSjADAlCkyndVZ3LsnY2O2bpU1ZjZulIG79oxFz4iIyO64uUmrwFdfAcOH646m4CQkAM8/L2vMeHsDv/0mY2XsHZMRIiKyS0WLAiNG6I6i4CQmAj17yjRmLy/5+cgjuqOyDHbTEBGR3XjvPZkxY/u1wy0rKUmmL69ZI4N0f/4ZePxx3VFZDltGiIjILqxeDXz8sWw3bQq0bKk3noKSlAT07y+DdN3d5d8hKEh3VJbFlhEiIrJ5J09KhVUAePNN50hEYmOBefOkbkjaWUOdO+uOzPLYMkJERDbtzh2gWzcp8NW6tcyecWRRUcDnnwNz5qSWti9ZEvjiC5lF44iYjBARkc1SChgwADhxQtZcWbFCZtI4ogsXZDzMokXSKgLIyrtvvgm8/LLMnnFUDvqREhGRI5g1C/jxR0lAfvpJ1l5xNKGhwLRp8j6Tk+Wxhx4C3nlHSt07avKVlhO8RSIisldubnKbORNo0UJ3NJajlKwjM3WqVE81CQqSJKRNG8Bg0BdfQWMyQkRENmvIEOCJJ4A6dXRHYhlJSdLCM22atIgAMjC1Rw/g7beBRo20hqcNkxEiIrIpCQlS8tzHR+7Xras3Hku4exdYvFjGhFy4II95ewOvvCKF26pU0RmdfkxGiIjIprz1lpQ7X73aflehNbl2DZg7V243bshjpUoBQ4cCr70ms2SIyQgREdmQ77+XKa0AcO6c/SYj585JK8jixdLKAwDVqkmi1a8fUKiQ1vBsDpMRIiKyCUePAgMHyvbYscBTT+mNJy8OHpTxICtXAkajPNa4sQxK7dZNxodQZkxGiIgcTFKS/U0HjY6Wi3VsLNCuHTBhgu6IzPPHH8BHHwFbt6Y+1qGDJCGPPeZcM2PyguXgiYgcyE8/SY2Ka9d0R5J79+4Bzz0HnD0LVKokXTX21ILwyy8yFXfrVkkCX3wROHIE+O03WcyOicj9MRkhIrJzt28D27bJRX3UKOnu6NgRiInRHVnujB0LbNkCFC4MrFolAzztxeXLqWvm9OghY0W++QZ44AGtYdkdJiNERHZu0SL5C/zll+Wv8VKlgJAQoGvX1MGTtmzsWKBVK2D9ehlfYS+SkoAXXpBZMoGBwLJl0rJD5mMyQkRkx5KSpDopIAlJ7drAxo1AkSIyjqFXL9nHlpUoIS07jz2mOxLzfPghsGMHULQosHw54OmpOyL7xWSEiMiOrVwJXLoElC4N9OkjjwUGAuvWAR4ewNq1wKBBUn7cViglNTbmzk19zN7GVfzxhyQjgKymW6OG3njsHZMRIiI7pZRMIwWAN95IX7vi8cflr3UXF6l18eOPemLMSCmptTF/PjBsmKzGa2+uXZNBqkoB/ftL6xPlj51N/iIiIpPt24FDhwAvL2lpyOiZZ4AvvwSOHQOef77g48vKuHFSDAwAFi60v1LvRqMMWL16VdbLmT1bd0SOgckIEZGdmj5dfr78cvYzUPr3L7h47ueTT1K7NmbPBgYM0BtPXsyaBWzYIONDVqyQGUCUf+ymISKyQ3fvSl0Og0EWWsuN+HjpUlizxrqxZWXOHGD0aNmeMkVW47U3Bw6kvoeZMzl915KYjBAR2aHChYHjx4Hdu4GaNXN3zPz5Mo6kZ08ZgFlQQkJkYTgAeP99qYVib2Ji5N8tMRF49llg8GDdETkWg1K2NMY6azExMfD19UV0dDR8TGtKExGRWZKSgO7dpWWkaFFJSAIDC+a1J08GoqJkwK29zZxRCujdWxK5ypWB0FCgeHHdUdmH3F6/mYwQEdmZkBBZzdbLy/xj790DOnWSRKRUKWDnTqlNYg1KpU88Mt63F4sXy/gWV1epK9K8ue6I7Edur9/spiEisiN37wJBQfIX+qlT5h/v5SW1RwIDgevX5Vz//GPxMLFxI9C+vZSqN7HHROT4cZk2DchCeExErIPJCBGRHVm6VMqPFy2a90JbPj5SNr5WLSmY9vTTqcvdW8K2bTKtODg4dcaPPYqLk/Vm4uJkJeF33tEdkeNiMkJEZCeSk1NrdIwcmb+VbUuXlmQhIECmq7pY6GqwZw/QpYt0B3XpIuvO2KuRI6VGS9mysvidpf6NKDPWGSEishNr1wLnz8taLqaVYvOjUiVZ6t5SF9lDh2S14Lt3gSeeAH76SUrS26OVK4EFC2T7m28kISHrYZ5HRGQH0pZ+f+01wNvbMudNm4j89Rfw+uvSAmOuv/+W8SfR0cAjj0jilJcBtrbgwgXglVdke/Ro6aIh62LLCBGRHdi9G9i3T1oaTAMqLenuXbnoRkZKMjJ/fu4HnBqNUoMjKgpo0gT49Vf7rUyamCiF4aKjZbDqxIm6I3IObBkhIrIDpiJlfftap8ugcGFZRddgkFVo338/98e6uAA//CAtIxs3ygBZe/X++8DevUCxYsD33wPu7rojcg6sM0JEZCeOHQOKFAGqVLHea3zxRWp10Rkzci41b691Q7KzebNMRwZkzMizz+qNxxGwzggRkYMJCLBuIgIAgwYBH38s2yNHAl9/nfV+kZFAs2aycrAjiIgA+vSR7VdfZSJS0JiMEBHZsJs3rVOULCdjxqS2iPTvD/z+e/rnb9yQLpn9+4H//U/KzNszo1ESkchIWfzu0091R+R8mIwQEdmwOXOAqlVTWysKgsEgxcr69gVatwYefjj1uZgYmb575IiMXfnlF8DNzqdCfPIJsGWLzFBavhwoVEh3RM7Hzv8LERE5rrg4GVSalJT3aqt55eICfPWVtBp4espjsbFSyGz/fqBkSbmA16pVsHFZ2u7dqYN1584F6tbVG4+zYjJCRGSjvv0WuHZN1qHRMYYh7UwSozF1uq6PD7Bpk4xhsWc3b8o03uRkWZXXEoXkKG/YTUNEZIOMxtSxC8OH6+8KmTNHfhYuLOvaBAbqjSe