Day_079_HW.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Work\n",
    "1. 請比較 SGD optimizer 不同的 momentum 及使用 nesterov 與否的表現"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import keras\n",
    "\n",
    "# 本作業可以不需使用 GPU, 將 GPU 設定為 \"無\" (若想使用可自行開啟)\n",
    "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "train, test = keras.datasets.cifar10.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "## 資料前處理\n",
    "def preproc_x(x, flatten=True):\n",
    "    x = x / 255.\n",
    "    if flatten:\n",
    "        x = x.reshape((len(x), -1))\n",
    "    return x\n",
    "\n",
    "def preproc_y(y, num_classes=10):\n",
    "    if y.shape[-1] == 1:\n",
    "        y = keras.utils.to_categorical(y, num_classes)\n",
    "    return y    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train, y_train = train\n",
    "x_test, y_test = test\n",
    "\n",
    "# 資料前處理 - X 標準化\n",
    "x_train = preproc_x(x_train)\n",
    "x_test = preproc_x(x_test)\n",
    "\n",
    "# 資料前處理 -Y 轉成 onehot\n",
    "y_train = preproc_y(y_train)\n",
    "y_test = preproc_y(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_mlp(input_shape, output_units=10, num_neurons=[512, 256, 128]):\n",
    "    \"\"\"Code Here\n",
    "    建立你的神經網路\n",
    "    \"\"\"\n",
    "    input_layer = keras.layers.Input(input_shape)\n",
    "    \n",
    "    for i, n_units in enumerate(num_neurons):\n",
    "        if i == 0:\n",
    "            x = keras.layers.Dense(units=n_units, activation=\"relu\", name=\"hidden_layer\"+str(i+1))(input_layer)\n",
    "        else:\n",
    "            x = keras.layers.Dense(units=n_units, activation=\"relu\", name=\"hidden_layer\"+str(i+1))(x)\n",
    "    \n",
    "    out = keras.layers.Dense(units=output_units, activation=\"softmax\", name=\"output\")(x)\n",
    "    \n",
    "    model = keras.models.Model(inputs=[input_layer], outputs=[out])\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"Code Here\n",
    "設定超參數\n",
    "\"\"\"\n",
    "LEARNING_RATE = 1e-4\n",
    "EPOCHS = 50\n",
    "BATCH_SIZE = 512\n",
    "MOMENTUM = [0.2, 0.5, 0.75, 0.95]\n",
    "NESTEROV = [True, False]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'MTM-0.95-False'"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"MTM-%s-%s\" % (str(momentum_val), str(nesterov_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Experiment with MOMENTUM = 0.200000\n",
      "NESTEROV =  True\n",
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 3072)]            0         \n",
      "_________________________________________________________________\n",
      "hidden_layer1 (Dense)        (None, 512)               1573376   \n",
      "_________________________________________________________________\n",
      "hidden_layer2 (Dense)        (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "hidden_layer3 (Dense)        (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 1,738,890\n",
      "Trainable params: 1,738,890\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.3476 - accuracy: 0.1072 - val_loss: 2.3268 - val_accuracy: 0.1106\n",
      "Epoch 2/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.3157 - accuracy: 0.1225 - val_loss: 2.3038 - val_accuracy: 0.1346\n",
      "Epoch 3/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2978 - accuracy: 0.1392 - val_loss: 2.2890 - val_accuracy: 0.1517\n",
      "Epoch 4/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2853 - accuracy: 0.1512 - val_loss: 2.2780 - val_accuracy: 0.1606\n",
      "Epoch 5/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2755 - accuracy: 0.1576 - val_loss: 2.2691 - val_accuracy: 0.1644\n",
      "Epoch 6/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2673 - accuracy: 0.1607 - val_loss: 2.2614 - val_accuracy: 0.1709\n",
      "Epoch 7/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2602 - accuracy: 0.1638 - val_loss: 2.2545 - val_accuracy: 0.1741\n",
      "Epoch 8/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2537 - accuracy: 0.1657 - val_loss: 2.2482 - val_accuracy: 0.1755\n",
      "Epoch 9/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2477 - accuracy: 0.1678 - val_loss: 2.2424 - val_accuracy: 0.1779\n",
      "Epoch 10/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2421 - accuracy: 0.1706 - val_loss: 2.2369 - val_accuracy: 0.1810\n",
      "Epoch 11/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2368 - accuracy: 0.1744 - val_loss: 2.2316 - val_accuracy: 0.1833\n",
      "Epoch 12/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2317 - accuracy: 0.1775 - val_loss: 2.2266 - val_accuracy: 0.1861\n",
      "Epoch 13/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2268 - accuracy: 0.1808 - val_loss: 2.2219 - val_accuracy: 0.1894\n",
      "Epoch 14/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2221 - accuracy: 0.1833 - val_loss: 2.2173 - val_accuracy: 0.1933\n",
      "Epoch 15/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2176 - accuracy: 0.1866 - val_loss: 2.2128 - val_accuracy: 0.1953\n",
      "Epoch 16/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2132 - accuracy: 0.1904 - val_loss: 2.2085 - val_accuracy: 0.2008\n",
      "Epoch 17/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2089 - accuracy: 0.1926 - val_loss: 2.2044 - val_accuracy: 0.2028\n",
      "Epoch 18/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.2048 - accuracy: 0.1966 - val_loss: 2.2003 - val_accuracy: 0.2051\n",
      "Epoch 19/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2007 - accuracy: 0.1982 - val_loss: 2.1963 - val_accuracy: 0.2063\n",
      "Epoch 20/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1968 - accuracy: 0.2009 - val_loss: 2.1925 - val_accuracy: 0.2082\n",
      "Epoch 21/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1929 - accuracy: 0.2037 - val_loss: 2.1887 - val_accuracy: 0.2108\n",
      "Epoch 22/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1891 - accuracy: 0.2065 - val_loss: 2.1850 - val_accuracy: 0.2127\n",
      "Epoch 23/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1854 - accuracy: 0.2087 - val_loss: 2.1814 - val_accuracy: 0.2143\n",
      "Epoch 24/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1818 - accuracy: 0.2109 - val_loss: 2.1778 - val_accuracy: 0.2176\n",
      "Epoch 25/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1782 - accuracy: 0.2133 - val_loss: 2.1743 - val_accuracy: 0.2197\n",
      "Epoch 26/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1746 - accuracy: 0.2159 - val_loss: 2.1708 - val_accuracy: 0.2215\n",
      "Epoch 27/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1711 - accuracy: 0.2187 - val_loss: 2.1674 - val_accuracy: 0.2238\n",
      "Epoch 28/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1677 - accuracy: 0.2207 - val_loss: 2.1640 - val_accuracy: 0.2252\n",
      "Epoch 29/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1643 - accuracy: 0.2226 - val_loss: 2.1606 - val_accuracy: 0.2260\n",
      "Epoch 30/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1609 - accuracy: 0.2248 - val_loss: 2.1573 - val_accuracy: 0.2281\n",
      "Epoch 31/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1575 - accuracy: 0.2271 - val_loss: 2.1540 - val_accuracy: 0.2312\n",
      "Epoch 32/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1542 - accuracy: 0.2288 - val_loss: 2.1507 - val_accuracy: 0.2326\n",
      "Epoch 33/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1508 - accuracy: 0.2306 - val_loss: 2.1475 - val_accuracy: 0.2342\n",
      "Epoch 34/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1476 - accuracy: 0.2324 - val_loss: 2.1443 - val_accuracy: 0.2362\n",
      "Epoch 35/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1443 - accuracy: 0.2345 - val_loss: 2.1410 - val_accuracy: 0.2387\n",
      "Epoch 36/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1411 - accuracy: 0.2362 - val_loss: 2.1379 - val_accuracy: 0.2408\n",
      "Epoch 37/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1378 - accuracy: 0.2383 - val_loss: 2.1347 - val_accuracy: 0.2428\n",
      "Epoch 38/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1346 - accuracy: 0.2396 - val_loss: 2.1316 - val_accuracy: 0.2453\n",
      "Epoch 39/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1315 - accuracy: 0.2422 - val_loss: 2.1284 - val_accuracy: 0.2463\n",
      "Epoch 40/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1283 - accuracy: 0.2431 - val_loss: 2.1254 - val_accuracy: 0.2476\n",
      "Epoch 41/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1253 - accuracy: 0.2450 - val_loss: 2.1224 - val_accuracy: 0.2503\n",
      "Epoch 42/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1222 - accuracy: 0.2469 - val_loss: 2.1194 - val_accuracy: 0.2521\n",
      "Epoch 43/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1192 - accuracy: 0.2489 - val_loss: 2.1164 - val_accuracy: 0.2539\n",
      "Epoch 44/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1162 - accuracy: 0.2499 - val_loss: 2.1135 - val_accuracy: 0.2552\n",
      "Epoch 45/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1133 - accuracy: 0.2521 - val_loss: 2.1106 - val_accuracy: 0.2570\n",
      "Epoch 46/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1104 - accuracy: 0.2536 - val_loss: 2.1077 - val_accuracy: 0.2578\n",
      "Epoch 47/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1075 - accuracy: 0.2548 - val_loss: 2.1049 - val_accuracy: 0.2589\n",
      "Epoch 48/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1046 - accuracy: 0.2563 - val_loss: 2.1020 - val_accuracy: 0.2603\n",
      "Epoch 49/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1018 - accuracy: 0.2575 - val_loss: 2.0992 - val_accuracy: 0.2614\n",
      "Epoch 50/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0990 - accuracy: 0.2589 - val_loss: 2.0965 - val_accuracy: 0.2629\n",
      "Experiment with MOMENTUM = 0.200000\n",
      "NESTEROV =  False\n",
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 3072)]            0         \n",
      "_________________________________________________________________\n",
      "hidden_layer1 (Dense)        (None, 512)               1573376   \n",
      "_________________________________________________________________\n",
      "hidden_layer2 (Dense)        (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "hidden_layer3 (Dense)        (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 1,738,890\n",
      "Trainable params: 1,738,890\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.3633 - accuracy: 0.1175 - val_loss: 2.3437 - val_accuracy: 0.1182\n",
      "Epoch 2/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.3335 - accuracy: 0.1229 - val_loss: 2.3212 - val_accuracy: 0.1253\n",
      "Epoch 3/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.3155 - accuracy: 0.1264 - val_loss: 2.3070 - val_accuracy: 0.1306\n",
      "Epoch 4/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.3034 - accuracy: 0.1324 - val_loss: 2.2968 - val_accuracy: 0.1366\n",
      "Epoch 5/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2942 - accuracy: 0.1375 - val_loss: 2.2887 - val_accuracy: 0.1423\n",
      "Epoch 6/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2865 - accuracy: 0.1411 - val_loss: 2.2818 - val_accuracy: 0.1445\n",
      "Epoch 7/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2798 - accuracy: 0.1464 - val_loss: 2.2755 - val_accuracy: 0.1519\n",
      "Epoch 8/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2738 - accuracy: 0.1517 - val_loss: 2.2698 - val_accuracy: 0.1547\n",
      "Epoch 9/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2682 - accuracy: 0.1565 - val_loss: 2.2645 - val_accuracy: 0.1597\n",
      "Epoch 10/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2628 - accuracy: 0.1613 - val_loss: 2.2594 - val_accuracy: 0.1652\n",
      "Epoch 11/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2578 - accuracy: 0.1662 - val_loss: 2.2544 - val_accuracy: 0.1703\n",
      "Epoch 12/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.2529 - accuracy: 0.1713 - val_loss: 2.2497 - val_accuracy: 0.1739\n",
      "Epoch 13/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.2482 - accuracy: 0.1764 - val_loss: 2.2451 - val_accuracy: 0.1777\n",