/lJLCZpcuAdWrm1dXhSyPLSNERDbo119lVV5fX1m+XrdevYB//wW6dk0/hsReLVgArF4trT/Ll9t3bRRHwGSEiMgGmVa7HTxYVujVrUwZYNIk3VFYxpEjqbOFpk4FGjfWGw+xm4aIyOZcuQKEhEjXzJAhuqNxLHfvAj16APHxMhh32DDdERHAlhEiIptToYKMZdi5U7bJcoYMke6v8uWBJUs4TsRWsGWEiMgGlSgBPPWU7igcy3ffSQLi4iLrzpQqpTsiMmEyQkRkQy5d0h2BYzp7NnXNnQ8+kGJuZDuYjBAR2YibN4F69YAWLYDr13VH4zj+/ht48kngzh1JQt57T3dElBGTESIiG/HFFzLA8u5dqXBK+aOUTOFt3Bg4eRLw85NCcq6uuiOjjJiMEBHZgPh4YPZs2X7rLQ6szK+oKKBbN1mB9949oEMHIDQUqFhRd2SUFSYjREQ24IcfgPBwmT3To4fuaOzbtm1Aw4bA2rVS1OzTT6WIXNmyuiOj7DAZISLSTKnUImfDhgEeHnrjsVdJSbLoXZs2UqulVi1g715g5EiZQUO2i3VGiIg027RJBlkWLQr873+6o7FPFy7IYnd79sj9l1+Wbq8iRbSGRbnEXJGISLPvvpOfAwfKWjRknuXLpVtmzx5ZY2b5cmDxYiYi9oQtI0REmi1ZIgXOmjXTHYl9uXNHKqouXSr3mzeXYmZVquiMivKCyQgRkWZubsDzz+uOwr4cOgT07AmcOSPjQcaOlWJmbryq2aU8ddPMmzcPVatWhZeXFwIDA7Fjx45s9922bRsMBkOm28mTJ/McNBGRI4iJARISdEdhX4xGmR3TrJkkIhUrAlu3AhMnMhGxZ2YnIytWrMDw4cMxduxYhIaG4tFHH0XHjh1x6T41jE+dOoXw8PCUW82aNfMcNBGRI5g4EahaFfjxR92R2IeICKBjR6nDkpgIPPMMcOQIS7s7ArOTkRkzZmDAgAF45ZVXULduXcyaNQv+/v6YP39+jseVKVMG5cqVS7m5sgQeETmx6Ghg4ULg6lUOtMyN336TQaqbNwNeXlJZddUqWVCQ7J9ZyUhCQgJCQkIQFBSU7vGgoCDs3r07x2MffPBB+Pn5oW3btvjjjz9y3Dc+Ph4xMTHpbkREjuTLL4Hbt2Utmg4ddEdju+LjpU5Ip05AZCTQoAFw8CAwaBCr1DoSs5KR69evIzk5GWUzlLErW7YsIiIisjzGz88PCxcuxKpVq7B69WrUrl0bbdu2xZ9//pnt60yePBm+vr4pN39/f3PCJCKyaQkJwKxZsv3mmyzIlZ1Tp2SGzMyZcv+NN4D9+4H69fXGRZaXp+E+hgzpqFIq02MmtWvXRu3atVPuN2/eHJcvX8b06dPRqlWrLI8ZM2YMRo4cmXI/JiaGCQkROYwff5QKoWXLAi+8oDsa26OU1AkZOhSIjZVFA5cskZV3yTGZlYyUKlUKrq6umVpBIiMjM7WW5KRZs2b49ttvs33e09MTnp6e5oRGRGQXlJLZIIBcbPlVl96tW9IFYxrU26YN8M03QPnyWsMiKzOrcdDDwwOBgYEIDg5O93hwcDBatGiR6/OEhobCz8/PnJcmInIIISHA4cOAtzcweLDuaGzLrl1Ao0aSiLi5AZMny4BVJiKOz+xumpEjR6JPnz5o3LgxmjdvjoULF+LSpUsY/N9v1ZgxY3DlyhV8/fXXAIBZs2ahSpUqqF+/PhISEvDtt99i1apVWLVqlWXfCRGRHWjcWAp2HTvGmSBpLV0KDBggdUSqVZNVjB9+WHdUVFDMTkZ69OiBqKgoTJw4EeHh4QgICMCGDRtQuXJlAEB4eHi6miMJCQl46623cOXKFRQqVAj169fHr7/+ik6dOlnuXRAR2ZEHH5QbiUuXZHCq0SiL3c2fL2vMkPMwKKWU7iDuJyYmBr6+voiOjoYP/4cSkZ26c4c1RTJSStblWb8eaNkS+PNPzi5yJLm9fvMjJyIqAIcPy+yZgQNZAj6t1aslEXF3lyJwTEScEz92IiIrU0q6IWJjpdCZh4fuiGxDdLSsugsAo0ZJAThyTkxGiIis7LvvZKaItzcwfbruaGzHu+8C4eFAzZqy6i45LyYjRERWFBMDvP22bL/3nqwyS8CePTJQFZB1Zry89MZDejEZISKyog8/lNVma9SQNVZIVtz93/+k++qll6SwGTk3JiNERFZy8mTqGjSffcZqqyaffip1VkqWZLcViTytTUNERPd38qSME2nVSladJeDcOWDCBNmeMQMoVUpvPGQbmIwQEVlJ167A6dOcymuiFPDqq8C9e0DbtkCfProjIlvBZISIyIrMWEPU4X3/PRAcLN1V8+cD2Sz2Tk6IY0aIiCxswQJg40bdUdiWGzeAESNk+/33ZTovkQlbRoiILCgsDBg+HIiPl9Lmjz6qOyLb8PbbwLVrQP36qVOdiUycumUkOVm+LKKjdUdCRI5i5EhJRNq0AR55RHc0tmH7dmDxYtn+4gtWoKXMnDoZeeIJoHVrYN063ZEQkSPYtAlYuxZwdQVmz+aYCEASs0GDZHvQIFkMjygjp05GTM2na9bojYOI7F9CAjB0qGwPGSLdEQRMmQKcOiUDeSdP1h0N2SqnTka6dZOfGzfKAlZERHn12WcyjbdMGWD8eN3R2IaTJ4FJk2T7s8+A4sX1xkO2y6mTkYYNgSpVgLg4aV4lIsqL69eBiRNl+5NPAF9fvfHYAqWAwYOlxahjR6B7d90RkS1z6mTEYACeeUa2V6/WGwsR2a+SJYFvvwV69gT69tUdjW1YskQGrnp7A/PmcfwM5cypkxEgtatm/XpZvImIyFwGA/D008APPwAuTv+tCkRGAm+9JdsTJkgLNFFOnP7Xpnlz6eO9dQvYtk13NERkT5KSgKgo3VHYnpEjgZs3gUaNpOYK0f04fTLi6irrRwDsqiEi83zxhVQSXbZMdyS2IzgY+O47aSFauBBwY2lNygWnT0aA1HEja9cCRqPWUIjITly7Brz3nrQA3L2rOxrbEBsrg1YB4I03gCZN9MZD9oPJCKRSoo8PEBEB7N2rOxoisgdjx0r3bsOGqUW9nN2HHwLnzwMVKwIffaQ7GrInTEYgpYm7dJFtFkAjovs5eBD46ivZnjtXunud3dGjwPTpsj13LlC0qN54yL4wGfmPaVbN6tUyP56IKCtGo3RBKAW8+CLXnwHk3+R//5MBvc88IzOLiMzBZOQ/HToAXl7SxHj0qO5oiMhWff01sG8fUKQIMHWq7mhsw4IF0sVdtCgwZ47uaMgeMRn5T+HCQFCQbLOrhoiyc+6c1BUZNw7w89MdjX5XrwJjxsj2pElAhQp64yH7xGQkjbRdNUREWfnwQyA0NHVRPGc3bBgQEwM8/DDw6qu6oyF7xWQkjSeflIFof/0lf/0QEWWlYUMZ+O7s1q8HVq6U782FCzmQl/KOyUgaJUoAjz0m2+yqISITpYB335VVaEncuQO8/rpsjxwpCRpRXjEZycBUAI3JCBGZrFwJTJ4MNG0qXRIEfPABcOmSrDszbpzuaMjeMRnJwFQafvduIDxcayhEZAPu3gXefFO2R4yQAonOLiQE+Owz2Z4/XyYAEOUHk5EMKlSQv34A4Oef9cZCRPpNngxcviwtAKNG6Y5Gv6QkqSliNAI9e0pZBKL8YjKSBXbVEBEAnD0LTJsm2zNmAIUK6Y3HFsyZAxw6BBQrBsyapTsachRMRrJgSka2bpVFsIjIOY0YASQkSA0iUxeuM7t0CXj/fdmeOhUoW1ZvPOQ4mIxkoVYtoH59aY789Vfd0RCRDsHBMnXVzU3GRxgMuiPSKz5eZs/cvSsl8AcM0B0RORI33QHYqm7dgL//lgJoL76oOxoiKmitWwOffALExQF16uiOxvKMRmn5jYxMf/v336wfM80icneXmiIu/FOWLMiglO0vCxcTEwNfX19ER0fDp4CGsoeGAg89JH3E168D3t4F8rJERHmWmCjl2XOTYFy7Jq2/5ihUSMbODB5snfjJ8eT2+s2WkWw0aiSj5y9cADZtSh1HQkSO7eZNWQTP3V13JOb5+Wdg0CBJPMxRrJiM/ShTJvMt4+PFirG7iqyDyUg2DAZJQGbOlFk1TEaInMPgwcCxY8DixanT/G3ZnTtSAfXLL+W+h0fuk4vSpVnWnmwDk5EcdOsmycgvv0jzp739pURE5vn9d+DHH2U8hKen7mjub98+GdN29qz8AfX228DEifYRO1FaHIKUg+bN5a+HW7eAbdt0R0NE1hQWBvTqJduDBklXra1KSpKko2VLSUT8/aUUwSefMBEh+8RkJAeursDTT8s2C6AROa5bt4DOnWVQZ6NGUkPDVp07B7RqJevBJCdLFdQjR1IX+SSyR0xG7qNbN/m5dq1MhSMix5KYCDz3HHDihCwHsX69DGC1NUoBS5ZIsrRnj6yR8913wA8/AMWL646OKH+YjNxHmzbySx8eLv2zRORYPvxQxooULiyJSIUKuiPKLCpKEqb+/WXAaqtWwF9/Ab17646MyDKYjNyHhwfQpYtsr16tNxYisrzhw+WPjuXLbXOcyObNQIMG8v3j7g5MmSLjQypX1h0ZkeUwGcmFtAvn2X6JOCIyR4kSwJYtqX902Iq4OEmU2reXltk6dYC9e2XlYFdX3dERWRaTkVzo0AHw8pKBY8eO6Y6GiPJr3z7giy9S79taIa8jR4AmTWRNHEDWhAkJkarQRI6IyUguFCkiq3YC7KohsncXLgBPPSXFzZYs0R1NekYjMH068PDDsjZW2bKyWOfcuVySghwbk5FcSttVQ0T2KTpapvBGRgING8qgUFtx+TLwxBNSuCwhQcoKHD0KdOqkOzIi62MykktPPin9tEeOAOfP646GiMyVmAg8/zxw/DhQvrzMnClaVHdUYsUK4IEHgD/+kBaQhQvlD5/SpXVHRlQwmIzkUsmSsqQ4wNYRInujlIy7CA5OncJbsaLuqKSlpk8fKVx265Z0zxw+DAwcaHvjWIisicmIGUwF0DhuhMi+TJ8uC8m5uEiRsAcf1B0RsGOHdBV9+63E9cEHwM6dQM2auiMjKnhMRszQtav83LMHiIjQGgoRmcHUyjBjhnS56pSQALz7rrS0XrwIVKsmSciECVyMk5wXkxEzVKggS4orBfz8s+5oiCi33noLCA0Fhg7VG8fp07IA5+TJ8j3y8svSLdO8ud64iHRjMmIm06wadtUQ2bZ//gFu306936iR3nEYcXFSIuDQISm0tnIlsHix7QyiJdKJyYiZTMnI1q0y4IyIbE90tFQufeQRmTJrC2bNkm4Zf3+Zsvvss7ojIrIdTEbMVKsWUL8+kJQkI/KJyLakncJ7/bptzEqJjJSuGQCYNEmmFhNRKiYjecACaES2SSngjTdkCq+3N/DLL7YxhXfcOOkyatyYK+0SZYXJSB6Ypvj+9hsQG6s3FiJK9emnUjDMYJApvLawlsvx4xITIPG58FuXKBP+WuRBo0ayfHdcnCzvTUT6rV4NvPOObM+cKevP2IK335Y1Z7p2BVq10h0NkW1iMpIHBkNq6wi7aoj0S0wERo1KrbSqewqvyZYtwIYNgJsb8MknuqMhsl1MRvLING5k3Tr5IiQifdzdZV2XESNk1ootDFpNTgbefFO2X3tNBr8TUdaYjORRixZAmTIyvXf7dt3REDknpVK3K1aUCqtubvriSevrr4G//gKKFZNS70SUPSYjeeTqKkt8AyyARqRDUpJ0l/70k+5IMrt7Fxg7Vrbfe08W2iSi7DEZyQdTV83atTJAjYgKhmkK79q1QP/+