      "Epoch 14/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2437 - accuracy: 0.1808 - val_loss: 2.2407 - val_accuracy: 0.1824\n",
      "Epoch 15/50\n",
      "98/98 [==============================] - 2s 19ms/step - loss: 2.2392 - accuracy: 0.1860 - val_loss: 2.2364 - val_accuracy: 0.1869\n",
      "Epoch 16/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.2349 - accuracy: 0.1902 - val_loss: 2.2322 - val_accuracy: 0.1918\n",
      "Epoch 17/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2307 - accuracy: 0.1945 - val_loss: 2.2281 - val_accuracy: 0.1964\n",
      "Epoch 18/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2266 - accuracy: 0.1988 - val_loss: 2.2241 - val_accuracy: 0.2015\n",
      "Epoch 19/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2225 - accuracy: 0.2026 - val_loss: 2.2202 - val_accuracy: 0.2048\n",
      "Epoch 20/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2186 - accuracy: 0.2067 - val_loss: 2.2163 - val_accuracy: 0.2075\n",
      "Epoch 21/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2147 - accuracy: 0.2101 - val_loss: 2.2125 - val_accuracy: 0.2107\n",
      "Epoch 22/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2109 - accuracy: 0.2129 - val_loss: 2.2088 - val_accuracy: 0.2153\n",
      "Epoch 23/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2071 - accuracy: 0.2155 - val_loss: 2.2051 - val_accuracy: 0.2194\n",
      "Epoch 24/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2034 - accuracy: 0.2186 - val_loss: 2.2015 - val_accuracy: 0.2218\n",
      "Epoch 25/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1998 - accuracy: 0.2219 - val_loss: 2.1980 - val_accuracy: 0.2252\n",
      "Epoch 26/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1963 - accuracy: 0.2245 - val_loss: 2.1945 - val_accuracy: 0.2273\n",
      "Epoch 27/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1928 - accuracy: 0.2279 - val_loss: 2.1911 - val_accuracy: 0.2303\n",
      "Epoch 28/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1893 - accuracy: 0.2291 - val_loss: 2.1877 - val_accuracy: 0.2332\n",
      "Epoch 29/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1860 - accuracy: 0.2323 - val_loss: 2.1844 - val_accuracy: 0.2364\n",
      "Epoch 30/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1826 - accuracy: 0.2347 - val_loss: 2.1811 - val_accuracy: 0.2394\n",
      "Epoch 31/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1794 - accuracy: 0.2370 - val_loss: 2.1779 - val_accuracy: 0.2420\n",
      "Epoch 32/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1762 - accuracy: 0.2385 - val_loss: 2.1747 - val_accuracy: 0.2441\n",
      "Epoch 33/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1730 - accuracy: 0.2405 - val_loss: 2.1716 - val_accuracy: 0.2439\n",
      "Epoch 34/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1699 - accuracy: 0.2424 - val_loss: 2.1685 - val_accuracy: 0.2452\n",
      "Epoch 35/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1668 - accuracy: 0.2440 - val_loss: 2.1654 - val_accuracy: 0.2474\n",
      "Epoch 36/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1637 - accuracy: 0.2456 - val_loss: 2.1623 - val_accuracy: 0.2489\n",
      "Epoch 37/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1607 - accuracy: 0.2477 - val_loss: 2.1593 - val_accuracy: 0.2492\n",
      "Epoch 38/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1577 - accuracy: 0.2486 - val_loss: 2.1563 - val_accuracy: 0.2527\n",
      "Epoch 39/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1546 - accuracy: 0.2503 - val_loss: 2.1533 - val_accuracy: 0.2539\n",
      "Epoch 40/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1516 - accuracy: 0.2514 - val_loss: 2.1503 - val_accuracy: 0.2572\n",
      "Epoch 41/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1486 - accuracy: 0.2537 - val_loss: 2.1473 - val_accuracy: 0.2578\n",
      "Epoch 42/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1457 - accuracy: 0.2548 - val_loss: 2.1443 - val_accuracy: 0.2596\n",
      "Epoch 43/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1427 - accuracy: 0.2560 - val_loss: 2.1414 - val_accuracy: 0.2602\n",
      "Epoch 44/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1398 - accuracy: 0.2568 - val_loss: 2.1385 - val_accuracy: 0.2611\n",
      "Epoch 45/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1369 - accuracy: 0.2581 - val_loss: 2.1356 - val_accuracy: 0.2625\n",
      "Epoch 46/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1340 - accuracy: 0.2599 - val_loss: 2.1327 - val_accuracy: 0.2642\n",
      "Epoch 47/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1312 - accuracy: 0.2608 - val_loss: 2.1299 - val_accuracy: 0.2640\n",
      "Epoch 48/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1283 - accuracy: 0.2613 - val_loss: 2.1271 - val_accuracy: 0.2651\n",
      "Epoch 49/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1256 - accuracy: 0.2620 - val_loss: 2.1243 - val_accuracy: 0.2665\n",
      "Epoch 50/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1228 - accuracy: 0.2630 - val_loss: 2.1216 - val_accuracy: 0.2686\n",
      "Experiment with MOMENTUM = 0.500000\n",
      "NESTEROV =  True\n",
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 3072)]            0         \n",
      "_________________________________________________________________\n",
      "hidden_layer1 (Dense)        (None, 512)               1573376   \n",
      "_________________________________________________________________\n",
      "hidden_layer2 (Dense)        (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "hidden_layer3 (Dense)        (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 1,738,890\n",
      "Trainable params: 1,738,890\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "98/98 [==============================] - 2s 16ms/step - loss: 2.4135 - accuracy: 0.0986 - val_loss: 2.3564 - val_accuracy: 0.1034\n",
      "Epoch 2/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.3247 - accuracy: 0.1075 - val_loss: 2.3036 - val_accuracy: 0.1131\n",
      "Epoch 3/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2883 - accuracy: 0.1205 - val_loss: 2.2768 - val_accuracy: 0.1238\n",
      "Epoch 4/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2677 - accuracy: 0.1325 - val_loss: 2.2597 - val_accuracy: 0.1371\n",
      "Epoch 5/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2532 - accuracy: 0.1479 - val_loss: 2.2467 - val_accuracy: 0.1517\n",
      "Epoch 6/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2416 - accuracy: 0.1599 - val_loss: 2.2359 - val_accuracy: 0.1634\n",
      "Epoch 7/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2316 - accuracy: 0.1697 - val_loss: 2.2264 - val_accuracy: 0.1731\n",
      "Epoch 8/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2227 - accuracy: 0.1765 - val_loss: 2.2179 - val_accuracy: 0.1800\n",
      "Epoch 9/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2145 - accuracy: 0.1827 - val_loss: 2.2100 - val_accuracy: 0.1843\n",
      "Epoch 10/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2069 - accuracy: 0.1871 - val_loss: 2.2026 - val_accuracy: 0.1900\n",
      "Epoch 11/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1998 - accuracy: 0.1919 - val_loss: 2.1957 - val_accuracy: 0.1948\n",
      "Epoch 12/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1932 - accuracy: 0.1951 - val_loss: 2.1893 - val_accuracy: 0.2000\n",
      "Epoch 13/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1869 - accuracy: 0.1991 - val_loss: 2.1831 - val_accuracy: 0.2025\n",
      "Epoch 14/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1809 - accuracy: 0.2005 - val_loss: 2.1773 - val_accuracy: 0.2062\n",
      "Epoch 15/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1752 - accuracy: 0.2045 - val_loss: 2.1717 - val_accuracy: 0.2079\n",
      "Epoch 16/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1697 - accuracy: 0.2058 - val_loss: 2.1664 - val_accuracy: 0.2095\n",
      "Epoch 17/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1645 - accuracy: 0.2085 - val_loss: 2.1612 - val_accuracy: 0.2115\n",
      "Epoch 18/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1594 - accuracy: 0.2097 - val_loss: 2.1563 - val_accuracy: 0.2138\n",
      "Epoch 19/50\n",
      "98/98 [==============================] - 2s 19ms/step - loss: 2.1545 - accuracy: 0.2122 - val_loss: 2.1515 - val_accuracy: 0.2176\n",
      "Epoch 20/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.1497 - accuracy: 0.2152 - val_loss: 2.1468 - val_accuracy: 0.2206\n",
      "Epoch 21/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1451 - accuracy: 0.2168 - val_loss: 2.1422 - val_accuracy: 0.2226\n",
      "Epoch 22/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1405 - accuracy: 0.2186 - val_loss: 2.1378 - val_accuracy: 0.2272\n",
      "Epoch 23/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1361 - accuracy: 0.2215 - val_loss: 2.1334 - val_accuracy: 0.2290\n",
      "Epoch 24/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1318 - accuracy: 0.2233 - val_loss: 2.1292 - val_accuracy: 0.2306\n",
      "Epoch 25/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1276 - accuracy: 0.2257 - val_loss: 2.1250 - val_accuracy: 0.2320\n",
      "Epoch 26/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1235 - accuracy: 0.2260 - val_loss: 2.1210 - val_accuracy: 0.2334\n",
      "Epoch 27/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.1195 - accuracy: 0.2280 - val_loss: 2.1170 - val_accuracy: 0.2360\n",
      "Epoch 28/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1155 - accuracy: 0.2290 - val_loss: 2.1131 - val_accuracy: 0.2382\n",
      "Epoch 29/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1116 - accuracy: 0.2310 - val_loss: 2.1093 - val_accuracy: 0.2401\n",
      "Epoch 30/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1078 - accuracy: 0.2338 - val_loss: 2.1055 - val_accuracy: 0.2415\n",
      "Epoch 31/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1041 - accuracy: 0.2342 - val_loss: 2.1018 - val_accuracy: 0.2431\n",
      "Epoch 32/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1005 - accuracy: 0.2364 - val_loss: 2.0982 - val_accuracy: 0.2442\n",
      "Epoch 33/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0969 - accuracy: 0.2376 - val_loss: 2.0947 - val_accuracy: 0.2448\n",
      "Epoch 34/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0933 - accuracy: 0.2397 - val_loss: 2.0912 - val_accuracy: 0.2476\n",
      "Epoch 35/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0898 - accuracy: 0.2421 - val_loss: 2.0877 - val_accuracy: 0.2488\n",
      "Epoch 36/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0864 - accuracy: 0.2443 - val_loss: 2.0843 - val_accuracy: 0.2504\n",
      "Epoch 37/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0830 - accuracy: 0.2453 - val_loss: 2.0810 - val_accuracy: 0.2521\n",
      "Epoch 38/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0797 - accuracy: 0.2475 - val_loss: 2.0777 - val_accuracy: 0.2539\n",
      "Epoch 39/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0764 - accuracy: 0.2501 - val_loss: 2.0744 - val_accuracy: 0.2558\n",
      "Epoch 40/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0732 - accuracy: 0.2517 - val_loss: 2.0712 - val_accuracy: 0.2564\n",
      "Epoch 41/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0700 - accuracy: 0.2523 - val_loss: 2.0680 - val_accuracy: 0.2577\n",
      "Epoch 42/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0669 - accuracy: 0.2547 - val_loss: 2.0649 - val_accuracy: 0.2605\n",
      "Epoch 43/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0638 - accuracy: 0.2567 - val_loss: 2.0618 - val_accuracy: 0.2603\n",
      "Epoch 44/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0608 - accuracy: 0.2580 - val_loss: 2.0588 - val_accuracy: 0.2624\n",
      "Epoch 45/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0579 - accuracy: 0.2598 - val_loss: 2.0559 - val_accuracy: 0.2632\n",
      "Epoch 46/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0549 - accuracy: 0.2616 - val_loss: 2.0530 - val_accuracy: 0.2645\n",
      "Epoch 47/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0521 - accuracy: 0.2629 - val_loss: 2.0501 - val_accuracy: 0.2651\n",