wLVruiNKb/p0IDwcqFpV4iSinOUpGZk3bx6qVq0KLy8vBAYGYseOHbk6bteuXXBzc0OjRo3y8rI2p00bwMdHvnT27dMdDZHzmDED+OILGRvy7bdA6dK6I0p19SowdapsT5kCeHrqjYfIHpidjKxYsQLDhw/H2LFjERoaikcffRQdO3bEpUuXcjwuOjoaffv2Rdu2bfMcrK3x9AQ6d5ZtzqohKhhr18p0WUDqdpi6S23F++9L/aHmzaUSLBHdn9nJyIwZMzBgwAC88sorqFu3LmbNmgV/f3/Mnz8/x+MGDRqE3r17o7mDLU+Zdopv2sF0RGRZR48CXboAPXvK79qrrwLDh+uOKr0jR4AlS2T7009tY1YPkT0wKxlJSEhASEgIgoKC0j0eFBSE3bt3Z3vckiVLcO7cOYwbNy5XrxMfH4+YmJh0N1vVoYO0kJw9Cxw7pjsaIvsXEwNs3CgDP9MODvfyAn79FYiPl4Jms2fb1sVeKeCtt+Rn9+7SMkJEuWPWJLjr168jOTkZZcuWTfd42bJlERERkeUxZ86cwejRo7Fjxw645XLO3eTJkzFhwgRzQtOmSBFZHXTdOmkdadBAd0RE9iU8HNixA9i5U37+9VfqgPBnn01tfaxRA5g7F2jaFAgMtK1EBJAEassWwMNDxooQUe7laQCrIcO3gFIq02MAkJycjN69e2PChAmoZUbFnzFjxiA6OjrldtlW1gDPhmlWDaf4EuVMKSAqKvX+vXuytEKPHsCcOcDhw5KIVKsGvPRS+jEXBoNUV23c2PYSkaQkaRUBpPpr1ap64yGyN2a1jJQqVQqurq6ZWkEiIyMztZYAwO3bt3Hw4EGEhobijf/mtxmNRiil4Obmhs2bN6NNmzaZjvP09ISnHQ1Bf/JJqTty5Ahw/rx8kRKRXKRDQ1NbPXbuBCpVAg4elOe9vICHH5a6HI8+CjzyiNzKl9cbt7kWLZIF8UqUSK0vQkS5Z1Yy4uHhgcDAQAQHB+MZU3MAgODgYDydxZB2Hx8fHD16NN1j8+bNw9atW7Fy5UpUdZA/H0qWBFq3BrZula4aUwloImc1dy7w88/Anj2SaKQVGystIl5ecn/bNtupmpoXMTGpFVbHjZOKq0RkHrO/AkaOHIk+ffqgcePGaN68ORYuXIhLly5h8ODBAKSL5cqVK/j666/h4uKCgICAdMeXKVMGXl5emR63d888w2SEyGTXLhk/AcjFuWXL1JaPxo3T196w50QEkAXwIiOBmjWB/74GichMZn8N9OjRA1FRUZg4cSLCw8MREBCADRs2oHLlygCA8PDw+9YccURduwJDhgC7dwMREUC5crojItKnf3+gVStJPurXB1wctNbz5ctSgA2QQmceHnrjIbJXBqVsvzpGTEwMfH19ER0dDR8fH93hZKtpU2D/fmDBAmDQIN3REBWcv/4CFi+W2hqurrqjKTh9+kgF2FatpLvJ1gbWEumW2+u3g/69okfaAmhEzuLsWSAoCPjsM+Cjj3RHU3AOHpREBGCBM6L8YjJiQaYxvb//Dty6pTUUogJx5QrQrh3w779Aw4bAsGG6IyoYpgJnAPDiizIOhojyzumTkfh4GeNhCbVqAfXqyXTGX3+1zDmJbFVUlLSIXLggBck2bXKemSTr1gHbt8uMoEmTdEdDZP+cOhlZvVrqGQwdarlzmrpqWACNHNmdO0CnTlJbo3x5IDgYyKLUkENKTATeeUe2R44E/P31xkPkCJw6GaleHbhxQ1YBvX7dMuc0ddVs3AjExVnmnES2xLT2yv79UuQrOBioUkV3VAVnwQLg9GmgTBlg1Cjd0RA5BqdORho2BB56SP7S+e47y5zzwQelvHVsLLB5s2XOSWRLTGXZS5cGfvtNuiadxa1bgGnZrAkTABue3EdkV5w6GQGAAQPk56JF8hdffhkMXKuGHF/nzkBYmJRydyYffyxjZerVA155RXc0RI7D6ZORXr2kGuTRo0BIiGXOaRo38ssv0upCZO+UkkqjZ8+mPla4sL54dAgLA2bPlu1p0+y/ciyRLXH6ZKR4cVmmHJCiTZbQooU0Yd+8KSPuiezdlCnA6NFS0j06Wnc0eoweDSQkAE88AXTsqDsaIsfi9MkIIKWrAeD772Wqb365ukp5eIAF0Mj+LVgAvPuubL/zDuDrqzceHfbsAX78Ubphp09ngTMiS2MyAuDxx4GJE+ULJ+0CXvlhGjeyZg1gNFrmnEQFbfly4LXXZPu994ARI/TGo4NSqYtfvvyyDHwnIsvi2jRWEh8vU/9iYmSmTu/euiMiMs9vvwFPPSVF/F57DZg71zlbBH76SaYye3sDZ85IXRUiyh2uTaOZp2dqYaQ333TefnayT/v2yViqpCQZ5D1njnMmIvHxqbVE3n6biQiRtTAZSePgQaBHD+DDDy1zvrfekhLxERHA++9b5pxEBaF6dSAgQKqsLlsGuDjpN8XcuTKLxs9PkhEisg5206Tx44+SjFSoAFy8aJml0LdskYXEXFyAAwekyBqRPbh9W34HvL11R6JHVJSsuXPrltQhMg10J6LcYzdNHjz9tJS3vnLFctVTn3gC6NlTBrG++ioHs5LtunIF+Oab1PtFizpvIgLIoPZbt4AHHgBeekl3NESOjclIGp6eshw4YLmaIwAwY4Z8se/fD3z5peXOS2QpphV4+/aVqbzO7vRpYN482f70U8u0khJR9piMZGBqiv35Z+DaNcuc088P+Ogj2R4zBoiMtMx5iSwh7Qq8FSoAHTrojki/UaNk8G6nTtK6SUTWxWQkg4YNgcBAyy6eB8jUyEaNpCorV/okWxEfLwX6TCvwbt7sXCvwZuXPP2Ulb1dXKftORNbHZCQLll48D5B1LObPl+mRS5cCO3ZY5rxEeZWUJPVvfv8dKFLE+VbgzYrRmFrgbOBA/nsQFRQmI1no1UtaR/r1A5KTLXfeZs3kCw6QlhIuoke6KAUMGiQrS3t4SLeks63AaxIdDfzxBzB1qhR5O3hQkrPx43VHRuQ8uO5kFooVky8ka5g0SS4Ax44Bn30mtUiIdKhRQ7oiVqwA2rTRHU3BiIsDQkPl9/vAAbmdOpV5vwkTgLJlCz4+ImfFOiMaLFkiA2ULFwZOnAD8/XVHRM7q9GkpzGdJV6/KrVw5uaC7u1v2/LmVmAgcPSoJhyn5OHYs69bOypWBxo2BJk2ARx4BWrYs+HiJHFFur99MRnJw9y6wapXMhmnXznLnNRqB1q2BnTul5PbKlZY7N1FOgoOB5s2lG8IawsJkoHZMjNw3GIBSpeR3yM9PEhTTdsbHChfO++smJ0sLh6m14+BB4PDhrFfhLltWkg5T8tG4sawjRUSWx2TEAj7+WFYqfewx6VO2pKNHgQcflC/RDRuAjh0te36ijDZvBjp3lmRhyxbA19ey5zcagbZtgW3bJLGIj5dBsrlVtGjWSUrGx4oXBy5cSE08DhwADh2SKcoZFSuWPulo0gSoWNE519kh0oHJiAVcuiTTHJUCzp6V9Tos6a23pKBStWrSfFyokGXPT2QSGgq0aiUX7F69gG+/tfx6M/PmAa+/LlVb//oLqFoVuH5d1mYKD898S/t4bGzuX8fFJetKxt7estxCkyapyUeNGkw8iHRiMmIhHToAmzYBY8emFi6zlNu3gbp1pQz3Bx/IoDkiS7t4UWZyRUQAjz8uU3g9PS37GmFhQIMG0rU5ezYwZEjuj1VKfheySlIyJi83bsgxHh5SE8jU2tGkifwusVIqkW1hMmIhP/0EdO9u2cXz0lq1CnjuOflyPXYMqFnTsucn53bjhgzGPHlSkoUdO6zbPdO6NbB1q/VW+Y2Pl8rIpUtbPqEiIsvjQnkW8tRTQMmSll08L61u3aT1JSFBmrhtPzUke3Hvniz+ePKkjJPYsMHyiQggxfy2bZNuksWLrZeIAJKAVKzIRITI0TAZuY+0i+ctWmT58xsMwJw58jrBwdISQ2QJFy8CZ85IAvLbb3IRt7Tz54F33pHtTz6R8U9EROZiMpILpsXz7t7NeuBcftWoIQvoAcDw4anTIonyo3ZtYM8eYP16ICDA8uc3GuV3IzZWZpy99prlX4OInAOTkVx44AHg8mX569JaTdCjRklSEh7OMtSUP2lXha5aVYp4WcO8ecD27dI9s2iRdbtniMix8esjl6zRxJ2Wlxfw+eeyPXs2cOSIdV+PHNPy5dJVsn69dV/n3LnU1afZPUNE+cVkxEwREcA//1jn3EFBwPPPSyG0V1+1TpcQOa5t24CXXpLuxN9/t97rsHuGiCyNyYgZZs2SFhJL1xtJa+ZMKdW9Z4+sYUOUG0ePAl27yqysZ58Fpk+33mvNmwf8+adUWWX3DBFZAr9GzNCwobRa/PCDeRUjzVGhAjBxomy/845UsCTKyT//yHIC0dEyPuTbb61X/IvdM0RkDUxGzNC6tXz5xsRIsTJrGTJEBs3euAGMHm291yH7Fx0ticiVK1KB9OefZfyRNWTsnnn1Veu8DhE5HyYjZnBxAV5+WbatUXPExM1NmsJNr7N7t/Vei+zbrFlSudfPT2Z7lShhvdf6/PPU7hlrFzcjIufCrxMz9esnhcq2b5fF86ylZcvU+iavvmre6qfkPMaOBUaMkOqqlStb73XOnUttpZs6VaYMExFZCpMRM1WsCLRvL9vWHmD6ySfyl+5ffwFz51r3tci+mJYNcHMDZswAGjWy3mul7Z55/HFg8GDrvRYROScmI3kwYID8XLHCumvJlColCQkAvP++jAsgmjNH/g8mJhbM682dy9kzRGRd/FrJgyeflMXBDhyQLhtr6t8faN4cuHMHGDnSuq9Ftm/1amDYMGmVW7nS+q939mxq98y0aeyeISLrYDKSB56e0lRdvLj1X8vFRRIfFxfgxx+ts3Iw2Yddu4AXXpDWuFdfBXr2tO7rmbpn4uKke2bQIOu+HhE5LyYjFmDtSqkNGwJDh8r266/L0vDkXE6elBa5e/eAp56Srhprt8rNnQvs2MHuGSKyPn695MO6dUDTpjK90tomTJDpm2fPymwGch4REUCHDsDNm0CzZlJ0z1pFzUzYPUNEBYnJSD5cvQrs3y9/NVpzICsA+PhIqXgAmDRJplqS4zMapcz7xYtAzZrAL7/IKrnWfk1T90ybNuyeISLrYzKSD716SbXL48clKbG27t2Bdu2A+HjgjTesnwCRfi4usjxA9epS1KxUKeu/5pw50j1TpAi7Z4ioYPBrJh98fYHnnpPtxYut/3oGg/Tje3gAGzfKzApyfEFBwIkTkpBY25kzwJgxsj1tGlClivVfk4iIyUg+mWqO/PCDLN1ubbVqpS5UNmwYcPu29V+TCt6sWcCpU6n33d2t/5ppu2fatmX3DBEVHCYj+dS6tfzFevu2dRfPS2vMGFmw78oVKYZGjuP2bUkyR4yQJQGuXSu41549G9i5M7V7xtqzdYiITJiM5JPBkLp4XkF01QBAoUKp5eE/+0zGFHD8iH1TSrrd6taVpACQ6dylSxfM6585A7z7rmxPn27ddW6IiDJiMmIB/foBnTun1gIpCB07yqwaABg3TlpLmJDYpwsXpHbIs89Ka1e1ajIm6IMPCub1k5MloTZ1z/zvfwXzukREJm66A3AEFSoA69cX/OuOGSOtJCNGyBo2sbEy1oCzH+zHxYtA/fry2bm7y3igd9+Vz7WgzJkj1V3ZPUNEujAZsXPDh8uF69VX5aISFwcsWGD9olhkGZUrA506ydiQ+fOlm6YgsXuGiGwB/4a2oEuXZPzGH38U7OsOGgQsXSotIl99BfTtCyQlFWwMlDs3bgBDhgDh4amPLV0q/2cKOhFJ2z3zxBPsniEifZiMWNDMmTJ+Y86cgn/tvn2B5csBNzfg+++BHj2AhISCj4OyphTwzTdAnToy+DjtCsyFC+vpGpk9W7pnihaVJJbdM0SkC5MRCzLVHPnlFyAysuBf//nnZUaGh4f8fOYZ+auX9Dp1SgaG9u0r3TH168uChzqdPs3uGSKyHUxGLCggAGjSRLpIvvlGTwxPPimDaQsVAjZsALp0KZhibJRZXJzMiHngAemGKVQImDIFOHQIeOQRfXGZumfu3ZPlBQYO1BcLERHAZMTiTK0jixfrm2rbrp1MDS1SBNi6FWjfHoiO1hOLM5s+HfjwQ+ku69QJ+PtvmS3j4aEvJqMRGD8e2L2b3TNEZDuYjFhYz57yF/Dx48C+ffriaNUK2LIFKFZMxgU88YQMnqSCM3w40LQpsHKltFZVrao3nvPnpbvoo4/k/qefApUq6Y2JiAhgMmJxBb14Xk6aNpWWkVKlgIMHgccf1zOWxRkkJwPz5kk3mdEojxUtCuzZI8XMdLY+GI0yqLpBA2DbNsDbWwbRvvKKvpiIiNJiMmIFAwbIDAlvb92RAA8+KBegcuWAv/6StXSuXNEdlWMJDQVatJBBqevXA2vWpD6nuwvk7FlJQocOlcJqjz0GHD0qseqOjYjIhMmIFbRqBURESDVUW1C/PvDnn4C/P3DypMR38aLuqOzf7dsyRbdxY2D/fsDHR1ocunbVHZm0hnz2mQye/fNPSY7nzQN+/13KzRMR2RImI1ZgMMjgUVtSs6ZclKpVk7EDjz4q1TfJfKZF7erVk9oyRqPUdTlxQlocdFe/PX1aEs7hw2VGT5s2wLFjUqWXSwUQkS3iV5OVffWV/rEjJlWqSEJSuzZw+bJcsI4f1x2V/TEYgBkzgH/+keTut9+k4Fz58nrjSk6WuBo2TF1r5osvZCBzlSp6YyMiykmekpF58+ahatWq8PLyQmBgIHbs2JHtvjt37kTLli1RsmRJFCpUCHXq1MHMmTPzHLA92bdPSmwPGAB8/rnuaESFCsD27TKYMSJCxpCEhuqOyrbFxwNffglERaU+9sEHwHvvSYtDhw76YjM5dUpau958M7V+yLFj8v+PY0OIyOYpMy1fvly5u7urL7/8Uh0/flwNGzZMFS5cWF28eDHL/Q8dOqS+//57dezYMRUWFqa++eYb5e3trb744otcv2Z0dLQCoKKjo80NVyujUanhw5WShn2lPvlEd0SpoqKUatxY4ipWTKm9e3VHZHtiYpSaNk0pPz/5dxo/XndEmSUlKTV1qlKenhJj0aJKffml/N8jItItt9dvs5ORhx9+WA0ePDjdY3Xq1FGjR4/O9TmeeeYZ9eKLL+Z6f3tNRpSSi8LYsakJybhxtnOhuHVLqZYtJa4iRZTavl13RLbh2jWl3n9fqeLFUz+3ChXkIm9Ljh9XqmnT1Bjbt1cqm78JiIi0yO3126xumoSEBISEhCAoKCjd40FBQdi9e3euzhEaGordu3ejdevW2e4THx+PmJiYdDd7ZTBIkalJk+T+hAnAO+/oq86alq+vVGpt0wa4c0e6G4KDdUel16hRsk7Lhx8CN28CtWoBixbJoF9bqcuRlAR88olM2963T2bxLFokY1dYxIyI7JFZycj169eRnJyMsmXLpnu8bNmyiIiIyPHYihUrwtPTE40bN8brr7+OV3L4Zp88eTJ8fX1Tbv7+/uaEaZPGjEmd6jt9uqwbYwuKFJHaGJ06ycyLLl1koT9nde2a1ON46CHgp59kgG///npLuKf1999S02T0aBnLYioz378/x4YQkf3K0wBWQ4ZvPaVUpscy2rFjBw4ePIgFCxZg1qxZ+OGHH7Ldd8yYMYiOjk65Xb58OS9h2pxhw4CFC6VlpFMn3dGkKlRICnV16ybrqHTrJhdiRxcSItVy//or9bGxY4FNm6Ri7XPP6Z+ma5KUBEyeLEnSgQPSqrV0qSSSFSvqjo6IKH/czNm5VKlScHV1zdQKEhkZmam1JKOq/y3M0aBBA/z7778YP348evXqleW+np6e8PT0NCc0u5FxhdTYWMDdXW46eXgAK1YAL70EfP+9rLFz7x7Qp4/euCxNKVlBd/JkmfIKAJ6ewHffyXb16nKzJUePyiq7ISFyv3NnmbJboYLeuIiILMWslhEPDw8EBgYiOMPAguDgYLRo0SLX51FKIT4+3pyXdkj37gFPPy1rl9y7pzsawM0N+PprmYpsNEpiMnQoEBamO7L8MxqBtWuBZs1ksbgtW6TVo08f4N13dUeXtcREGW8UGCiJSLFi8vn88gsTESJyMOaOjDVN7V20aJE6fvy4Gj58uCpcuLC6cOGCUkqp0aNHqz59+qTsP3fuXLVu3Tp1+vRpdfr0abV48WLl4+Ojxo4da/HRuPZm716lvLxkJkS7dkrdvas7IpGcrNSQIamzNFxclHruOaV279YdWd61bZv6fry8lHr9daXCwnRHlb0jR5R68MHUmJ96SqmrV3VHRURkntxev83qpgGAHj16ICoqChMnTkR4eDgCAgKwYcMGVK5cGQAQHh6OS5cupexvNBoxZswYhIWFwc3NDdWrV8eUKVMwaNAgS+VTdqtpUxnI+uSTMoulQwcZA+DjozcuFxdZ16RLF1lmfvNmYOVKuTVrJoW1unaVlpS82LcPmDhR1klxdwemTQMGD5bn/vpLSqp7e8tYFm/v9NudOkmhNgC4cUPGd2S1n5ubVEc1jflo107GWrz+uozduU+vYoFTSsq4b9woty1bZJxIiRKy4m6vXhygSkQOrICSo3xx1JYRk127lPL1lb+AmzSRgmS25OhRpfr3V8rDI/Uv9SpVlJo5UwqD5dbu3VILw3QO023BgtR9tmzJ/Hza27Rpqfvu25fzvitXpu57547UVbElMTFKrV2r1ODB8u+ZMf6uXZUKD9cdJRFR3uX2+m1QyhYqXuQsJiYGvr6+iI6Oho/uZgMrOXQICAqSkuMPPCAtJWXK6I4qvYgIWfl13rzU0ug+PlJyfMiQ7GtcxMbK2BjTgFFXV6BvX1nIrXBhoGRJGQ8BAJGRwI4dckxcnPxMu/3UU6ktI0ePyjnSPp/2mCefBH780Yr/IGZSCjhyJLX1Y9cuaf0w8fCQku4dOsgtIEBfrERElpDb6zeTERty7Jh0J8TFAdu2AY0a6Y4oa3FxwDffyKJsp07JY66uQPfuwMiRQOPGmY8JCpJZLC+9JANGnWUZ+6goSSw3bpQupYzleGrUSE0+HntMkjMiIkfBZMROnTkjYyGaNtUdyf0ZjVL189NPJdEwadBAkpNff01dyfbkScDLy/FXj01OlrEpptaP/fvTV9v19paKtx06AO3bSzJCROSomIw4iL17ZRBjrVq6I8nZoUNSSv3331MvvsWKydTUfv0c+y/+8HBp9di4UQb73ryZ/vkGDSTx6NABeOQRqWtCROQMmIw4gMOHpeney0vGW9jiGAKlpBtiwgTAtDyRi4vMkjGVkileHBg0SMaVmFpK7FlkpIz92LJFkpAjR9I/X6yYdLd16CDdU6yQSkTOismIA7h2TS5qR45I68jmzVIAy1YYjcATT6R20Xh6ymDWUaOkXPmyZcDMmcC5c/K8u7tUdh050nbHw6R1546s+3L0qIznMf2MjEy/n8Eg42RMYz8efjjv056JiBwJkxEHcfOmXOD275eZK7/9Jgul6aJU+noXQ4YAX30ldULefjtzy0dyslQMnTFDZsmYPP641PyoVUtqfpQsqW8dmMREGYibNuE4ejT7yrMGgwzAbdEC6NhREsZSpQo2ZiIie8BkxIHExEgBsh07ZOzFunUyCLIgKSVJxYcfytTeJk3k8WvXJOEoV+7+59i/X1pKfvpJjknLxUUu6GXLpr+VKZP5sdKl87aKrtEIXLyYOek4dUoSkqyUKydjPgICUn/Wq+fYY2CIiCyFyYiDiY0FnnlGumq8vGTqb0HMuFEK+PlnqZgaGiqPPfusVGPNq0uXgNmzZcBnRERqzRJzFC9+/6Tl7t30icexY9L1kpWiRdMnHKafbPEgIso7JiMOKD5eannExUnriJdX/s6XlCTnvHdPfvr5pXbBnDkDHDwIfPJJ6gDNwoWBN96QcvClS+fvtdNKTASuXwf+/Tf9LTIy82Omlpi8cncH6tbNnHhUqsRy60RElpbb6zeH2dkRT09pkUhMTE1E5syRFov4+PSJRXy87LdnT+rxAwYAa9akPp/xoh4Xl3reDz+UwmaAtBoMGQKMGGGdlgJ3d0mE/Pzuv6/RKHVY7pe0/PuvdOUEBKQmHA0aADVryusREZHtYDJiZ9zd019Mg4NlLEd2kpJSZ3bExmaugWHi6iqJjCkZqVgRqF9fuoZGjJDZPLbANLakVCmJj4iI7B+7aezcjz/K1FlPT0kkPD3Tb3fqlDpL5dIlSUgy7uPpyamoRERkeeymcRLdu+d+3+wWsiMiItLJRXcARERE5NyYjBAREZFWTEaIiIhIKyYjREREpBWTESIiItKKyQgRERFpxWSEiIiItGIyQkRERFoxGSEiIiKtmIwQERGRVkxGiIiISCsmI0RERKQVkxEiIiLSyi5W7VVKAZCliImIiMg+mK7bput4duwiGbl9+zYAwN/fX3MkREREZK7bt2/D19c32+cN6n7pig0wGo24evUqihYtCoPBoDscq4mJiYG/vz8uX74MHx8f3eFYlTO9V8C53i/fq+NypvfL92oZSincvn0b5cuXh4tL9iND7KJlxMXFBRUrVtQdRoHx8fFx+P/8Js70XgHner98r47Lmd4v32v+5dQiYsIBrERERKQVkxEiIiLSismIDfH09MS4cePg6empOxSrc6b3CjjX++V7dVzO9H75XguWXQxgJSIiIsfFlhEiIiLSiskIERERacVkhIiIiLRiMkJERERaMRkpIJMnT0aTJk1QtGhRlClTBl27dsWpU6dyPGbbtm0wGAyZbidPniygqPNm/PjxmWIuV65cjsds374dgYGB8PLyQrVq1bBgwYICijb/qlSpkuXn9Prrr2e5vz19rn/++SeefPJJlC9fHgaDAWvXrk33vFIK48ePR/ny5VGoUCE89thj+Pvvv+973lWrVqFevXrw9PREvXr1sGbNGiu9g9zL6b0mJiZi1KhRaNCgAQoXLozy5cujb9++uHr1ao7nXLp0aZaf9b1796z8bu7vfp9tv379MsXdrFmz+57X3j5bAFl+RgaDAdOmTcv2nLb62ebmWmOLv7dMRgrI9u3b8frrr2Pv3r0IDg5GUlISgoKCcPfu3fsee+rUKYSHh6fcatasWQAR50/9+vXTxXz06NFs9w0LC0OnTp3w6KOPIjQ0FO+++y6GDh2KVatWFWDEeXfgwIF07zU4OBgA8Pzzz+d4nD18rnfv3kXDhg0xd+7cLJ+fOnUqZsyYgblz5+LAgQMoV64c2rVrl7KeVFb27NmDHj16oE+fPjhy5Aj69OmD7t27Y9++fdZ6G7mS03uNjY3FoUOH8P777+PQoUNYvXo1Tp8+jaeeeuq+5/Xx8Un3OYeHh8PLy8sab8Es9/tsAaBDhw7p4t6wYUOO57THzxZAps9n8eLFMBgMePbZZ3M8ry1+trm51tjk760iLSIjIxUAtX379mz3+eOPPxQAdfPmzYILzALGjRunGjZsmOv933nnHVWnTp10jw0aNEg1a9bMwpEVjGHDhqnq1asro9GY5fP2+rkCUGvWrEm5bzQaVbly5dSUKVNSHrt3757y9fVVCxYsyPY83bt3Vx