      "Epoch 48/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0492 - accuracy: 0.2644 - val_loss: 2.0473 - val_accuracy: 0.2660\n",
      "Epoch 49/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0464 - accuracy: 0.2654 - val_loss: 2.0446 - val_accuracy: 0.2670\n",
      "Epoch 50/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0437 - accuracy: 0.2673 - val_loss: 2.0418 - val_accuracy: 0.2689\n",
      "Experiment with MOMENTUM = 0.500000\n",
      "NESTEROV =  False\n",
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 3072)]            0         \n",
      "_________________________________________________________________\n",
      "hidden_layer1 (Dense)        (None, 512)               1573376   \n",
      "_________________________________________________________________\n",
      "hidden_layer2 (Dense)        (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "hidden_layer3 (Dense)        (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 1,738,890\n",
      "Trainable params: 1,738,890\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "98/98 [==============================] - 2s 16ms/step - loss: 2.3313 - accuracy: 0.0885 - val_loss: 2.3147 - val_accuracy: 0.1004\n",
      "Epoch 2/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.3039 - accuracy: 0.1133 - val_loss: 2.2938 - val_accuracy: 0.1273\n",
      "Epoch 3/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2870 - accuracy: 0.1306 - val_loss: 2.2797 - val_accuracy: 0.1396\n",
      "Epoch 4/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2747 - accuracy: 0.1418 - val_loss: 2.2684 - val_accuracy: 0.1467\n",
      "Epoch 5/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2643 - accuracy: 0.1527 - val_loss: 2.2585 - val_accuracy: 0.1595\n",
      "Epoch 6/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2551 - accuracy: 0.1612 - val_loss: 2.2496 - val_accuracy: 0.1702\n",
      "Epoch 7/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2467 - accuracy: 0.1704 - val_loss: 2.2414 - val_accuracy: 0.1768\n",
      "Epoch 8/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2390 - accuracy: 0.1770 - val_loss: 2.2339 - val_accuracy: 0.1832\n",
      "Epoch 9/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2318 - accuracy: 0.1840 - val_loss: 2.2270 - val_accuracy: 0.1894\n",
      "Epoch 10/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2251 - accuracy: 0.1887 - val_loss: 2.2204 - val_accuracy: 0.1945\n",
      "Epoch 11/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2187 - accuracy: 0.1951 - val_loss: 2.2141 - val_accuracy: 0.1995\n",
      "Epoch 12/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2125 - accuracy: 0.2001 - val_loss: 2.2080 - val_accuracy: 0.2038\n",
      "Epoch 13/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2064 - accuracy: 0.2041 - val_loss: 2.2021 - val_accuracy: 0.2088\n",
      "Epoch 14/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2005 - accuracy: 0.2079 - val_loss: 2.1963 - val_accuracy: 0.2137\n",
      "Epoch 15/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1946 - accuracy: 0.2118 - val_loss: 2.1905 - val_accuracy: 0.2172\n",
      "Epoch 16/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1888 - accuracy: 0.2157 - val_loss: 2.1848 - val_accuracy: 0.2210\n",
      "Epoch 17/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1830 - accuracy: 0.2192 - val_loss: 2.1791 - val_accuracy: 0.2232\n",
      "Epoch 18/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1773 - accuracy: 0.2227 - val_loss: 2.1735 - val_accuracy: 0.2258\n",
      "Epoch 19/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1718 - accuracy: 0.2258 - val_loss: 2.1680 - val_accuracy: 0.2296\n",
      "Epoch 20/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1664 - accuracy: 0.2291 - val_loss: 2.1627 - val_accuracy: 0.2309\n",
      "Epoch 21/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1611 - accuracy: 0.2324 - val_loss: 2.1575 - val_accuracy: 0.2323\n",
      "Epoch 22/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1561 - accuracy: 0.2352 - val_loss: 2.1526 - val_accuracy: 0.2361\n",
      "Epoch 23/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1512 - accuracy: 0.2383 - val_loss: 2.1478 - val_accuracy: 0.2391\n",
      "Epoch 24/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1464 - accuracy: 0.2400 - val_loss: 2.1431 - val_accuracy: 0.2405\n",
      "Epoch 25/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1418 - accuracy: 0.2424 - val_loss: 2.1384 - val_accuracy: 0.2438\n",
      "Epoch 26/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1372 - accuracy: 0.2448 - val_loss: 2.1339 - val_accuracy: 0.2474\n",
      "Epoch 27/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1327 - accuracy: 0.2468 - val_loss: 2.1294 - val_accuracy: 0.2499\n",
      "Epoch 28/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1281 - accuracy: 0.2489 - val_loss: 2.1249 - val_accuracy: 0.2525\n",
      "Epoch 29/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1237 - accuracy: 0.2515 - val_loss: 2.1204 - val_accuracy: 0.2547\n",
      "Epoch 30/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1193 - accuracy: 0.2541 - val_loss: 2.1160 - val_accuracy: 0.2563\n",
      "Epoch 31/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1149 - accuracy: 0.2561 - val_loss: 2.1117 - val_accuracy: 0.2586\n",
      "Epoch 32/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1106 - accuracy: 0.2571 - val_loss: 2.1075 - val_accuracy: 0.2604\n",
      "Epoch 33/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1064 - accuracy: 0.2591 - val_loss: 2.1033 - val_accuracy: 0.2628\n",
      "Epoch 34/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1023 - accuracy: 0.2608 - val_loss: 2.0992 - val_accuracy: 0.2646\n",
      "Epoch 35/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0983 - accuracy: 0.2623 - val_loss: 2.0953 - val_accuracy: 0.2660\n",
      "Epoch 36/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0944 - accuracy: 0.2642 - val_loss: 2.0913 - val_accuracy: 0.2681\n",
      "Epoch 37/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0905 - accuracy: 0.2657 - val_loss: 2.0875 - val_accuracy: 0.2707\n",
      "Epoch 38/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0867 - accuracy: 0.2670 - val_loss: 2.0838 - val_accuracy: 0.2726\n",
      "Epoch 39/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0830 - accuracy: 0.2685 - val_loss: 2.0801 - val_accuracy: 0.2745\n",
      "Epoch 40/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0794 - accuracy: 0.2699 - val_loss: 2.0765 - val_accuracy: 0.2752\n",
      "Epoch 41/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0758 - accuracy: 0.2717 - val_loss: 2.0729 - val_accuracy: 0.2766\n",
      "Epoch 42/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0723 - accuracy: 0.2739 - val_loss: 2.0694 - val_accuracy: 0.2781\n",
      "Epoch 43/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0689 - accuracy: 0.2740 - val_loss: 2.0660 - val_accuracy: 0.2792\n",
      "Epoch 44/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0655 - accuracy: 0.2760 - val_loss: 2.0627 - val_accuracy: 0.2809\n",
      "Epoch 45/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0622 - accuracy: 0.2778 - val_loss: 2.0594 - val_accuracy: 0.2826\n",
      "Epoch 46/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0589 - accuracy: 0.2790 - val_loss: 2.0561 - val_accuracy: 0.2840\n",
      "Epoch 47/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0557 - accuracy: 0.2798 - val_loss: 2.0529 - val_accuracy: 0.2858\n",
      "Epoch 48/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0525 - accuracy: 0.2808 - val_loss: 2.0497 - val_accuracy: 0.2873\n",
      "Epoch 49/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0494 - accuracy: 0.2828 - val_loss: 2.0466 - val_accuracy: 0.2877\n",
      "Epoch 50/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0463 - accuracy: 0.2830 - val_loss: 2.0436 - val_accuracy: 0.2880\n",
      "Experiment with MOMENTUM = 0.750000\n",
      "NESTEROV =  True\n",
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 3072)]            0         \n",
      "_________________________________________________________________\n",
      "hidden_layer1 (Dense)        (None, 512)               1573376   \n",
      "_________________________________________________________________\n",
      "hidden_layer2 (Dense)        (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "hidden_layer3 (Dense)        (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 1,738,890\n",
      "Trainable params: 1,738,890\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "98/98 [==============================] - 2s 16ms/step - loss: 2.3397 - accuracy: 0.1019 - val_loss: 2.2984 - val_accuracy: 0.1162\n",
      "Epoch 2/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2771 - accuracy: 0.1328 - val_loss: 2.2599 - val_accuracy: 0.1453\n",
      "Epoch 3/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2482 - accuracy: 0.1588 - val_loss: 2.2368 - val_accuracy: 0.1688\n",
      "Epoch 4/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2282 - accuracy: 0.1800 - val_loss: 2.2187 - val_accuracy: 0.1897\n",
      "Epoch 5/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2116 - accuracy: 0.1952 - val_loss: 2.2032 - val_accuracy: 0.2010\n",
      "Epoch 6/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1969 - accuracy: 0.2043 - val_loss: 2.1891 - val_accuracy: 0.2089\n",
      "Epoch 7/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1835 - accuracy: 0.2137 - val_loss: 2.1762 - val_accuracy: 0.2156\n",
      "Epoch 8/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1711 - accuracy: 0.2211 - val_loss: 2.1642 - val_accuracy: 0.2237\n",
      "Epoch 9/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1595 - accuracy: 0.2275 - val_loss: 2.1529 - val_accuracy: 0.2302\n",
      "Epoch 10/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1486 - accuracy: 0.2333 - val_loss: 2.1423 - val_accuracy: 0.2353\n",
      "Epoch 11/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1382 - accuracy: 0.2381 - val_loss: 2.1321 - val_accuracy: 0.2404\n",
      "Epoch 12/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.1283 - accuracy: 0.2432 - val_loss: 2.1223 - val_accuracy: 0.2466\n",
      "Epoch 13/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1188 - accuracy: 0.2476 - val_loss: 2.1128 - val_accuracy: 0.2498\n",
      "Epoch 14/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1096 - accuracy: 0.2510 - val_loss: 2.1038 - val_accuracy: 0.2556\n",
      "Epoch 15/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.1008 - accuracy: 0.2543 - val_loss: 2.0950 - val_accuracy: 0.2609\n",
      "Epoch 16/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.0922 - accuracy: 0.2579 - val_loss: 2.0866 - val_accuracy: 0.2642\n",
      "Epoch 17/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0839 - accuracy: 0.2617 - val_loss: 2.0783 - val_accuracy: 0.2655\n",
      "Epoch 18/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0758 - accuracy: 0.2647 - val_loss: 2.0704 - val_accuracy: 0.2688\n",
      "Epoch 19/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0680 - accuracy: 0.2681 - val_loss: 2.0627 - val_accuracy: 0.2705\n",
      "Epoch 20/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0604 - accuracy: 0.2705 - val_loss: 2.0553 - val_accuracy: 0.2744\n",
      "Epoch 21/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0531 - accuracy: 0.2743 - val_loss: 2.0481 - val_accuracy: 0.2767\n",
      "Epoch 22/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0460 - accuracy: 0.2762 - val_loss: 2.0411 - val_accuracy: 0.2814\n",
      "Epoch 23/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0391 - accuracy: 0.2800 - val_loss: 2.0343 - val_accuracy: 0.2807\n",
      "Epoch 24/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0324 - accuracy: 0.2828 - val_loss: 2.0276 - val_accuracy: 0.2833\n",
      "Epoch 25/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0258 - accuracy: 0.2851 - val_loss: 2.0212 - val_accuracy: 0.2854\n",
      "Epoch 26/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0196 - accuracy: 0.2865 - val_loss: 2.0151 - val_accuracy: 0.2886\n",