06dEj3WPv27VXPnj0tHnNeZXyvWdm/f78CoC5evJjtPkuWLFG+vr6WDc4Ksnq/L730knr66afNOo+jfLZPP/20atOmTY772Mtnm/FaY6u/t2wZ0SQ6OhoAUKJEifvu++CDD8LPzw9t27bFH3/8Ye3QLOLMmTMoX748qlatip49e+L8+fPZ7rtnzx4EBQWle6x9+/Y4ePAgEhMTrR2qRSUkJODbb79F//7977uooz1+rmmFhYUhIiIi3Wfn6emJ1q1bY/fu3dkel93nndMxtig6OhoGgwHFihXLcb87d+6gcuXKqFixIrp06YLQ0NCCCdACtm3bhjJlyqBWrVoYOHAgIiMjc9zfET7bf//9F7/++isGDBhw333t4bPNeK2x1d9bJiMaKKUwcuRIPPLIIwgICMh2Pz8/PyxcuBCrVq3C6tWrUbt2bbRt2xZ//vlnAUZrvqZNm+Lrr7/Gpk2b8OWXXyIiIgItWrRAVFRUlvtHRESgbNmy6R4rW7YskpKScP369YII2WLWrl2LW7duoV+/ftnuY6+fa0YREREAkOVnZ3ouu+PMPcbW3Lt3D6NHj0bv3r1zXFisTp06WLp0KdatW4cffvgBXl5eaNmyJc6cOVOA0eZNx44d8d1332Hr1q349NNPceDAAbRp0wbx8fHZHuMIn+2yZctQtGhRdOvWLcf97OGzzepaY6u/t3axaq+jeeONN/DXX39h586dOe5Xu3Zt1K5dO+V+8+bNcfnyZUyfPh2tWrWydph51rFjx5TtBg0aoHnz5qhevTqWLVuGkSNHZnlMxlYE9V9h4Pu1LtiaRYsWoWPHjihfvny2+9jr55qdrD67+31ueTnGViQmJqJnz54wGo2YN29ejvs2a9Ys3aDPli1b4qGHHsKcOXMwe/Zsa4eaLz169EjZDggIQOPGjVG5cmX8+uuvOV6o7fmzBYDFixfjhRdeuO/YD3v4bHO61tja7y1bRgrYkCFDsG7dOvzxxx+oWLGi2cc3a9bMpjLv3ChcuDAaNGiQbdzlypXLlF1HRkbCzc0NJUuWLIgQLeLixYvYsmULXnnlFbOPtcfP1TRDKqvPLuNfUBmPM/cYW5GYmIju3bsjLCwMwcHBZi+37uLigiZNmtjdZw1Ii17lypVzjN2eP1sA2LFjB06dOpWn32Fb+2yzu9bY6u8tk5ECopTCG2+8gdWrV2Pr1q2oWrVqns4TGhoKPz8/C0dnXfHx8Thx4kS2cTdv3jxlBorJ5s2b0bhxY7i7uxdEiBaxZMkSlClTBp07dzb7WHv8XKtWrYpy5cql++wSEhKwfft2tGjRItvjsvu8czrGFpgSkTNnzmDLli15SpSVUjh8+LDdfdYAEBUVhcuXL+cYu71+tiaLFi1CYGAgGjZsaPaxtvLZ3u9aY7O/txYZBkv39eqrrypfX1+1bds2FR4ennKLjY1N2Wf06NGqT58+Kfdnzpyp1qxZo06fPq2OHTumRo8erQCoVatW6XgLufbmm2+qbdu2qfPnz6u9e/eqLl26qKJFi6oLFy4opTK/z/Pnzytvb281YsQIdfz4cbVo0SLl7u6uVq5cqestmC05OVlVqlRJjRo1KtNz9vy53r59W4WGhqrQ0FAFQM2YMUOFhoamzCCZMmWK8vX1VatXr1ZHjx5VvXr1Un5+fiomJiblHH369FGjR49Oub9r1y7l6uqqpkyZok6cOKGmTJmi3Nzc1N69ewv8/aWV03tNTExUTz31lKpYsaI6fPhwut/h+Pj4lHNkfK/jx49XGzduVOfOnVOhoaHq5ZdfVm5ubmrfvn063mI6Ob3f27dvqzfffFPt3r1bhYWFqT/++EM1b95cVahQweE+W5Po6Gjl7e2t5s+fn+U57OWzzc21xhZ/b5mMFBAAWd6WLFmSss9LL72kWrdunXL/k08+UdWrV1deXl6qePHi6pFHHlG//vprwQdvph49eig/Pz/l7u6uypcvr7p166b+/vvvlOczvk+llNq2bZt68MEHlYeHh6pSpUq2Xwi2atOmTQqAOnXqVKbn7PlzNU1Dznh76aWXlFIyTXDcuHGqXLlyytPTU7Vq1UodPXo03Tlat26dsr/JTz/9pGrXrq3c3d1VnTp1bCIRy+m9hoWFZfs7/Mcff6ScI+N7HT58uKpUqZLy8PBQpUuXVkFBQWr37t0F/+aykNP7jY2NVUFBQap06dLK3d1dVapUSb300kvq0qVL6c7hCJ+tyRdffKEKFSqkbt26leU57OWzzc21xhZ/bw3/BU9ERESkBceMEBERkVZMRoiIiEgrJiNERESkFZMRIiIi0orJCBEREWnFZISIiIi0YjJCREREWjEZISIiIq2YjBAREZFWTEaIiIhIKyYjREREpBWTESIiItLq/5aJpvMDmMYwAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "original_model = history_original.history[\"val_loss\"]\n",
    "drop_model = history_dropout.history[\"val_loss\"]\n",
    "epochs = range(1,21)\n",
    "plt.plot(epochs, original_model, \"b--\", label=\"Original Model\")\n",
    "plt.plot(epochs, drop_model, \"b-\", label=\"L2 Model\")\n",
    "plt.legend()"
   ]
  }
 ],
 "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": "3.10.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}