      "Epoch 27/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0135 - accuracy: 0.2900 - val_loss: 2.0090 - val_accuracy: 0.2903\n",
      "Epoch 28/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0077 - accuracy: 0.2924 - val_loss: 2.0033 - val_accuracy: 0.2929\n",
      "Epoch 29/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.0020 - accuracy: 0.2946 - val_loss: 1.9977 - val_accuracy: 0.2941\n",
      "Epoch 30/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9965 - accuracy: 0.2957 - val_loss: 1.9923 - val_accuracy: 0.2970\n",
      "Epoch 31/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9911 - accuracy: 0.2991 - val_loss: 1.9871 - val_accuracy: 0.2967\n",
      "Epoch 32/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9860 - accuracy: 0.2994 - val_loss: 1.9820 - val_accuracy: 0.2997\n",
      "Epoch 33/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9810 - accuracy: 0.3015 - val_loss: 1.9771 - val_accuracy: 0.3001\n",
      "Epoch 34/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.9761 - accuracy: 0.3026 - val_loss: 1.9724 - val_accuracy: 0.3024\n",
      "Epoch 35/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.9714 - accuracy: 0.3047 - val_loss: 1.9677 - val_accuracy: 0.3030\n",
      "Epoch 36/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.9668 - accuracy: 0.3064 - val_loss: 1.9632 - val_accuracy: 0.3065\n",
      "Epoch 37/50\n",
      "98/98 [==============================] - 2s 19ms/step - loss: 1.9623 - accuracy: 0.3086 - val_loss: 1.9589 - val_accuracy: 0.3071\n",
      "Epoch 38/50\n",
      "98/98 [==============================] - 2s 18ms/step - loss: 1.9580 - accuracy: 0.3097 - val_loss: 1.9546 - val_accuracy: 0.3084\n",
      "Epoch 39/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.9538 - accuracy: 0.3108 - val_loss: 1.9504 - val_accuracy: 0.3112\n",
      "Epoch 40/50\n",
      "98/98 [==============================] - 2s 18ms/step - loss: 1.9497 - accuracy: 0.3139 - val_loss: 1.9465 - val_accuracy: 0.3128\n",
      "Epoch 41/50\n",
      "98/98 [==============================] - 2s 19ms/step - loss: 1.9456 - accuracy: 0.3144 - val_loss: 1.9425 - val_accuracy: 0.3141\n",
      "Epoch 42/50\n",
      "98/98 [==============================] - 2s 19ms/step - loss: 1.9417 - accuracy: 0.3160 - val_loss: 1.9387 - val_accuracy: 0.3168\n",
      "Epoch 43/50\n",
      "98/98 [==============================] - 2s 18ms/step - loss: 1.9379 - accuracy: 0.3179 - val_loss: 1.9349 - val_accuracy: 0.3193\n",
      "Epoch 44/50\n",
      "98/98 [==============================] - 2s 18ms/step - loss: 1.9342 - accuracy: 0.3186 - val_loss: 1.9312 - val_accuracy: 0.3197\n",
      "Epoch 45/50\n",
      "98/98 [==============================] - 2s 18ms/step - loss: 1.9305 - accuracy: 0.3197 - val_loss: 1.9276 - val_accuracy: 0.3206\n",
      "Epoch 46/50\n",
      "98/98 [==============================] - 2s 19ms/step - loss: 1.9270 - accuracy: 0.3214 - val_loss: 1.9242 - val_accuracy: 0.3233\n",
      "Epoch 47/50\n",
      "98/98 [==============================] - 2s 18ms/step - loss: 1.9235 - accuracy: 0.3222 - val_loss: 1.9208 - val_accuracy: 0.3232\n",
      "Epoch 48/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.9202 - accuracy: 0.3244 - val_loss: 1.9174 - val_accuracy: 0.3251\n",
      "Epoch 49/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.9169 - accuracy: 0.3260 - val_loss: 1.9143 - val_accuracy: 0.3260\n",
      "Epoch 50/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.9137 - accuracy: 0.3272 - val_loss: 1.9111 - val_accuracy: 0.3285\n",
      "Experiment with MOMENTUM = 0.750000\n",
      "NESTEROV =  False\n",
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 3072)]            0         \n",
      "_________________________________________________________________\n",
      "hidden_layer1 (Dense)        (None, 512)               1573376   \n",
      "_________________________________________________________________\n",
      "hidden_layer2 (Dense)        (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "hidden_layer3 (Dense)        (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 1,738,890\n",
      "Trainable params: 1,738,890\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "98/98 [==============================] - 2s 17ms/step - loss: 2.3356 - accuracy: 0.0944 - val_loss: 2.3038 - val_accuracy: 0.1094\n",
      "Epoch 2/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2900 - accuracy: 0.1182 - val_loss: 2.2769 - val_accuracy: 0.1362\n",
      "Epoch 3/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.2677 - accuracy: 0.1482 - val_loss: 2.2577 - val_accuracy: 0.1618\n",
      "Epoch 4/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2502 - accuracy: 0.1724 - val_loss: 2.2417 - val_accuracy: 0.1839\n",
      "Epoch 5/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2350 - accuracy: 0.1855 - val_loss: 2.2272 - val_accuracy: 0.1979\n",
      "Epoch 6/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2212 - accuracy: 0.1980 - val_loss: 2.2138 - val_accuracy: 0.2089\n",
      "Epoch 7/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2082 - accuracy: 0.2074 - val_loss: 2.2013 - val_accuracy: 0.2200\n",
      "Epoch 8/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1961 - accuracy: 0.2160 - val_loss: 2.1897 - val_accuracy: 0.2259\n",
      "Epoch 9/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1847 - accuracy: 0.2219 - val_loss: 2.1787 - val_accuracy: 0.2332\n",
      "Epoch 10/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1740 - accuracy: 0.2287 - val_loss: 2.1682 - val_accuracy: 0.2385\n",
      "Epoch 11/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1637 - accuracy: 0.2369 - val_loss: 2.1582 - val_accuracy: 0.2446\n",
      "Epoch 12/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1539 - accuracy: 0.2417 - val_loss: 2.1485 - val_accuracy: 0.2502\n",
      "Epoch 13/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1443 - accuracy: 0.2466 - val_loss: 2.1392 - val_accuracy: 0.2554\n",
      "Epoch 14/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1350 - accuracy: 0.2514 - val_loss: 2.1300 - val_accuracy: 0.2607\n",
      "Epoch 15/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1259 - accuracy: 0.2566 - val_loss: 2.1210 - val_accuracy: 0.2675\n",
      "Epoch 16/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1169 - accuracy: 0.2605 - val_loss: 2.1122 - val_accuracy: 0.2711\n",
      "Epoch 17/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1082 - accuracy: 0.2646 - val_loss: 2.1035 - val_accuracy: 0.2744\n",
      "Epoch 18/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0996 - accuracy: 0.2694 - val_loss: 2.0951 - val_accuracy: 0.2776\n",
      "Epoch 19/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0914 - accuracy: 0.2724 - val_loss: 2.0870 - val_accuracy: 0.2793\n",
      "Epoch 20/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0834 - accuracy: 0.2756 - val_loss: 2.0790 - val_accuracy: 0.2827\n",
      "Epoch 21/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0757 - accuracy: 0.2790 - val_loss: 2.0714 - val_accuracy: 0.2852\n",
      "Epoch 22/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0684 - accuracy: 0.2810 - val_loss: 2.0641 - val_accuracy: 0.2876\n",
      "Epoch 23/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0612 - accuracy: 0.2833 - val_loss: 2.0571 - val_accuracy: 0.2878\n",
      "Epoch 24/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0542 - accuracy: 0.2870 - val_loss: 2.0502 - val_accuracy: 0.2907\n",
      "Epoch 25/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.0474 - accuracy: 0.2898 - val_loss: 2.0435 - val_accuracy: 0.2919\n",
      "Epoch 26/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0408 - accuracy: 0.2915 - val_loss: 2.0369 - val_accuracy: 0.2942\n",
      "Epoch 27/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0344 - accuracy: 0.2941 - val_loss: 2.0304 - val_accuracy: 0.2957\n",
      "Epoch 28/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0280 - accuracy: 0.2960 - val_loss: 2.0241 - val_accuracy: 0.2985\n",
      "Epoch 29/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0219 - accuracy: 0.2978 - val_loss: 2.0180 - val_accuracy: 0.3002\n",
      "Epoch 30/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0158 - accuracy: 0.2991 - val_loss: 2.0121 - val_accuracy: 0.3006\n",
      "Epoch 31/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0100 - accuracy: 0.3014 - val_loss: 2.0063 - val_accuracy: 0.3036\n",
      "Epoch 32/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0043 - accuracy: 0.3031 - val_loss: 2.0007 - val_accuracy: 0.3068\n",
      "Epoch 33/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9988 - accuracy: 0.3048 - val_loss: 1.9954 - val_accuracy: 0.3079\n",
      "Epoch 34/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.9935 - accuracy: 0.3062 - val_loss: 1.9901 - val_accuracy: 0.3087\n",
      "Epoch 35/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.9883 - accuracy: 0.3077 - val_loss: 1.9851 - val_accuracy: 0.3104\n",
      "Epoch 36/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9833 - accuracy: 0.3092 - val_loss: 1.9802 - val_accuracy: 0.3103\n",
      "Epoch 37/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.9784 - accuracy: 0.3107 - val_loss: 1.9754 - val_accuracy: 0.3122\n",
      "Epoch 38/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9736 - accuracy: 0.3120 - val_loss: 1.9707 - val_accuracy: 0.3160\n",
      "Epoch 39/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9690 - accuracy: 0.3141 - val_loss: 1.9664 - val_accuracy: 0.3181\n",
      "Epoch 40/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9645 - accuracy: 0.3142 - val_loss: 1.9618 - val_accuracy: 0.3183\n",
      "Epoch 41/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9600 - accuracy: 0.3165 - val_loss: 1.9575 - val_accuracy: 0.3186\n",
      "Epoch 42/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9557 - accuracy: 0.3181 - val_loss: 1.9533 - val_accuracy: 0.3211\n",
      "Epoch 43/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.9513 - accuracy: 0.3188 - val_loss: 1.9490 - val_accuracy: 0.3217\n",
      "Epoch 44/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.9470 - accuracy: 0.3207 - val_loss: 1.9448 - val_accuracy: 0.3216\n",
      "Epoch 45/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9428 - accuracy: 0.3234 - val_loss: 1.9407 - val_accuracy: 0.3255\n",
      "Epoch 46/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9387 - accuracy: 0.3237 - val_loss: 1.9366 - val_accuracy: 0.3264\n",
      "Epoch 47/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9346 - accuracy: 0.3257 - val_loss: 1.9328 - val_accuracy: 0.3258\n",
      "Epoch 48/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9307 - accuracy: 0.3269 - val_loss: 1.9289 - val_accuracy: 0.3272\n",
      "Epoch 49/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9269 - accuracy: 0.3280 - val_loss: 1.9254 - val_accuracy: 0.3282\n",
      "Epoch 50/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9233 - accuracy: 0.3286 - val_loss: 1.9220 - val_accuracy: 0.3304\n",
      "Experiment with MOMENTUM = 0.950000\n",
      "NESTEROV =  True\n",
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 3072)]            0         \n",
      "_________________________________________________________________\n",
      "hidden_layer1 (Dense)        (None, 512)               1573376   \n",
      "_________________________________________________________________\n",
      "hidden_layer2 (Dense)        (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "hidden_layer3 (Dense)        (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 1,738,890\n",
      "Trainable params: 1,738,890\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2830 - accuracy: 0.1455 - val_loss: 2.2229 - val_accuracy: 0.1955\n",
      "Epoch 2/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1875 - accuracy: 0.2172 - val_loss: 2.1537 - val_accuracy: 0.2361\n",
      "Epoch 3/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 2.1270 - accuracy: 0.2491 - val_loss: 2.1014 - val_accuracy: 0.2631\n",
      "Epoch 4/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0814 - accuracy: 0.2668 - val_loss: 2.0613 - val_accuracy: 0.2826\n",
      "Epoch 5/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0446 - accuracy: 0.2800 - val_loss: 2.0276 - val_accuracy: 0.2952\n",
      "Epoch 6/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0137 - accuracy: 0.2931 - val_loss: 1.9996 - val_accuracy: 0.3040\n",
      "Epoch 7/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9875 - accuracy: 0.3024 - val_loss: 1.9751 - val_accuracy: 0.3177\n",
      "Epoch 8/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9648 - accuracy: 0.3153 - val_loss: 1.9544 - val_accuracy: 0.3178\n",
      "Epoch 9/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.9452 - accuracy: 0.3210 - val_loss: 1.9361 - val_accuracy: 0.3289\n",
      "Epoch 10/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9278 - accuracy: 0.3285 - val_loss: 1.9205 - val_accuracy: 0.3363\n",
      "Epoch 11/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9128 - accuracy: 0.3334 - val_loss: 1.9062 - val_accuracy: 0.3408\n",
      "Epoch 12/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8991 - accuracy: 0.3360 - val_loss: 1.8935 - val_accuracy: 0.3433\n",
      "Epoch 13/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8866 - accuracy: 0.3416 - val_loss: 1.8816 - val_accuracy: 0.3469\n",
      "Epoch 14/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.8755 - accuracy: 0.3447 - val_loss: 1.8712 - val_accuracy: 0.3534\n",
      "Epoch 15/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8650 - accuracy: 0.3485 - val_loss: 1.8610 - val_accuracy: 0.3552\n",
      "Epoch 16/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8553 - accuracy: 0.3517 - val_loss: 1.8521 - val_accuracy: 0.3605\n",
      "Epoch 17/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8466 - accuracy: 0.3548 - val_loss: 1.8435 - val_accuracy: 0.3596\n",
      "Epoch 18/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8386 - accuracy: 0.3586 - val_loss: 1.8358 - val_accuracy: 0.3627\n",
      "Epoch 19/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.8306 - accuracy: 0.3611 - val_loss: 1.8281 - val_accuracy: 0.3682\n",
      "Epoch 20/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.8228 - accuracy: 0.3629 - val_loss: 1.8206 - val_accuracy: 0.3662\n",
      "Epoch 21/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8159 - accuracy: 0.3653 - val_loss: 1.8141 - val_accuracy: 0.3680\n",
      "Epoch 22/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8090 - accuracy: 0.3680 - val_loss: 1.8076 - val_accuracy: 0.3711\n",
      "Epoch 23/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.8023 - accuracy: 0.3711 - val_loss: 1.8006 - val_accuracy: 0.3747\n",
      "Epoch 24/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7961 - accuracy: 0.3726 - val_loss: 1.7953 - val_accuracy: 0.3784\n",
      "Epoch 25/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.7899 - accuracy: 0.3749 - val_loss: 1.7886 - val_accuracy: 0.3786\n",
      "Epoch 26/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7837 - accuracy: 0.3771 - val_loss: 1.7834 - val_accuracy: 0.3806\n",
      "Epoch 27/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7776 - accuracy: 0.3792 - val_loss: 1.7769 - val_accuracy: 0.3822\n",
      "Epoch 28/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7723 - accuracy: 0.3816 - val_loss: 1.7709 - val_accuracy: 0.3835\n",
      "Epoch 29/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7664 - accuracy: 0.3839 - val_loss: 1.7656 - val_accuracy: 0.3851\n",
      "Epoch 30/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7611 - accuracy: 0.3858 - val_loss: 1.7605 - val_accuracy: 0.3894\n",
      "Epoch 31/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7556 - accuracy: 0.3881 - val_loss: 1.7554 - val_accuracy: 0.3926\n",
      "Epoch 32/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7506 - accuracy: 0.3907 - val_loss: 1.7505 - val_accuracy: 0.3919\n",
      "Epoch 33/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7453 - accuracy: 0.3913 - val_loss: 1.7455 - val_accuracy: 0.3958\n",
      "Epoch 34/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7407 - accuracy: 0.3930 - val_loss: 1.7402 - val_accuracy: 0.3965\n",
      "Epoch 35/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.7359 - accuracy: 0.3950 - val_loss: 1.7371 - val_accuracy: 0.3982\n",
      "Epoch 36/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7313 - accuracy: 0.3968 - val_loss: 1.7325 - val_accuracy: 0.4004\n",
      "Epoch 37/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7266 - accuracy: 0.3985 - val_loss: 1.7271 - val_accuracy: 0.4025\n",
      "Epoch 38/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7223 - accuracy: 0.3992 - val_loss: 1.7232 - val_accuracy: 0.4048\n",
      "Epoch 39/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7180 - accuracy: 0.4012 - val_loss: 1.7181 - val_accuracy: 0.4031\n",
      "Epoch 40/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7134 - accuracy: 0.4024 - val_loss: 1.7144 - val_accuracy: 0.4050\n",
      "Epoch 41/50\n",
      "98/98 [==============================] - 2s 21ms/step - loss: 1.7094 - accuracy: 0.4033 - val_loss: 1.7107 - val_accuracy: 0.4071\n",
      "Epoch 42/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.7055 - accuracy: 0.4049 - val_loss: 1.7064 - val_accuracy: 0.4080\n",
      "Epoch 43/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.7013 - accuracy: 0.4064 - val_loss: 1.7036 - val_accuracy: 0.4087\n",
      "Epoch 44/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.6975 - accuracy: 0.4077 - val_loss: 1.6989 - val_accuracy: 0.4118\n",
      "Epoch 45/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.6935 - accuracy: 0.4096 - val_loss: 1.6956 - val_accuracy: 0.4133\n",
      "Epoch 46/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.6896 - accuracy: 0.4103 - val_loss: 1.6916 - val_accuracy: 0.4126\n",
      "Epoch 47/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.6857 - accuracy: 0.4128 - val_loss: 1.6876 - val_accuracy: 0.4143\n",
      "Epoch 48/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.6821 - accuracy: 0.4130 - val_loss: 1.6843 - val_accuracy: 0.4161\n",
      "Epoch 49/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.6786 - accuracy: 0.4146 - val_loss: 1.6809 - val_accuracy: 0.4144\n",
      "Epoch 50/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.6747 - accuracy: 0.4156 - val_loss: 1.6787 - val_accuracy: 0.4163\n",
      "Experiment with MOMENTUM = 0.950000\n",
      "NESTEROV =  False\n",
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 3072)]            0         \n",
      "_________________________________________________________________\n",
      "hidden_layer1 (Dense)        (None, 512)               1573376   \n",
      "_________________________________________________________________\n",
      "hidden_layer2 (Dense)        (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "hidden_layer3 (Dense)        (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 1,738,890\n",
      "Trainable params: 1,738,890\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "98/98 [==============================] - 2s 16ms/step - loss: 2.2940 - accuracy: 0.1357 - val_loss: 2.2368 - val_accuracy: 0.1957\n",
      "Epoch 2/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.2044 - accuracy: 0.2148 - val_loss: 2.1701 - val_accuracy: 0.2337\n",
      "Epoch 3/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 2.1457 - accuracy: 0.2390 - val_loss: 2.1210 - val_accuracy: 0.2495\n",
      "Epoch 4/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.1005 - accuracy: 0.2558 - val_loss: 2.0806 - val_accuracy: 0.2548\n",
      "Epoch 5/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 2.0625 - accuracy: 0.2678 - val_loss: 2.0455 - val_accuracy: 0.2711\n",
      "Epoch 6/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.0295 - accuracy: 0.2808 - val_loss: 2.0153 - val_accuracy: 0.2861\n",
      "Epoch 7/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 2.0022 - accuracy: 0.2930 - val_loss: 1.9903 - val_accuracy: 0.2977\n",
      "Epoch 8/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9788 - accuracy: 0.3031 - val_loss: 1.9685 - val_accuracy: 0.3049\n",
      "Epoch 9/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.9582 - accuracy: 0.3124 - val_loss: 1.9502 - val_accuracy: 0.3125\n",
      "Epoch 10/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.9402 - accuracy: 0.3195 - val_loss: 1.9337 - val_accuracy: 0.3155\n",
      "Epoch 11/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.9242 - accuracy: 0.3267 - val_loss: 1.9180 - val_accuracy: 0.3266\n",
      "Epoch 12/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.9092 - accuracy: 0.3322 - val_loss: 1.9039 - val_accuracy: 0.3349\n",
      "Epoch 13/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8957 - accuracy: 0.3384 - val_loss: 1.8911 - val_accuracy: 0.3390\n",
      "Epoch 14/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8834 - accuracy: 0.3424 - val_loss: 1.8798 - val_accuracy: 0.3438\n",
      "Epoch 15/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8727 - accuracy: 0.3461 - val_loss: 1.8703 - val_accuracy: 0.3489\n",
      "Epoch 16/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8628 - accuracy: 0.3505 - val_loss: 1.8590 - val_accuracy: 0.3519\n",
      "Epoch 17/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.8530 - accuracy: 0.3543 - val_loss: 1.8497 - val_accuracy: 0.3567\n",
      "Epoch 18/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.8440 - accuracy: 0.3584 - val_loss: 1.8422 - val_accuracy: 0.3583\n",
      "Epoch 19/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.8360 - accuracy: 0.3604 - val_loss: 1.8336 - val_accuracy: 0.3620\n",
      "Epoch 20/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.8282 - accuracy: 0.3649 - val_loss: 1.8273 - val_accuracy: 0.3617\n",
      "Epoch 21/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8205 - accuracy: 0.3673 - val_loss: 1.8200 - val_accuracy: 0.3657\n",
      "Epoch 22/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.8139 - accuracy: 0.3698 - val_loss: 1.8126 - val_accuracy: 0.3685\n",
      "Epoch 23/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.8072 - accuracy: 0.3719 - val_loss: 1.8064 - val_accuracy: 0.3722\n",
      "Epoch 24/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.8006 - accuracy: 0.3747 - val_loss: 1.8003 - val_accuracy: 0.3733\n",
      "Epoch 25/50\n",
      "98/98 [==============================] - 2s 18ms/step - loss: 1.7945 - accuracy: 0.3766 - val_loss: 1.7938 - val_accuracy: 0.3766\n",
      "Epoch 26/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7889 - accuracy: 0.3790 - val_loss: 1.7890 - val_accuracy: 0.3770\n",
      "Epoch 27/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7829 - accuracy: 0.3805 - val_loss: 1.7822 - val_accuracy: 0.3805\n",
      "Epoch 28/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7770 - accuracy: 0.3822 - val_loss: 1.7767 - val_accuracy: 0.3830\n",
      "Epoch 29/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7719 - accuracy: 0.3833 - val_loss: 1.7711 - val_accuracy: 0.3864\n",
      "Epoch 30/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7663 - accuracy: 0.3859 - val_loss: 1.7664 - val_accuracy: 0.3860\n",
      "Epoch 31/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7613 - accuracy: 0.3884 - val_loss: 1.7612 - val_accuracy: 0.3894\n",
      "Epoch 32/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7557 - accuracy: 0.3898 - val_loss: 1.7560 - val_accuracy: 0.3883\n",
      "Epoch 33/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7506 - accuracy: 0.3929 - val_loss: 1.7514 - val_accuracy: 0.3907\n",
      "Epoch 34/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7460 - accuracy: 0.3932 - val_loss: 1.7469 - val_accuracy: 0.3931\n",
      "Epoch 35/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7410 - accuracy: 0.3950 - val_loss: 1.7432 - val_accuracy: 0.3929\n",
      "Epoch 36/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7367 - accuracy: 0.3958 - val_loss: 1.7377 - val_accuracy: 0.3964\n",
      "Epoch 37/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7324 - accuracy: 0.3980 - val_loss: 1.7338 - val_accuracy: 0.3993\n",
      "Epoch 38/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.7276 - accuracy: 0.3991 - val_loss: 1.7291 - val_accuracy: 0.4015\n",
      "Epoch 39/50\n",
      "98/98 [==============================] - 2s 15ms/step - loss: 1.7232 - accuracy: 0.4020 - val_loss: 1.7248 - val_accuracy: 0.4003\n",
      "Epoch 40/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7186 - accuracy: 0.4042 - val_loss: 1.7213 - val_accuracy: 0.4000\n",
      "Epoch 41/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7146 - accuracy: 0.4041 - val_loss: 1.7174 - val_accuracy: 0.4033\n",
      "Epoch 42/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7106 - accuracy: 0.4058 - val_loss: 1.7132 - val_accuracy: 0.4040\n",
      "Epoch 43/50\n",
      "98/98 [==============================] - 2s 17ms/step - loss: 1.7063 - accuracy: 0.4078 - val_loss: 1.7089 - val_accuracy: 0.4054\n",
      "Epoch 44/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.7026 - accuracy: 0.4080 - val_loss: 1.7067 - val_accuracy: 0.4065\n",
      "Epoch 45/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.6989 - accuracy: 0.4104 - val_loss: 1.7030 - val_accuracy: 0.4071\n",
      "Epoch 46/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.6951 - accuracy: 0.4112 - val_loss: 1.6994 - val_accuracy: 0.4079\n",
      "Epoch 47/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.6910 - accuracy: 0.4127 - val_loss: 1.6961 - val_accuracy: 0.4111\n",
      "Epoch 48/50\n",
      "98/98 [==============================] - 1s 15ms/step - loss: 1.6873 - accuracy: 0.4149 - val_loss: 1.6918 - val_accuracy: 0.4119\n",
      "Epoch 49/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.6837 - accuracy: 0.4153 - val_loss: 1.6877 - val_accuracy: 0.4140\n",
      "Epoch 50/50\n",
      "98/98 [==============================] - 2s 16ms/step - loss: 1.6803 - accuracy: 0.4170 - val_loss: 1.6852 - val_accuracy: 0.4164\n"
     ]
    }
   ],
   "source": [
    "results = {}\n",
    "\"\"\"Code Here\n",
    "撰寫你的訓練流程並將結果用 dictionary 紀錄\n",
    "\"\"\"\n",
    "for momentum_val in MOMENTUM:\n",
    "    for nesterov_val in NESTEROV:\n",
    "        keras.backend.clear_session() # 把舊的 Graph 清掉\n",
    "        print(\"Experiment with MOMENTUM = %.6f\" % (momentum_val))\n",
    "        print(\"NESTEROV = \", nesterov_val)\n",
    "        model = build_mlp(input_shape = x_train.shape[1:])\n",
    "        model.summary()\n",
    "        optimizer = keras.optimizers.SGD(lr = LEARNING_RATE, nesterov = nesterov_val, momentum = momentum_val)\n",
    "        model.compile(loss=\"categorical_crossentropy\", metrics=[\"accuracy\"], optimizer=optimizer)\n",
    "\n",
    "        model.fit(x_train, y_train, \n",
    "                  epochs=EPOCHS, \n",
    "                  batch_size=BATCH_SIZE, \n",
    "                  validation_data=(x_test, y_test), \n",
    "                  shuffle=True)\n",
    "\n",
    "        # Collect results\n",
    "        train_loss = model.history.history[\"loss\"]\n",
    "        valid_loss = model.history.history[\"val_loss\"]\n",
    "        train_acc = model.history.history[\"accuracy\"]\n",
    "        valid_acc = model.history.history[\"val_accuracy\"]\n",
    "\n",
    "        exp_name_tag = \"MTM-%s-%s\" % (str(momentum_val), str(nesterov_val))\n",
    "        results[exp_name_tag] = {'train-loss': train_loss,\n",
    "                                 'valid-loss': valid_loss,\n",
    "                                 'train-acc': train_acc,\n",
    "                                 'valid-acc': valid_acc}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeMAAAF1CAYAAADbSIJmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOzdd3zV9b3H8df37JNzTs7JDhkQCEOQKSAIKCIoQysOrLjRWr1qq7bqdRTrLXK1drhq6yiO1mutddYJaJUhKgqK7CkBErL3SXL29/7xzS/nhL0kGL7PPM4jyTm//M7vJA94n893CiklmqZpmqZ1HFNHX4CmaZqmHe90GGuapmlaB9NhrGmapmkdTIexpmmapnUwHcaapmma1sF0GGuapmlaB9NhrGmapmkdTIexpv3ACSGKhBATOvo6NE07dDqMNU3TNK2D6TDWtE5KCPFTIcRmIUSNEOJtIURO6/1CCPGIEKJCCFEvhFgphOjf+tgUIcRaIUSjEKJECHF7x74KTTs+6DDWtE5ICHEG8CDwY6ALsA34Z+vDZwGnAb0BH3AxUN362LPA9VJKD9Af+PgoXramHbcsHX0BmqZ9Ly4DnpNSfg0ghLgbqBVCFABhwAOcAHwppVyX8HNhoJ8Q4lspZS1Qe1SvWtOOU7oy1rTOKQdVDQMgpfSjqt9cKeXHwBPAn4FyIcQzQojk1kMvBKYA24QQC4UQpxzl69a045IOY03rnHYC3YxvhBAuIA0oAZBSPi6lHAqciGquvqP1/q+klFOBTOAt4F9H+bo17bikw1jTOgerEMJh3FAherUQYrAQwg48ACyVUhYJIYYLIUYIIaxAExAAokIImxDiMiGEV0oZBhqAaIe9Ik07jugw1rTO4X2gJeF2KnAv8DpQChQC01uPTQb+iuoP3oZqvv5D62NXAEVCiAbgv4DLj9L1a9pxTUgpO/oaNE3TNO24pitjTdM0TetgOow1TdM0rYPpMNY0TdO0DqbDWNM0TdM6mA5jTdM0TetgHbYcZnp6uiwoKOiop9c0TdO0o2758uVVUsqMXe/vsDAuKChg2bJlHfX0mqZpmnbUCSG27el+3UytaZqmaR1Mh7GmaZqmdTAdxpqmaZrWwXQYa5qmaVoH02GsaZqmaR1Mh7GmaZqmdTAdxpqmaZrWwXQYa5qmaVoH02GsaZqmaR1Mh7GmaZqmdTAdxpqmaZrWwTpFGNfXR/j3v2toaop29KVomqZp2kHrFGG8dKmf887bwOefN3b0pWiapmnaQesUYTxypBuTCT79VIexpmma9sOz3zAWQuQLIT4RQqwTQqwRQtyyj2OHCyGiQohpR/Yy9y052cLAgUksWaLDWNM0TfvhOZDKOALcJqXsC4wEbhJC9Nv1ICGEGXgImHdkL/HAjB7t4fPPG4lEZEc8vaZpmqYdsv2GsZSyVEr5devXjcA6IHcPh/4ceB2oOKJXeIDGjEmmqSnGypVNHfH0mqZpmnbIDqrPWAhRAAwBlu5yfy5wPvDUkbqwgzV6tAfQ/caapmnaD88Bh7EQwo2qfG+VUjbs8vCjwJ1Syn3OLRJCXCeEWCaEWFZZWXnwV7sP+fl2una16X5jTdM07QfHciAHCSGsqCB+SUr5xh4OGQb8UwgBkA5MEUJEpJRvJR4kpXwGeAZg2LBhR7xzd/RoDwsXNiClpPVaNE3TNO2YdyCjqQXwLLBOSvnwno6RUnaXUhZIKQuA14Abdw3io2HMmGR27gxTVBQ82k+taZqmaYfsQCrj0cAVwCohxIrW++4BugJIKTusn3hXRr/xkiWNdO/u6OCr0TRN07QDs98wllJ+Chxwm6+UcsbhXNDh6N8/ieRkM59+2sjll2d01GVomqZp2kHpFCtwGcxmwSmnuFmyZNfxZZqmaZp27OpUYQyq33j16hZqayMdfSmapmmadkA6XRgb/cZ60whN0zTth6LThfHJJ7uxWIRe/EPTNE37weh0YexymRkyJEn3G2uapmk/GJ0ujEH1G3/5pZ9QKNbRl6JpmqZp+9Upw3j0aA+BgOTrr/WmEZqmadqxr9OGMehNIzRN07Qfhk4TxiUl8a+zs20UFtp1v7GmaZr2g9ApwnjtWnj6aVi9On7fmDHJLFnSiJRHfD8KTdM0TTuiOkUY9+kDubnwzjvQ2NoyPXq0h8rKCJs2BTr24jRN0zRtPzpFGJvNcOGFEInAm2+ClDBmjO431jRN034YOkUY19VF+PnPN2I2V7N5M3z1FfTp4yQ11aL7jTVN07RjXqcI4+RkM19+6eejj8ooLIR586CmRjB6tEdXxpqmadoxr1OEsckkuOaaTD75pIGBA1uwWOD112HUKA8bNwaorAx39CVqmqZp2l51ijBuaYEzz8zEZIJXX63kRz9SU51cLtVvvGSJro41TdO0Y1enCGOrFTIybIwfn8Lzz1dwwgmSgQOhvNyN3S50v7GmaZp2TLN09AUcCWYzrFwJw4Zl079/PS0tMc4+20xRkYmcHDeLF+vKWNM0TTt2dYrKuJlmMkdtxp3iYuzYbjidZpxOOP986NLFw/LlTbS0RDv6MjVN0zRtjzpFGJuFiaaMIoaOr+OrryS/+10tpaUhCgthwgQPkYjkzTf1phGapmnasalThPEXfMHN3ExwyBd4PEF+9av1zJ5dAcANN6hBXC++2EBzc0depaZpmqbtWacI46HRoTgjSbzOa/zsVjO9eyfz1lsVhMOS7GwrvXo52bKlkf/7P2jSBbKmaZp2jOkUYfzR29tpfK4HX4mv2GRfwz33ZLJzZ5C3324gFILTT/dQXt7Izp2Sv/4Vqqs7+oo1TdM0La5ThPF55/Wj//IJmGJmng2+wMUXp+HzmfnLX8p59FEYONBDQ0OUMWNaCATgr3+F7ds7+qo1TdM0TekUYQyC3952JXnrBrHItoBtTTu44ooMtmxpIhiUlJQkA7BxYwM//Sk4nfDCC7BmTYdetKZpmqYBnSSMhYDsbAcXN/+UkAjyqw2PcP/9eaxdO5iJEwVOp53sbCePPlqKxxPj2muhSxf417/gs886+uo1TdO0412nCGOA5GSY2vM0+ocGsXXoZzz+1AKSkgRdu8Y44wwYN64bmzYFePzxclwumDED+vaFuXPh/fchFuvoV6BpmqYdrzpNGAOk+AQXRy+n2d7Ik6Uv8uSTaxg06Gtqavz84hc+hg3z8sADO6iuDmM2w49/DKNGwRdfwCuvQCjU0a9A0zRNOx51mjCurPRTVtbAJMc4Ciigyz1bmXX/v2lsbObNNytISxP8/vcF1NdHueOOYp56CsrLYdIkmDIF1q9XA7uKijr6lWiapmnHm04RxlJKzj33RcaNm0NjeZSpTKU5s4rgiK34fCv5178qsdmidO2axKWXZvH3v5exdWsLzzwDCxfC8OFw6aUQCMBzz8Frr0GD3ltC0zRNO0o6RRgLIfjd7yZRXNzA+DOeZWzTeFJJ5aQnqikt3Ulz8wbmzasiMxNuvjmfpCQzq1YV0bcv/Oc/KoDT0+HnP4exY2HtWnj8cVi0CCKRjn51mqZpWmfXKcIY4NRTu/Pee1eybVsd/33NQiYzmZLcdUz8ZQawkT/8YRVuNwwYYOWmm3KZN6+OpKQ6pk2DykpYvRpsNhg/XoVyYSF89BE88QRs2NDRr07TNE3rzISUct8HCJEP/B3IBmLAM1LKx3Y5Zipwf+vjEeBWKeWn+zrvsGHD5LJlyw7j0vfsk0+2cPbZf+e1beO4N+M2RgdP49WCXAKBIKtX30xurge/P8aJJ67A4TCxdOkgTCaBy6W2Yly9WjVRDxumFgZ5/32oqoJevWDyZFVBa5qmadqhEEIsl1IO2/X+A6mMI8BtUsq+wEjgJiFEv12O+Q8wSEo5GLgGmHO4F3wwokT5Pb/nWZ5l3LhC5s6dQaGzK2dwBh/bP+S5uRMIBEKceuozrFlTjttt4o9/7MbGjS386U/lxGJgav1NbNqkpjs98giUlcG118LEiSqYn3gCXn0VSkqO5qvTNE3TOrv9hrGUslRK+XXr143AOiB3l2P8Ml5iu4B9l9tHmAkTO9nJIzzCl3zJaad1J8+dxUQmImOSpX3e5e23r6a0tJmhQ//CO++s54ILUhk7NplHH91BUVGE6mqQUu2BfM01kJ0N8+ervmOXC26+GUaOhI0b4emnYc4c1bes5ydrmqZph+ug+oyFEAXAEGDpHh47XwixHngPVR3v6eevE0IsE0Isq6ysPPir3Ydf8AtyyeU2bqOSSpJIIjWUQeHOwbwS/Re9hzkZN+5sgsEkpk79O/fcs5CHHupKbW2EOXOKaWiA0lIIh6GgAK66Cn76U8jNBasVPB447TS4/no1HaqhAf75T3jsMTVPORg8oi9H0zRNO47st8+47UAh3MBC4H+llG/s47jTgF9LKSfs63xHss84RhM7+YL1BLmDXzGQgTzLs5RRxpfRr/iV+R5sj4/i7amPMn78BkpKviIQ2ME55wzA7R7C66/XsXTpIHw+JwBeL/h88aZrw4IF6ta7NwwZokZaf/EF7NgBDgcMHar6mtPSjsjL0jRN0zqZw+kzRghhBV4HXtpXEANIKRcBhUKIozbUyUQNeayjAAd3cAcrWMFa1uLDR3dzAb0bBtB84deMnfAE//M/HmAI3bsP5r33VvP11/Mxm1uYNWs7+fmQlAR1dVBcDM3N7Z9nwAC1YteOHfDyyzBvHvTsqfqVe/aEzz9XlfJf/wrLlkFLy9H6DWiapmk/ZPsNYyGEAJ4F1kkpH97LMT1bj0MIcRJgA47irsFZSAQ+mhnEIN7nfQYyEBcurFi5LHkaltwASVd+x4AByTz9dCG1tQU88cTFlJbWIsSnvPXWJj79tJ6sLLWJhBBqAFdZWXyucVoanHUW3H47TJ8OmZmqabtrV7W05kUXqalRgQC8/Tb8/vdqM4qNGyEaPXq/DU3TNO2HxXIAx4wGrgBWCSFWtN53D9AVQEr5FHAhcKUQIgy0ABfLA23/PiJsCNLx0cIOQtiwAfAe75FFFr3oxXCG883ML2gUW7hy0BhWr/6G4cP78eWXN/CjH73I5s2fc9FFzSxadC6FhQ7y8qC+HmprVSWckqKar4VQU6D69VM3I2Rra9X61nY7nHCCarKurlZTpVavVn3OAwfCoEGQlaXOo2mapmlwEH3GR9qRnmcs5UJgHdvFRHykEiPGJCaRQw53cRdd6MIt3MI2tvFo/ZNc1n8BpaWNTJgwlFmzTuOWW97iiy+2YLen8Oc/X8iPftSD1FR17qoq1WRttaq+ZLd79zCNxWDLFhW869ap6tjhUNVyOAwrVqgKORaD1NR4mOfm6mDWNE07Xuytz7hThPHHmz/mHyt+w5xpFwEXAVnqfj7mJm5iAhP4Jb/EjZvLuZx66vlLwxx+e+0aXn31W5zOFBYtuoy3397G7NnzkTLI1Kknce+9kykocOHzqdHS1dUqWC0WVSV7PLsP8gLVrP3ddyqYzzpLhffKlWolL5cLKirUhhSxmDpP375w4omQn7/n82mapmmdQ6cO481Vmxn39HB2/Op+4FSi9KeSStJJ5zEeYw5zuIEbuImbKKWUS7kUgeAf/IO//KqYBx74AIfDRE3N3SxcWMePfvQ2kch3eL0O7rjjLG67bTgOhwkpVYVcV6fC2WxW+yh7vfsP0U8/VZtSBIOqwu7eXYW036+COxJR3/ftq0Zrd++ulufUNE3TOo9OHcZSSjJ+k8GGO+4hzdWPFsaynvVkkEE22VzFVaxmNa/xGr3oxUY2ciVX4sXLS7zE7FureeyxjcyZczIzZmTw0EPLuf/+WoRYQ0tLJUOH5vKXv5xLQUF+25zjcFiFckuLamb2etXNbN77dYbDqiLesEFt2ejxqHnLwaBaB7uyUvVPG9V39+4qmHv3Vn3WmqZp2g9bpw5jgLOfO5ubRvVkygmDgRlsYzu11NKPftRRx9u8zSmcwgmcgBkz3/ANP+EnFFDAc9EXuGhSMYsXN/DnPzu59tpX6NUrkx07CvH5Yki5hoqKJqZPH8qNN55BTo4Pt1uFL6hQbmpSoexyqWrZbt93X7BRZbtcKnwffFBVx16vGqUNKpzr6tTXmZlqfezevdXo7X2FvqZpmnZs6vRhPOvDWVQ2/Yc/nXcRcCUhHKxlLSmk0I1u+PGziU0UUUQXujCSkXzKp9zIjQxkIH9qeZpvPo9w+ukeXnllFffcM5+iolrM5ky6dDmRs88O8NxzXyIEXHbZMK699nSys71kZKgKNxRSo6/9fhW0VqsKZbf7wIKzoUEN8Nq0STVbB4NwxhnQvz+sWaOW3iwrU+e22VTVXFio5jenpelBYJqmaT8EewvjA5na9IMwsttI7nz/r63flWKjDxlkUEEFmWTixk0WWdzLvexgBw/zMOMZz2/5LbdzO79y3sFjZzyGCRNmcy7Z2WcxY0Ydf/zjQkpKvuLjjyexZMnPmTNnCc899xUvvbSMK68czsyZY/F4vITDKoDz8lTTdWOjGvBVU6OqX49Hja7eW2gmJ6vVu4YNU9OlduxQI7d9PlUV/+c/4HRCRobqny4ri2/t6PWqYC4shB491PNpmqZpPxydpjKua6kj/TdpNM1+FLtlIDCWCBG2s50ccnDgQCJZxjJmMYutbGU2szmP83iZl5nFLKYwhQd5kHffaOSKKzaTlmbh//6vG1u31vGzn1WTlSUYObKE668fzIsvruD555djMgmuu2441103FpdLtVsnJanwNZtVpez3q5HTVquqlN1u9fWBamlRFfOWLbB5swp6gCuuUM3Y69ap8A4GVdhnZ6tQ7t4dunVTTeaapmlax+v0zdQAJ/z+BP556SUMzu0DTN/jMUGCrGAFD/MwK1nJ3dzNlVzJHObwR/7IUIbyJ/5E0Tc2zj13PTU1EV58sSfZ2TbOOWcRtbWfYzJJLr10EFdddRKvvLKSF174GrPZxE9+MpzrrjsVn89HNKoq2S5dVBA3NakQDQTUddjtKpRdLjVY60BJqeY9FxWpKloIeOst+PprVUW7XKrJvLqatq0hc3Pj4Zyff3BvBDRN07Qj57gI46teuYr+Wc3cPvZ0hPgptK7EFSJEDTVkkYVAUEMNm9nMHOaQQQazmY1A8B7vcQ/3kE02T/IkztI8zj9/A0uX+vnmm4Hk5tq4/PJVzJ//NSbTNqSMcv75J/K//3smf/zjp7zwwtdICRdd1J8bbxzD0KF5JCWpZueyMhWUDocKZL9fhSao+4xgPpSBWeXlqnLeuhW2bVPnTUmBc89V961bpwJcShX8eXlqZ6qCAh3OmqZpR9NxEcZPfv4kb6x+jA9/+jPgXFpX7KSKKnawg+50x4cPiWQb26iiil70wouXKqpIJZWVrORn/IwQIR7lUU4KjOT116u57LIMAEKhKC++WMXNN28gGt3KoEFRli5VO0YuWrSVt99ex1//+hUNDUHGjOnGbbedysSJJ1BTY2rbZtHhUMFrt6sR1U1NakQ1qGra5Tr0YI5GYedOdd4+fVQA/+536jmSk9X5g8H4KG2zWVXORjh37arnN2uapn1fjosw/rr4a8Y+PYr6WX/AJIYDIwCQSDawgSBBetGLJJKIEmU96wHIJZdpTGMwg7mf+6mhhhu4ge/4jl/za37MjwFYtaqJs89ezx/+0I0hQ5K44ootLF3q54or0rn77jQGDXqEwYO7cN11w6mrC/DEE5+zbVsdPXumceuto7j00qFIaWsL37w8FXyhkJrW1NKiQtPYmMLhUP3PLtehV6+xGJSUqGbtoqJ45Tx8uJom9d13qnKur1fBbTJBTo7qa+7WTYVzUtLh/V00TdM05bgI43A0TPK9yWy6czZ53kLgvLbHQoTYyEYkkt70xo69bbpTCinMZz6P8ii96c1jPEYaafySX7KYxcxgBrdzO+tWB5kxYzPLlzdx9tk+HnusOy++WMns2cXk5Fi46KIm3nvvazZsqKJLFw/XXz+crl19PP30VyxduoOUFCfXXjuM6647mW7d0toCtrxchbARvlarCsympnhTttUar5httkOfymQ0mTscakpUcTE884x6zOjHjkbVVKtYTN2fkdE+nH0+PZVK0zTtUPwgwjgcDlNcXEzAGOV0CMoay0i220iyWQE3EE8NiSRMGBMmLK2zuqKtHxYshAhRSy0AKaRgx04DDTTRhB07KaQgpInGxih1dap89fks2O2CqqoIkYjE7TbhcEBTU5CWlgi5uclYLCYCgQiNjUGam8OAxOGw4vHYcTotgEDKePhBfHcoKdvfEh9PvB2OaFTdIhF1i8VU6AsRr9oTr81kUtdmsajPZnP7a3A4HOTl5WHVndGapmnt/CDmGRcXF+PxeCgoKEAcYsK469xEYn66p6YBKUD7QIgSxYQJ0RrSEkmQIDFiOHC0TYcyY6aAAgSCaqoppRQ7dvLJx46dYDDG9u1BkpJM5ObaiUYlJSUhKirCmEyC3r2tpKaasNvVr3jTpiqcTklaWhKhUJTq6iZCoSjhsJmMDBfp6UnYbBZiMRV8iWEcDqvvjfWvjWMMxmMm05ELZ+Ncfr+qko03AkbwShnfPlIIVa3bbGC1ShobqykuLqZ79+6HdyGapmnHiWMqjAOBwGEFMYDL5qK4vqr1uzC7hrEZNSpKIokQwYIFGzYCBAgRwo6dHvQgRgyBIEIEL15s2CimmC1sIYccfHYfPXs62s7r90cRAvr2dbJzZ4iSkhCVlYK8PElKioXkZDvl5U1s21aLxWImLc2Jw2Ghri7Azp2N7NzZiM/nICPDRXKyvd3vQIj2ASxEvA/ZuN+obiEezEagHqzEgWNut2o6D4dVlRwKqa+zstTz1der7yMRWgeoCaRMo6KikhUr1Gjt/Pz4YiWapmna7o6pMAYOK4gB3DY3oWiUaExiNoX3elyUKGHCxIhha/0IESJAAAeOtmbsEkoIECCffAoppLj1o4kmskV2W7g3NcUoLw9TXR0hO9tKRoaVkpIQ330XxOUKk5+fxIABbhoaglRWNlFe3kRubjK9eqUTCISpqmqmqqqZuroWrFYzaWlJpKUl4XRasVppa8aOxeJVqhFuUrZv0t41uA+3ajaZVH+ysXiIlOo8FouqhqPR+GhwAKtVYDKpFcK++UbdZ7erUdt5eSqc8/L0SmGapmmGTlerWM1WrCYrLeEoqjLeM0vrhxHKRoWcJJK47IrLiKHSLDWSyikZpzD5nMk8/fzTXDD4AqYPnk6BrYB+A/oxaPAg7rrrLubP/wfDh3v49tuFFBeH2LYtyPLl7zF8uIf33nuD9etb2Lo1iMNho2fPNAYNyuaDD16jV69e9OvXl+eee57UVCf5+V5cLhtlZX7WrCnnmmtuolev3gwcOJBp086nqamu3QAuKeHnP7+Jk04azIAB/UhOdnLyyYMZMWIwb731GhAPS6OqNfqAD3W4QGKgezxquc4uXdSAsORkVUm73XDnnfG1uW02NVBt8WJ46SV46CF45BF49VX4/PP4blWapmnHo2OuMj5cQghcNhcNwRbcdgsQBfY8YdeKta25WiCwYsXlcrFm9RrqWurwOX18+uGn5OfmY8HCuKvHMfXqqeSQQ2FBIS988gLJ6clkk83bL7zNgAEDWLz4DS64YCIlJSHeeONfDBo0iPx8G9nZVsrLw9TWRkhNtWC3+5k9+36WLVtGc3OYUaNGcOqpZ+LxeElKspKT48FkEowceRrXXXcHVquVZ555iPvuu59HHvkDJpNKRLMZnnzyz8RisHVrEVOnnsOXX65oC+xoNB664XAE40++a5N2Yp/0oTCZ1AhtR7zlHilh4kQ1Yru4OL7RRd++qjIuLlabY6xaRdtrycpSjxm31FTdvK1pWufX6cIYVL9xbUslOckeVHW85zAWCGzYCBJsq44BJk+ezAfvfcAF0y7gHy//g0svuZTFixeTTTY11CBaP7rTnQABSimlhhrGnDqGTxd/isMRIzc3wo4d3zF48GAaG2PU1UXp2tVOS0uMysowc+e+w4gRZ+BweElNNTN58kSKi1dw1llTqalppra2hX79MrnuuoupqmqiqSlMnz6DmD//XVauLCUlxUlqahJutw0hRNvoZmMwlVG9nnbaGMaMGcuSJYs577wL+PrrZUybNo2pU88jFoOUFDfV1X4Afv/73/Lvf79BMBjgwguncd99vz6sv4PJBIMGqRuoyteYVpWRARUVao6zwWZTfdAVFfDll+o+h0PNezaauHNzVfWtaZrWmRyzYVxVFZ9ju38SVQGbABPRWAbmsJudO+0ktsTbbJCe3v4nBQI7diSybYT1pdMv5TezfsPkcybz7cpvufqaq1m8eDHppJNKKqbWc9ZSS296U0stAQLUi3rGThjLvHnzqK+v59xzz2Xr1q1Yreq827YFcTpN5OfbaGkpJyUlh7VrW/D5zGRm5lBeXkpWlpusLDfRaAwhBNFojO3b6wF4771XOe+8C/F47FRXN1NZ2YTVaiY1VQWzMU0tsRlZCGhqamDBgkVICVdeeTmxmKpCjYrTbIYPPnif4uLtLF68FCklU6dOYcGCzxg9etQRqZxBDTrLz49/n5kJd92lVgwrKVG34mKYPl3tRPXNN7BihQrwrVvjFb7Ho0I5Nzce1HphEk3TfsiO2TA+eMZcHxMmoSrhmAST2H/HqFHpGk4YeALbirbx5stvMnHKRMKEkUjj7IAajV1FFWbM5JFHOukIBKOmj+L5x58nWB/k4T8+zAMPPIDLZebEE51UV0coLw+zfXuISCRGdraVnBwb5eVhKisjhMMRGhujuN0mzGb1PCaToFevNH7zm/uRUjBy5CQaG0N065aCEFBT00JFRRPl5X6qqqqIRGI0N4dxOi1tg+EuuWR622YUxrzgRNEozJ8/n7lzP2DEiCGtU5r8bNmykVNOGbXbHOPEwWCHO40qKUntydyzp/reCFwh1Aph5eUqrBMHreXmQmUlrF8fP09KigpmI5xzcto3mWuaph3Ljtkw3rWC3TcBtAAhIB0QrCnfjtPtIc3laL1v/2WdEbgRIkw5dwr/fft/8/GCjymtLiXW+mGEsUDQla4ECLCVrQQJ4sXL2JPH8pvVv8HpdJLXO6/t3F9++SXXX389AP/93/fRp083vqEA4RgAACAASURBVPhiERkZFqJRid9fSrduo9mwoYWkJBOZmVZSUy2YTII33vgnS5b8hw8//JBw2ExNTTO33noDq1Z9S2ZmNs888zIWi4n6+nLC4Rhr15Zjt1tISXESi0mSEspGi8WClCpdY7EokUikdZqU5O67ZzJjxk/aFvMwpkwlDhbbdY5zYtV8JMI58ed79FA3KaG2VoXyzp0wfry6vrffhmXL1HMHAmqzjDVr4j+flqZCuUuXeFDrgNY07Vh0zIbxwbMDQVQfsQ2X1UVti781jCMYOzjti1EdW7BwxTVXkOxNZuCAgVQvqAYgQAA79rbpTC5c5JFHKaVtK3XlkMMDDz5Ao6OR7/iOFlqIEWPEiBGsWLGi7blqamqYNWsm27dXsW1bkE8++YjbbptFZqaN2toIRUVBiotDrF79Mb/97UMsWrQQt9sNQEqKkxdf/BsAlZV+tm+vR0pJMBjFYhFkZXloaQlRVuanuTnMpk3VpKfXk5LioFu3bixfvpwLLriAN998k2g0iskEkydPZPbs2Vx22XRsNhfFxcVYrQ58vvi7IiN0jWlUicGcuACIyRR//EgMvhJCDeRKTYX+/eP3n3aa2hbSCOnSUjV6+/zzVZP3t9+q6tkYIAbqHEYwG0HtdB7+NWqaph2OThTGRtgGARsuu4viuurWObFhDiSM42ey0T2vOzfeciNBgghE26pdQYJYExYSMZqpU0nFgSq7pk6eikRSQQUBApRTTjPNJBGvUFNTU7n33nuZOHEUUsKtt96DlMmUloZ46KGfcdNN/0Vu7iDuvvtWwuEgY8dOwGIRjBo1kqeeeqrtPBkZblJTk6irC+D3VxKJSKqrmxk0KJtoNIbNZsZuN1NR4ae8vJFTTz2f22//CXPnzmPSpInYWycPT5kyhfXr13PKKSMB8Hg8/OMf/yAzM71d8MZiaqCYEbiJC4sY85yNBUiKilQ/vTHK2m6PDzI7Enw+dRswIP78TU0qkHv1UtX0hg3tp0yFw2oa1erV8fuMJm6jgu7SRc+B1jTt6Dqm1qZet24dffv2PaTzRWIRTKIJk7AALlrCLawpX8PgLl2xmG2A76DPGSWKRLaNsk5cOtOMGRu2dn3NhmKKCRAgl1yiRCmhhDBhMlo/THtpMg+HY1RVRQiHJV27qpAsKwsRCMSoqYm0rhltIiMj3oS92zVHYwQCEVwuG1JKVq0qJxqNkZxsx2o1EwxGaGgIIqXEbDbh9TpISXGSnGxv66feF2PBD1DzlY2K2GAyqcBdu3Yd2dl9CYVUE3LicppGMDscKqy/z6lLUqoR2qWl6uZywYgRapnPhx9Wr8FYzSwxtL1eFcqJt+RkvUGGpmmH5wexNvWhCkVDrCpdRZ43jyxPFgAOiwOTMNESieAxC9SI64P7n9ScMCXKmItsx06YMBEiBAliw7ZbuHrw4MfPd3xHOun0oAcVVFBJJQ00kEMOLnYvvaxWE126xCv4UChGcbEaUu7zmbHZTNTXqybsHTtCpKVZSE+3kJQUv06z2YTLFT9H165e6uoC1NW1EImoEdq5ucnY7Rbq6lqorw9QU9OMEILkZDs+nxOfz4HVupfpYAm/QqNvOXHlL+MYk0ltCQnxhT9ABV4wqCpYg7G61/dRPQsRr6AT3+clJcGll6qALitTn6uq1DSs7Oz4HOjEQWJOZ7xyNm56HrSmaUdCpwhjm9mG3WKnPlBPlicTiCGEWS3+EWjBY7eg+o0PbRchY7cnicTa+mHC1LZ8ZmI/MoAXLy5clFHWFsD55JNMMjvZyVa2kkIKWWS1Vd17fF02E/37J1FZqZbZjESiOBwmunWz0dgYpbIyTEVFuK1aTkmxYDYnrmktWsPViZQ+/P4QtbUtJCVZSU52YLdbCAQi+HxWpJQ0Noaor69l2zaBy2XF53Pg9TpwOq17XaY0cdDWrqO0XS5VFTc1xYPa7VbbMEYi0NioKutQSH3d0BA/z64BfaQDz2RqP4ob1HVEoyp0q6uhrk6N5jb2lw4E1P1FRfEWAatVhXeXLvHPmZmHvv+0pmnHp04RxgA+h49yfzlS1iGEBFJx29zUNFeS6zUW/zi0/yEFAgcOQoTarWftwEGw9cOGrV2wWrCQRx5evJRTjhkzDhz0pCeVVFJNNY00kk02Xrx7bO4GcDhM5Ofbyc21UVMToaoqgs9nJiPDSm1thPr6CH5/jKKiINu3J1bLpl02mxB4PHY8HnvbfdFoDCmhqkqVqXa7hbS0JKxWM42NQUpKGigpacBms7QFs8dj32Pz+J6kpKjPxs5TgQBtU6xA9elCfK9moyI2qufm5vixVmv7cD6cPZ33xpYwrCAtDa6/XoVudXW8eh4yRFXDn38OH36ortWYJ53YMpCeHg9o49Y6/k7TNG03nSaMvU4vZf4yWsJRkmwxIIrL5qK0sZSYFJjE4S18bKzWFSFCmHDbhhJGIIcIESOGFWu7YPXgwY0bgUAiKaUUDx560IOd7KSYYuqoowtdsGPf6/ObTIL0dCvp6fE3FFVVYerro1itgtRUC7GYpKoqTGVlGIfDRFqahbQ0CzbbnstKj8dOv36ZhEIR6uoC1NcHqKsLMHBgNnl5XqqqmvD7Q4RCUaqqmqio8GMymUhOtuP1qnC22fbcnN3ud5ewxWL89aiACgbjwRuLqZW5UlJUlVpfHx8oFg6rY/z+9uc0Kmi7XQX2kQ5os1lVupmZMHBg/P5hw9TSnWVlqnouK1PN3Oecoyr89evVNKuVK+M/43K1D2ijmXvXFgVN044/nSaM3TY3ZmGmpqWJJJsTCOKyqX7ZUCSGw3r4uxAY61ebMLVtsQi060eOEsWOvV0/snFclCgBAtRRhxs3eeThx0855WxmM5lkkkbaXgd47aqw0EFDQ5SqqjA1NaotNS3Ngtttpro6QkmJ2srR4zGTnm7B52vfjG2w2SxkZrrJzHQTi8m2yre2VvUpg2q2djhU4Dc1hairUx3CSUm2tmB2ufbenL0rk0n12xpToKVUzcFGMIVCKngTF/uw21XfbzSqquxdm7dNpvYB7XDE918+0pxOtShJ797x+8LheHWfkaHmQJeWGltLqub6xka1mpjRzG2EvRHSWVnqs55upWnHl04TxkIIuqd2x2GxA81ACKs5CZvZhj8UxGF1sK9NIw6GufUDVMBGiLQN5DL6kW3YMGNuVyVbsFBIIdVUU0EFm9lMOun0pCdllFFOeVuV7Gb/bZomk8DnUyEbDseoro5gswlSU1X/8XffBTCbBU1NUbZuDWIyBUlJUdWyx2PeY3AmNkH37JlGc3OY+npVNVdXN5Oc7GDAgGxaWsJUVDTR0hKmtLSR0tIGLBYzXq+qmmOxgxuln7hHM6gmXZdLBa5RPQeDKmzNZhV84bAKLaPp29idqr4+8fW0r57t9u8voBOvf+BAdZNSvVkoK1PXdfLJ6k3H88+rKVbRaLy6TlxMJTk5XkEbAa2raE3rvPYbxkKIfODvQDZqzclnpJSP7XLMZcCdrd/6gRuklN8e4WvdL5/TmL4URQVyrHXTiCbSXQ72tWmEQQjB5ZdfzosvvghAJBKhS5cujBgxggsvvJDHHlMvfe3atfTp0weT2cSESRPoc0Ifrr/6euZ/NJ9Tx59KiBDvvvkuF19wMa+++irTpk1T50eQTjpevPzpb3/i8dmPY8XKzJkzOf+q8ymjjCKKSEbtBvWrO37FO++8g81mo7CwkOeffx6fr/00rZtuuoklS5YQCoXYunUrffr0IRaDK6+8ndNPPw+zWeD1qtddVxehujqCxaKatlNSLLjdpj0GsxACl8uGy2UjJyeZcDjaNiLbajVTVdWEEAK324bNZiYajVFXp0K7qqqeG254hilT+jB5cm8GDsw+6L2qhWi/j3Iim00FcWLfshBQUBAPwHBYBVw4rAZjGYwtHY9GQAuhpkl5vfH7LBaYMUMt6ZnYzG2zqbAuK4NFi2DzZjWiO/G6MzNVOBsBnZWl+6I1rTPY7zxjIUQXoIuU8mshhAdYDpwnpVybcMwoYJ2UslYIMRn4HynliH2d90jPMzbUNtdiEjG8Tg9gpbyxguL6HZyU2xUhHMC+t/xxu9306tWLzz77DKfTyQcffMDdd99NXl4e7777bttxBQUFLFu2jPT0dKJEefaFZ3n84cc5+eSTeXbOs0SIcMnFl7Bxw0ZmzpzJj6f9uN3z1NTUMGzYML5Y9gVWYWXo0KG8u/xdClMKqaeeKqqQSNbMX8N5Z5yHzWLjzjvV+52HHnpoj9deVFTEOeecw+rWFS3UCOko1dURamsjhEIRBg70EAxKqqtVf7OUtPU5p6buPvBrb6SU+P2htqq5pUV1A3TrloLTaWHlyjXccMOnrFxZBkBubjKTJ/dm8uTeTJjQk+TkI7cuZSymQjkSUZtIgBpQZTQPG+Frs6kgNKrsxHnFiRW0EdRHcorVwfriCzUwrLxc7WIVjaqm7+RkFdaJU8OczngwG7fMzPZ99JqmHRsOeZ6xlLIUKG39ulEIsQ7IBdYmHPNZwo98AeTRQSr8FURiEbzONEBtpyiBUNSE3RLkQOYbT548mffee49p06bx8ssvc8kll7B48eK9Hm/GjAVL2xaKgXCAaDDK1s1bGTR4EGHChAi1G9w1b948zjzzTDJTMwEYd+Y43pv7HlMumUI66RRSSAUV9DurH1vZSjbZjBg5gtdfe/2AfxdCCKZMOZ2xY8eyaNFipkyZyhNPfMO0adMYMGASVqvglFOyWL68koqKMA899FsWLnyHSCTIhRdeyP3337ePKU3x0dl5eV5CoQj19UG8Xjs2mwWr1czatRWMGtWVnJxk6upaeOWVlcyZswyLxcSYMd2YPFlVzf37Zx101ZzIZNq9jzU7O97EbXw2mdQoaVBh7XTGm32NCtqYG22c1whm4/P3MUhsT0aOjH8di0FNjfo6PV1d41NPxUejt7Sofuht29o3dRsLlySGtG7q1rRj00H1GQshCoAhwNJ9HPYT4IO9/Px1wHUAXbt23e/z7dy5+30ul/pPJhZTFcJuj1uzKYtsoiXYQm21DSldeMJ9KCu10S2/HrWZxN5HLQNMnz6dWbNmcc4557By5UquueaafYYxqN2czMLMhAkT+GjeRzTUN/Cjc39E0dYiTJh2G9xVUlJCfsJ+ggV5BURLoiSTTCWV1FJLJpmkkEIZZexgB39+7s9cfvHl+/297aqhoYHFixcBcPnl6udTUszEYhIpwe+PsnTpfGprd/Laa4uor49wyy0X0rv3x4wbNwafb+9N2QabzUJGhiXhezO3334qc+du5LXXVKXepYuHV165hIULt/L++xu488653HnnXPLyvEya1OuIVs1mswrbxJA2GoFUa4AK6MTw9XrVSO5gUA20isVUtd3Q0H43KaPKNkL6+15FzGRqv3GK0wm/+IV6k1FZqSrnigooLFTXv3atmnZVXx8f2Z34e0lL272K9nr14iWa1pEOOIyFEG7gdeBWKWXDXo4ZhwrjMXt6XEr5DPAMqGbqg77aA5BkTYIgBCJ+wIsQArMwEY5FURVxgP2F8cCBAykqKuLll19mypQpB/zcAsGl0y/l8ccfp76+ntl/nM0fHvhD29KZxuAuK1ZiMrbbz1uFlXzySSONMsqopZYe9KCQQmb+70yERTD8suHsZCeZZO5zwZBE06dP3+2+1FQrqalWTCYoKLDzxBMfs3jxPC6+WK2VXVfXSHHxZioqRlBeHsZqVYPFUlLMuN3m/c41ttnMPPjgRB58cCKlpQ3Mm7eJ1avLW6vhPmzeXI3VaiYvz0tjY6Bd1Tx6dDcmT+7NpEmH1te8N8ZphFABBCpwQyF1M0LVbG4/hcpqVU3WNlv8eGNktMGYB50Y1N93BWqzxfd1TjRyJHTrpgK6vDx+GzZMvanYskVtnpG4gYbFoprBs7PjU7kyM/USoJp2tBzQ/+ZCCCsqiF+SUr6xl2MGAnOAyVLK6iNxcTk5e3/MZNrb41ZKAnaqW+romSMBJ9tra6lqrkLKXgjRghqHtu8y4Nxzz+X2229nwYIFVFcf+Ms5+eSTWb16NU6nk369+7Wt3rV06VJuuv4mJJKZs2aSlZfFkgVL2n6uuLiY008/HYAkkuhOd6JEEQie+9tzzH13LvP+M4+QCFFLLTddfRObvtlEfk4+H7y/x4aINq6EXQ8sFgux1rbMaFRtoZiebsXrNXPffffyk5/8hObmKOvWtSAlmM0Cp9OElKqfubIyjMWiBoR5vRa8XvMep0sl6tIlmRkzhra7b8SIfIqLG3jnnfVIKUlNdXLllUPIyUnmgw82ctdd87jrrnl06eJh0qR4X3NKypGd82MyxTeyMFitkJcXb+I21tdOTlZTsVpaVEWauDJYNKruN0Ic2g8UM0L6aDRzW61qlbNdG5+MdcVLSuDrr1XLUkVFvL/dYlHbUH7zTftz7RrSRt+1DmlNO3IOZDS1AJ5FDdB6eC/HdAXeAK6QUm7c0zFHk8/hoyHQgJQ2hFDzjSuaKghGwGEFtbPTvv9Tv+aaa/B6vQwYMIAFCxYc1PM/+OCDOBwOrFjbpjcNGTGEpSuWYsNGlChlNWXcd899lNeWY8XK/PnzefDBB9vOIRBYsDB37lx+99DvmLNwDuVJ5aSSSne689jzj9FAAxYs1FKL7wA3wigoKNhtC0WAiRPVForTp0/H5XKRnl5LKGTFbPZRXx8lGpX07u0gGoWamgh1dWpgmBDg8Zhbp1iZ97rAyK7uuOM07rjjNKqrm/nww03Mm7eJ/v2zuO22U7nvvjMYNeopsrM9NDeHeeON1Tz//HJMJsEpp3Rl0qReTJrUm5NOysH0PbSt7mmRksRxjsYo71Co/SCw3FwVaH6/CmZj7nRiU3jiuRNvR6Mf1wjPxGpadVOoUC4oUNfxySfw2WfxQW7GFpWJrFbV3G0s/6kraU07PAdSGY8GrgBWCSGMDXnvAboCSCmfAn4NpAF/aW1SjOxptNjRkuvNJc+bhxABIITbrirDxmAAh9XEgYRxXl4et9xyyyE9/+TJk9u+NhYKsWIlhqpILVjITc3l7nvvZtTwUQDM/PVMUlNTAbj22mv5r//6L4YNG8bPfvYzgsEgN515ExEinDjyRP7nqf8hnfS25uwSSqimmgCB/V7b9ddfz9SpU/nwww8566yzdttCceTI9lsoFhSoOcN+f7RtbnJjowpnp9OE2SwIBGJs3x5k+3ZISjLh81kIBmPtFhDZm7S0JKZPH8T06YPa7isv9+P1Ovjooy2Ew1FcLhujR3eld+8MVq0q4957P+Leez8iI8PFWWf1YtKkXpx1Vi8yM7+/OT6JAZNYSSc2c1utqlKORHZfytNsVlOQjKU+Exc0gfh63IkBfTSqaPVmKj4KHWDcODj9dNUUb/RHNzbCqFHq6/ffj0/H2nXchs2mBokZfdFGJe3z6T5pTduXTrOF4p5IGUWIGqT0sLJ0Ay6bi57p2ag5yGkciQVADup6kAgEMWKECGHDhkQSItS2VeOuy2nuKkCACioQCPLJRyJpoIFyygkRwo2bLLJw7ufNxuFobo5SWxulvj5Cc7N6g+F0quU36+qi+P1Rqqo2M2NGhMmTfUyZksKZZ3rx+Q5ujRm/P8gnn3zHvHmqcn7xxYsYObIrb765hiee+JxIJMbq1eXU1KjS86STcpg0qTcTJ/bilFO67nXnqe+bUREbIR0MqmZsoxotL1dhbYS0EPF9oEOh9ufaWxXdkdVnU5NqpjcGjxmBPHCg+v6bb3Z/HSZTPJjT0+Nfp6a2X69c0zq7vU1t6rRhXOGvoNJfSb+sfggh2F63nUp/JYNz+mM21QNuIOmIPNfBihIliJoEa2n9iLR+GJX0rqt37SpGDBMmAgTYznYyyCBKlEoqiaJGZWeSiYMjN593T0KhGPX1UWIxSVaW2kN57doWyss3ceedgo0bW2hoiGE2w+jRHqZMSeHss1M48UTnQQ/MklIihODpp5dy++0f4PeHMJtNDBiQRVpaEi0tYZYuLSYajeHx2Bk/vpCJE1XV3KNH6vf0Gzh4fn98Oc9QSFXXNpvqp5ZShZyxXKYx5SpxypIx5WrX27FSeX77bbxqNqrqtDQVvJWV7RdgMSpzYxETI6zT09v342taZ3HchXF1UzVba7fSN7MvLpsLf9DP+sr1dE/pTpoL1MjqjvsPOkaMMOG2QVo2VAelUSWbMO1xr+RdNdPMTna2bVyRQQYBAlRTTYwYPnxkktl2/u9bLCbZsSPEpk3rOfNM1VbbtauNIUNcbNsWZMUKdV9eno2JE32cdZaXCRO8pKYe3I5aoVCEzz/fzvz5m5k3b1PrDlN30tgY4s4757J2bQVbttSwc6ca+N+zZ1pbMI8b16Pd7lUdyaiIo9H4SmM7d6qwTuR2q9AKhVRVvacq2hjxfbSbuvfHaJp3u1WLwWuvqdaB2tr4mwyHI/7GxOB0tu+XTk9XYe3xdPxr0rRDddyFcSQaYUXpCnKTc+iSbEdKJ6vKNuO0OumVngM00RFN1buKEiVEqG36k0S27QwFHFDTtURSTz0VVLQ1VeeRRxVVVKNGgqeQQgYZWA9xG8mDtW7dOkymAubOrWPu3Dquuy6T889P45NP6vj5z4swmwXffRfA748hBAwf7mbiRC8TJ/oYMcKDxXJw/9s2N4dISlJvOE455Um++GIHAN27p5Cf7yUUirJqVTlNTSEsFhOjRnXlrLN6ceaZPRk6NBez+RgpK1sZO1Ul9kd7PCqsiorUMUKoADab1Wcp4yOjE1mtu4d0R64uZojFVJVcVaXmOaenq8VLXnklvnranpjNaj61MU/aqKZTU/U+0tqx77gLY4D1FeuJyRj9srIAEzvq/FT4KxiUMwCLqQ5wtd46lkT9DQSCKFFixDBjblc5H2jTdS21xIiRQQYSiR8/jTRSi1quKY000kk/4DnKh2pvf8tPPqnnxhu3sn696udNT7eQm2vDbIYVK5qJxSA52cz48apinjDBS69ejoNq0pZSsmZNOR9+uJkPP9zMggVbueyyQTzxxLl89tk2Zs36mNLSRjZsqAIgJcXJGWf04Mwze3LmmcdWk/aujMA1QtoIbJ9PBVo4rKYuWSyq2VqI+OIlRtM3xEd17xrUHd0fbWhuViFdVaWatrt3V69pzRr46qs9/4wQajS3zxefipWWpoJaV9PaseK4DOOyxjKK64sZktMLsylIcyiJtRXrKUgpIN1lRs03Tjtiz3ckhAi19R0fatO1wY+fIopw4SKVVBpppI46TJhIJfV7DeX9/S23bw/y4Yd1fPhhPQsWNLBx42CiUZg5czufftpIaWmIykpV4nXtamsNZh9nnJFMVtbBNbkHgxEaG4Okp7vYuLGKE054BCklXq+D3r3TMJtNbNtWR2mpWsWjR49UzjyzJxMmFHLGGYWkpnbM2IKDYcwhNnatSuyPBlVBOp3tFysxQnrX/ujEgDa+PlZCWk2tiwe1MYBs8GD1JmXTJvVmZFc2mwplI5yNW1qaXsNbO7qOyzAOhANUNlWS7UnDavYjpYfVZZuxW+z0zshDbTCVAkep6fZASCRRooQJI5GYMWPF2nYfHFjTNahKuYYaqqgiQgQ3blJIoYEG6qn/XkP5YP6WxsAsgF/+sognnigjHJZYrYKCAjsOh6C4OERtrSrtBg5MYsIEL2ec4eXUUz0kJx/ctVdVNfGf/2xpq5y3b6/j3/++nN6903nppRW8//5G1q+vpLk5jBCCIUO6MH58IRMm9GTMmG5tzeHHusQR2saKYH6/qjQT/9kba3ZLqeZEG9tT7nrM3irpY0lNjWrGN3bEqqpSS4KedJJqEi8u3r0J3NhoIzGg09P1dCzt+3FchnGcBKoBMyX1zZQ2ljKoy0Cs5jrUiOr4/NSD3ULRbDYzadIkTjjhBK6++mo++ugjxo8fD8Cbb77JBRdc0G4LxUR/+9vfmD17NgAzZ87kqquuSrhi1Xd8xx138ME7H2C32elR2IOnn38aj09NCrVixYIFgWDGjBksXLgQb+tefddccw0333wzoPqla6mlkkoEgt705vTTT+fOP9xJt2Hd2kI5jbQj1qd8OH/LpqYoixY18PHH9Xz8cQPp6Rbef78v33zTxC23FFFREWbbtgDhsAqDoUPdjBuXzLhxXsaM8eByHXhCSCnZsqWGnBwPSUk2fvvbhdx99zwACgtTychw0dAQZOPGKiKRGDabmVNO6cr48YWMH1/I8OF5HTaF6lAlhrRxS09XwVNTEx/tbDLFm7stlniT+J5CetegPpZCOhqNX8+GDbBunQrqmho1UM5kUmFsfG8wmr3T09UgsrS0eFC7XMdGS4H2w3PchrGUEn/Ij8tmwSQEzaEYayvW0tXXlUy3DYigmqrVv6xD2UIR4IUXXuDhhx/m5JNPZs6cOQBcfPHFbNiwgZkzZ+4WxsYWisuWLUMIwdChQ1m+/P/ZO/P4qOpz/7/P7FsyM9k3EvaEhIQAYYsgoAKKqKi0RbEV9F77c+nt9Va73UrVqtVWW221tdpbtNal4ooLiwsgm+wQtrCEhCUbSSbJJJPJrOf3xzdzZhICBGUJMO/zOq8sczJ75nOe5/s8z2czdru903FLli3hyiuuRKvR8uDPHgTgyaeexIePIEElnX3n3DuZMWNGt6IfIlQsZsTIpEmTePDpBykuKsaFS0lf27GTQMK3FuUzWoznl9FoJHy+IMOHl7Brl1hvtlhUpKXpUKsl9u9vV44bM0aI86RJVsaNs2Ay9VwZAoEgW7dW8fnnZXz++QHWrDkEwNGjP2Pjxkr+/vdNbNtWTVlZQ8d90DF+fF8mT+7P5Mn9GT48FY2mFynRaRLZHx1ak/b7xWhNSRJp4bY2IW6hqDEYFILXXftV15R3b0l3h2hvF5FzUpI4yVi6VAh2c3O4EC4k5JFr7hqNiJxDg00io+ru/LejRAnxjS0UL3RaPC3sq9/HBvFRIQAAIABJREFUwPiB2Iw2jFoZg8ZAY1sjSZYswAn4IKL153QtFENMmDCBVatW4fP58Hg8HDhwgMLCwm6PDVkohqZuTZkyhSVLlnDLLbd0Ou7qqVcr3xeNLeK9d97Dhw8tWmXutQePUvjVlbvvvpuNGzfidruZNWsWjzzyCCBS2G7cHA4c5tE7H2X3pt3IkswNd9zAD+7/Ac1lzTx878M01DVgMpl4+eWXycnJOeVzcDYIVVZrtSp27iykqsrLl182s3KlkxUrnNx3XzJ33JHEokUO5s8/yuHDHh5/vJLHHqtEq5UYPdrCxImxTJwYS3FxDBbLicVSrVZRVJRBUVEGP//5RNrbfZSW1hEfb+bqqwfzyCNfUFbWgE6nZuDAeMxmLXv2HGPJEjEFNjZWz+WX92Py5P5MmtSPYcNSe12l9snQaMRuilgmD61Hg/i9JIWFOhgUIpuVJcSqvj4sWn6/SAmfLN19vtekI6epSRJcfbXYQ2NC6+vF4xgwQGQM3nxTnJD4/eF16927O1+nXt95Clkooo4OOIlyMnrtW+PTT7u3SDxdZGJo9WSzQa2lb4bM9OltxJviqHRW4fVL6DQgxmOGxfibWCiCSHFfddVVLF26lObmZq6//nrKy8u7PbarhWJGRgaV3VWeRPCvf/yLWd+bhYyMB49S0BUkiIzMTx/8KY8/9jgSEq+99hr5+fk8/vjjxMXFEQgEuPLKKykpKaGgoAAVKjLJpH5bPTWVNSzcuRAjRoxNRmRk7rzrTua/OJ+CQQWUry/nnnvu4csvv+zJU37WSUvTcdttidx2WyKAMnZzwAADkgRHjogGXJNJRUqKFqfTz1NPVfLEE5VoNBIjR5oVcb7sshis1hP/GxgMWgoLw44ky5bdwerVFXz55UG+/LKMTZuquP76HL7++m5WrDjIn/+8jq1bq/j4Y+FbaLMZGD++LxMn9mPixH4XZOQcKZIWi9ghnO4OBsNtVpJ0fOW2wSCiSJ9PCFwg0L1IdxXokFvW+RDp7saExsXBvfeK+9/UBA0NQoyNRmFaU1cHCxeKx1ZdLfaumM1hC8vExHDaO2phGaXXivGZQkJCo9LgD/oRhVouEswWKp3Q6G4iOcaAsFW0EEpVf1MLRRBCHrJQfOaZZ3jiiSe6Pa675YGTte88/vjjaDQafjDnBwBKL7KMrKwdP/n7J7l+1vWAKPKSkXn77bd56aWX8Pv9VFdXs3v3bgoKCgDhwTyi/whqDtbw3I+eY/y147lt6m24W91sX7udB7/zIAHEp6rf48eN+6yO2fymhOZfjx4dw759w6ms9LBihZNVq1pYtcrJhx/mkJio5de/PsLChQ1UVnp55pkqfve7KiRJFIRNmBDLhAkxTJgQS2rqiQu0YmL0ig0kQFOTm8ZGNykpMUyZMog5c94mGJTR69X06xeH0aihpKRaEWeLRUdxcRYTJ/bj8sv7MmpUBnr9hflvGBLgSJKTxddAILzGrFaLiFqWhYBFolYLsVarhYi5XJ3T3SELy64ifT6HmYQ8oePjYfDg8O+Tk8XPDod4nA0NYrhJqAe6qgrWrROP8fDh46/Tbhdr1KFistBtRNuyLg167afAaWrgSal3uahorCA3KRfQoFX7MGqNONwOkmP6IyJjL5E+x2fCQnFwxH/q+vXr+eEPfwjAo48+SkZGRic3qEgLxa68+uqrfPzxx3zxxReKYN817y62bt1KWloan376qZKi1qNXRmseKD/A008/zYaNG4izxzF37lzau4x2stvtbN++naVLl/LKC6+w8u2VPPPsM1hsFt7f9j527AQI4MBBGWVYsJBIIuZe0J99ItLT9cyZk8icOYmdfj9ihJlVq5xs2eIiEBCRSHKylvh4DQsWHOP550UqZsAAfYc4i8h58OAT9znbbEZsNnGCEh9voqHhV6xefYgVKw6yYkU5W7dW8fLLN3LNNYN5552d/OMfmyktPcayZfsBMBg0jB6dwYQJfZkwoS/jxmUSG3vhz4FUq8NCG0KShDNUSKRDX00mITg+Hxw5Ev77UJ+0JB1vTwknFunzGWHqdEJMU1KOvywvD/LzwxF1yISjf3+R2j56FA4eFHskIYesuLhwJB0S6tCyQZQLn14rxmcSq0FUGDe3N2PS2YAWUmLiKXccxeMHvUZCCHJYjM+EhWIkY8aMYdu2bcrPDoeDX/7ylzQ2imEcXS0UQyxZsoSnnnqKlStXYopYyFuwYEGn40ItUR48SqTc6mzFaDais+o4WnuUxYsXHyf49fX16HQ6br75ZgYMGMDcuXOxxdro368/SxcuZfJ3JqOW1dSU1JA3LI9GGimnHBMmEkgghphTtlj1FkIC3dIS4OuvRdRcWenl//5vID5fkIkTd7FvXzter8w77zTwyit1ACQmaigujuGyy2K47LJYRo40o9d3/4lvsxmZMSOHGTPE+rrT2Y5KJWGx6MnMtLFr1zF8vgCSBJmZNmw2A01N7Tz55Fc8/vgKVCqJwsJUxo/PUgQ6OTmm29u6EFGphPB0V+SkVos11kix9vlEOtdiEYVjx46FRRqESLtcna8nNBa0q1if7zSwRiPmj2dkdH95ayvs2yeEOrIta+BAkfo/fPj49WmNRjw/oZR35Pq0sfclsaKchEtCjLVqLTmJOZh1oWiuFZtBpCIb3Y2kxJgQYhzOj50pC8UTERcXx0MPPcSoUaMAmD9//kktFKdMmQLA2LFjefHFF4+7vlA/sgoVPnxISIwcNpIRw0dQlFdE3/59GXvZWAIElIlfINau582bR7AjN/jb3/4WCYm3Xn+Lu+++m5cee4l2XztTZ09l7LCxJJJII43UU89hDqNHTwIJWLH2eBjJ+SYmRs2UKTamTAl7QGu1Km6/PYlly5pYvbqF1lbxfBQUmBgxwsyaNS18+KE4cdLrJYqKLIwfH0NxcQzjxsWQmNh99XlklHvDDbk0N89nw4YjfPVVBatWVbBu3WHKyh7AaNTy8MNfsGzZflpbvfztbxv505/WAaLFavz4vlx2WRaXXZZJTk7iWfFxPt+oVOH16BCRqzkajVhzDYl1KJ2dkhLuoQ5ZV3o8nW0sQRzTXTTdWyq8LRbRDx1J6DGqVGINeutWkfpuaBDDW/x+8bwcPgwlJZ3/VqcTa/WpqZ2j6bi4aMV3b+Sib23qnlYgwJ5jVciyTG7yQKAJiIWz7HJ0tpGRFROK0AjNroNEejpeMxIPHvQdmYMqqggSRI+eZppppx0tWuKJx44dNepz+FqeeWRZ5sCBdlavbsFmU3PjjfG0tQWwWjcSDMokJmqRJKir8ymFSoMGGRg3TohzcbGF3FwTavWpn9tAIKhUW//+91/xxz+uUSaBWSw6srJsDBwYz9q1h6mrEyGg3W6kuDizQ5yzGDUqA6Ox9wyuOVeE1qVDUW9LizCf6DqbOyFBiFpbm7is6+Wh4rGu6e7eYLJxMgIB0YIVGk6yfTts2iSE2uUKn8hYLMen+PV6sUYdas2KFOroRLKzyyXbZxxJTUsNHr+HLHsmICnjMocmD8WgbUUkCmynuJYLDw8eggSV3uFvI8oyMjXUKDOwLVgwY6aVVly4UKMmjjjq9tQxdMjQs/mwzileb5Avvmhm7doW1qxpYf36Vtragtx7bzJZWXqWL2/mq69acLlEKBMTo2bsWAvjxsUwdqyFMWMsPXKmkmWZiopGVq8+xJo1h2ht9fKvf30XWZYZN+6v1Na2YjLpaGpyU1UlRFujUTFiRBrFxVmMG9eH4uIsMjKsZ/X56M2EbCdDu81Gx8lTeBQoiN+p1SKd2zXaDtFVpENfe3tiIlTx3dQk2rK8XliyRAw8iUzrh7y0IzEYhFAnJoqitNBadVSozwxRMQaONB3hWOsxClIL0Kq1eANudlTvJi02jdTYGKANYat4cWXvIyuvQwIM306U/fhxdGx+/CSTjBkz9dTjxEntnlqWD1nOXOYykIFn8+GdF/x+me3bXaSm6khL07F4cSPTp4uKabUa4uI0HdXDfuXDLjvbwNixMYpA5+WZTsud6vHHl/Pll2WsX38Ul0u0b02e3J8xY/qwbt1hvv76CB6PCPv69LEyblwmxcWZjBuXSWFhKjrdxfW+Pl1kWUTFIZEOiW+oAry6WqxBQzh1HYqMfb7O16VWd46ie1vK+2T4/SKD0NAgnoPsbFEB/sEH3bdjRRJqU4tG1N+cqBgjZlXvrN1Jemw6qbEJQCNVzlYa29zkpeQgRmYagYunYCZE11R1qMjr26avgwRpphkzZnToaKMNJ07K95Rz95C78eBhIhOZxzxGM/qCKfb6Jhw75mP9+hbWrWtl3boWNm5s5auv8mhuDvDii7WsWNFMa2uQtjYRfpnNKkaNElGz2GNISzv1J5rfH6CkpIa1aw/Tp4+VG27IxeFoIz7+MXQ6Nenpseh0aurr22hoEAunBoOGoqJ0xo3LVPaUlIvvff5t6OqG5fOJ9djUVCHkR44I8Q5FxaE+60i6umFdKCnvED6fEOpQe5ZaLaavORzw/vvHn5REYjCIfunIqWRxcdE16q5ExbiD0mOl+AI+hqYMRZIc+AJ+tlcfIS85D6PWh+g5Pv8+x2eLkCirOrbIgq5vI8ohaqmljjqO7TnGV0O+Qo+ej/kYBw5yyWUe85jGtHPmq3w+8fvljkhJYsGCYzz9dBV79riVSNlmU9O/v54dO9z4fOKXGRk6Ro8OC/TIkZaTTgwL0dbm5dNP97Ju3RHWrTvM5s2VeL0BfvvbaQwYEMeyZfv57LMDVFY68fvFyUDfvnbGjevD2LFCnIcNS7nko+euRE4fa2rqLNTBYHiIh9crqp9DBALHp38jhTlSqHvTHO+TEYqoQ0JdVyfWo5OTRZX7yZpOzGaxJySI4yOF+lKr+o6KcQcNrgbKG8sZnDCYWIMGaGVPbQ2xhnjSrcmAg67mERczoVGaIQEOGVR8G1F242brnq38cMgPCRLkZm4mn3xe4RUOcpBUUvk+32cWs4i5CLMQJ6O52c+mTS7Wr2/h8GEvL77Yn/b2IFOm7GbduhZsNg0+XxCnUwimSgV5eSZGj7YwapSF0aMtDB1qRKs9+aKlx+Nn69Yq+vWzk5wcw7//XcLs2W8BoNOpSUsT0bPT2U5NjajuMRg0jByZztixfRg7tg/jxmWSnn7prj2fjMioWKMRwlxbG57lHcJiEWLj9YYHmnRdl44cERop1udr+tg3IRgMp74djnAPdWqq6CGvqjq+LSuEySQi6tjY44X6YjTkiIpxB8FgkENNh0i2JGPSGYAGmtvbOdLUTF5yHpLUjJhVHQ8XSKvOt+FEldbQOVLWdGw9FeU9e/YQNySO13mdeOK5ndvx4uUFXmALW9jEJkyYuJmb+T7fpw99Tn2lFzGLFzeyYoWTDRta2bSpldbWILm5Rm6+OY6NG1189ZVTSW/r9TBihBDmkEgPGGBQJpGdiKNHm/n668Odoufy8gcJBII8+eRKPv+8DK/XT2VlCz6fUJqMDKsizmPHZjJiRNolWbl9Oshy5zS3Xi/EuKrqeCE2m4UYh9LjJ4qmu4r1hRJNh/B6RZrf4RARdW2t+L5fP/EcVVZ2v14dGnhis4WLyux28bvY2N5fSNcdUTE+IS0EZTfbqo6QkzgEs97MbbfN4rXXXgVM581CMcSDDz7IRx99hE6nY8CAASxYsACb7fiK75NZKHbHpEmTePrppykqEu+JSFEO+SWHfufHrzhE9VSUu3stv+AL7uM+0khjClOopZbP+ZwgQa7kSm7ndkYw4qJeV+4JgYBMaakblyvA6NExBAIydvtGWlqEQBqNKgwGidbWoJLetlrVjBxpZtQoC0VFFoqKzGRl6U86YtXr9Stp6eeeW8Mf/7iGQ4eEf6JGo6JPHyujR2ewfv1RKioald8XFqYyZkxYoAcMiDvp7UQRhKLpyEpvi0WItcslBCqEJIVbrkJ/E0l30fSFtDbdFacTDhwQqf5QH7XTCTk54ROZru1ZkiREOuQ9HVn1bbeL56I3EhXjLrh9bvxBPzF6E/5AgG3VJaTEpJCTnsOgQf1Zu/ZTjMZ0Fi9ecl4tFJctW8YVV1yBRqPhZz/7GQBPPfXUcY9n7ty5p7RQjKSrGIcIrR9LSEoVdkh8Q6IMKIJ9IuHs7rUMEGAFK1jAAjazGRMmruZqrFh5j/doppl88rmd25nK1EtiXbmneDxBSkra2LixlY0bRfQ8Z04i11xjY/nyZn71qyNoNBItLQEl+kpI0CjCPHKkhZEjzWRk6E4qnNXVTtavP8rXXx/G4XDz0ks3AjB+/N/YubOWpCQzgUCQqqoW2ttFPjY+3qSI85gxfRg9OkMZERqlZ0RG05F7YqIQlebm8FzvyAlksnx8tB05gaxrNH0hCXVImiRJTCTbs0cIdX29eD48HjELvKVFpMS7FtPp9eHWrJBAh76ez/T3JWuheCIqGisIBAPkJeehUaux6q00uMS7/ZprruaTT5Yya9ac826hOHXqVOX7sWPH8s4775zW4zyRhWKIQCDAnXfeqZwU3HHHHdx///2UlZVxz733UFdXh9Fk5PmXn2dozlC0aPHhU+ZfhyLlnkzfUqPmyo5tF7v4J/9kBzv4gA+4j/t4hVdYxCIe4AFSSOFWbmUWs7BjP+V1X+zo9aLyetSocC2DLMtIkoTNpmHTJhebNrXidIY/kfLyjFRVeXniiSblAzsxUaMIs9gt9OkTFujU1Fhmzsxl5szcTrf/H/9RxIoV5Xz99WH27xdRcnFxJrffPoJ16w7z2WcH+PTTvcrxQ4YkdhLooUOTLzi3qnNJqAr7RO1BJpMQ4a5inZUVNuBobRXHyLLwae46gSxkutHd3hvT3pFi2d28b683nAnYvx927RIp8MZGkWnwiu4/ysogYhIxIB6vxSLctiJF2m4XUfb5sLrs1WL8j38c/7uhQ2H0aPFE/+tfx18+fLjYXS749787X3bHHeHvE0wJHGo6hMvrwqI30i/ORkWjePVmz57Do48+xIwZ1/QqC8V//OMffO973zvh5Q8++KCS9j6VhWKIbdu2UVlZyc6dOwFoahJpyrvuuosXX3yRQYMGsXb9Wu6/534+/fJTVKgwYFCmfIVEWY0aDRrUPaxCzyOPp3gKHz5UqJCQ+Cf/JIkk5jCH/eznD/yBv/AXruM6buM2BjP41Fd8CRES0KwsPf/61yAAnE4/W7e62LzZxY03xtGvn4GXXqrhhz8U70OXK8jatS0sXdqkRB4JCRpGjDB37BZGjDDTv3/nFPfcuSOZO3ckAA5HGxs2HMVg0DBpUn9mzy7AZvsNACaTlpQUCz5fkA8+2M0rr2xRfl9UlM7o0X0YMyaDMWP6kJFhjaa3e0hINCOJrPS2WITARAq1SgXp6eEqaK9XRI9+//HzvCOnkHXde+u6bOSJy6BBYg8RDIqTk9hY8fPGjVBaKtapnU7x/LhcIsrev//4qWxGoygqS0uDadPOTcV3rxbjs0mcKY4jzUeoc9Vh0fdFrZJIibESlIPk5xdQUXGUN99cyPTp007res+2heKcOXNOeMzvf//749LUJ7NQBOjfvz8HDx7kRz/6Eddeey1Tp06ltbWVtWvX8p3vfEc5zuMJj8MEkcaWkNCjV9aVQy1Tp5NeDh2rRs3P+Bmv8iqv8zo2bMxiFu20s4hFLGQhYxnLD/gBE5l4wczBPtfExmqYONHKxInhKuhbbkkkO9vIli1CpLdscVFa6uaTT3KoqPDwr3/VsW5dC5991qwIdGysihEjLAwfLkR6+HAz2dlGNBqJuDgTV18dPjGyWHTs3v1j1q8/yvr1R1i//gglJTX84Q/TufbaHD75pJQ//3kdFRVNrF17WGmtSkmJUYR59OgMRo3KuCgcq84VkR8XJpPYQ4TWpzUaIagejzje5wunc7VaERH6fCLVGxqE0vWjKTTT+0Jan1apwkIMMGqU2EO43SJzEB8vhPuzz6CiQrSvtbWJy4NBIdgzZpyb+9yrxTgyku2KTnfyy83mk1+uVqmJM8bhcDvoE+yDRmXCog8iAS2eFq6//gYeeOARVqz4iIaGk3S6d+FcWSjOmzevk4Vid5SXl/P000+zceNG7Hb7KS0UX3jhBd5++22effZZbDZbJ5eprgQJKlGxChU6dMjI+PDhwYMPHw4c2LD1SDh16LiRG5nJTDaykX/yT97jPV7ndX7JL3mTN3mbt7mHe8gkkznM4SZuwnKJtKB9G2Ji1McJtMsVwGwOZzH+9rdadu1qUz6ofT5oawvw17/W0N4uPp0NBomCAiHMYjeRn2/CaFSTk5NETk4St98unA7cbh+BQBCLRc/Ikel4PAEOHxZZF7VaIjU1hvz8FHbvPsaHH+4BxMloTk4Co0cLcR49OoOCgmjv8zehq9e0zSZ2ECLj94uvIXM5j0ektiPXn3W6sDGH2y2O6SrUIbHXajt/35uFGkSkG4p2VSoR/YYIBMInJ4mJ3f/92eCSfpcnWhJpaGugzdtGrMGCLLuQJDGz+o477sRqNZKfP4gVK3ac1vWeDwvF7nA6nZjNZqxWK7W1tT22UIyNjaVfv34sXLiQ73znO8iyTElJCcOGDVP+To0aI0YlVe3F2ymFDXCIQ1RSSSKJJJCApgdvNwmJ0R1bDTUkk4yERA012LEzmcnsYQ+/5bc8x3PMZCa3cisDGHDK644SJlKI77knhXvuSaG9PcjOnW1s3eqirs7HL3+Zgd8vM2bMDrZsceHzyezeLS4PVXGr1ZCTY6SwUAh0YaGZYcNMJCSEsyPFxVkcOvRTampa2LjxKBs2iAj6r3+9gawsO08+uZJHH/2C5GQLHk+A99/fxauvivS2Tqdm+PA0JXIeNSqdwYMTLkrXqnNFqBI7ktBI0Mi53iqViLZlWbhCdRVivV6IrtcrhPxkQt1VsHuzUKvV4ROXc8klLcYmrYlhacPQqMTTIEnCYlHGQ1xSBj/+8QNAPWIqV885HxaK3TFs2DCGDx9OXl4e/fv357LLLjvumO4sFAFef/117r77bh577DF8Ph+zZ8/uJMYghFOHDi1apdI6NNlLjZpMMqmnnmqqqaGGeOJJJBFDD52xUghXbIxgBNvYxlu8hQ0bM5mJCxcLWcgbvME4xjGHOUxiUo/XraN0xmBQdVRfh7MNGo3E668PYvPmVrZudbFtmxDjYcNM3HdfClu3unjxxVree8/B66+HR1Clp2spLDRH7Cb697dw3XVDuO66zlX248dncccdRWzceJRt26rxeoXf86uvzmLHjlqWLt3PSy9t5M9/FpaSsbEi2g6J86hRGWRm2qLrz2eA7vymJUkUikUKtd8vBN1kCnstRxJqzQoEwvO+I+kq1JE/X6ov4yXb2tSVoBxEJUkEgi52VO/HarTRL64f0AK4uZhHZJ5pggTZsWcH0hCJeOKJIYYmmnDgQEYmlliSSMKC5bR6imVkNrKR13iNL/mS7/E97uM+3uZt3uItaqklnXRu4RZu4qZoFfZZQpZlXK4gFouaQEBm9ux9bNni4uBBj3JMdrYBrVZi9263kvo0m0WaW0TPIoLOzzd1itI9Hj8lJTXs21fPnDmiE+HqqxewdOl+ABISTFitBny+ANXVrcpwksREsyLORUUZFBWlR2dvn0NCQh1pxBGaPuZ2dx7oETLgUKuFWHfXmtVd2luj6b3FZKdDtM/4BMiyzN66vRi1RrLsWQAcbjpMXWsd+Sn56DQqLrURmWeCXXt2YRpiogmxTmjDRiKJtNBCPfX48WPAQCKJigfy6VBJJWrUpJDCJjYxn/mMYATllLOFLejRcy3Xciu3kkfe2XiIUbrgdPopKWlj+/Y2cnONTJ5sZefONvLztwPig9RoVOH1ykqaG2DwYIMizgUFJgoKzGRmhtutnM52tmypYuPGox17Jf362Vm8eC4lJTXcddf7OJ0e2tq81Na2KunS9PRYiorSO/YMRo5MIzEx+j98rumuh9rvDzs9tbR0nusdacRxsmKyrmJ9oQj1NxZjSZL6AP8EUoAg8JIsy891OSYHWACMAP5XluWnT3WHeosYA5Q7yml0NzIsdRhqlRpfoJUWTz1tPg0Z1gygGfByqYzIPBOEXksvXuqoo4EGsslGjx4PHkWU3bhRoyaeeBJI6FSx3VM2spGneZoSSjBh4nIuJ0iQVazCjZsCCriFW7iaq3ucIo9yZggEZPbudbN9exslJa6Or2089FA6KSk6PvzQwRtv1KNSgdsd/iyKjVUxbJhZEeeCAhNDh5oU0wyPx49er0GWZWbNeoO1aw9TUyPMilUqiYkT+5KebmXTpkpKS+uU683MtDFyZBojRqQxcmQ6I0emk5QUFejzSSAgisMihdrnE3OtJUm0ZTmd4thQVN2dUEe2Z3UtLOtNA0++jRinAqmyLG+RJCkG2AzMlGV5d8QxSUAWMBNovNDEuMXTwt66vfS19yXBnIBIS7dQ1tBAX/tg1Kog0IiIjE0nva4ogq6vZWg9GaCMMty4SSABAwYaaVQiaCtWEkk87RQ2wA528AZv8CmfYsHCR3zEJ3zCW7zFQQ5ixcrN3Mz3+B6ZZJ65BxvltAkNLNm2zcWzz1azfbuL3bvdeL3i8+imm+KorfWxdatLmckN0L+/joICM/n5JuXrwIEG1GqJyspmNm+uYvPmSgoLU7nxxjyqqpykpz+JSiWRlGTBaNTQ2uqhri48ESMjw8rIkUKcR4wQQp2aGnvcfY5yfggViEUKtd8PffqIrw6HSIWfTKhDA0+6q/4+14YcZyxNLUnSh8Dzsix/1s1lDwOtF5oYy7LMrtpdqCU1OUk5SBIEg/W4/R5aPVqSY5IRYhxARMe95BSrF3Oy17KZZo5xjFZakZCwY8eGDRcuGmhQUtgJJBBH3GmnsBtp5AAHGMUoAgSYwxz60AcnTtawhgABJjCB2czmci7vUZV3lLOP3y+zf7+bkpI2rrvOjsmk5oknjvK//3tEOUbJtqLvAAAgAElEQVSjkdBowOORlQ9cvR6GDhXCnJ8vIuj8fBMpKVra2nx8/vkBNm+uZNOmSjZvruTYMRd/+9tMBg9O4JNPSnnnnV24XF7q6sKTMFJSYhRhDu3RIrHeQ+TAE7c7LNaRrlmpqeL7hgbx9URCDcevU0eK9plOf5+RcZiSJPUFhgPrv+GduAu4CyAzs/dEJpIkkWxJ5lDTIZztTqxGKyqVBbNOpt7ViCwnIUkmRLraTTQ6/nZYOzY3buqpx4EDPXrSSCOZZBpppIEGjnKUKqqII44EEjDSszE4duyMQlSpO3ESRxyfInqxxzCGeOJZz3ru5V5SSOHmji2V1LP2mKOcGo1GYsgQE0OGhP+/fvnLDO67L4UdO9qUvaysnffeG0xpaTs/+lE5a9e2snOnuCwUWQPEx2sUYR46tICrrx5Lbq4Bl6uNmBg9sbEG2tv9vPFGSSchFsVg6ZSXN7J48V7lwzsuzsiIEekUFqYyfHgqw4enMXhwAmp1dOnqXBN5ThTZMxwiGAy3cHWXBtfpxMAPn0+kwQOB7j2oATIzz814zB5HxpIkWYCVwOOyLL93gmMe5gKMjEFUUze0NRBvikclqQCZQLAOj99Lu99EnMkONAF+omvHp+Z0XssAoiJWjZpGGjnKUUV8m2mmkUZkZCxYSCABG7bTTmFXUcXbvM27vEs99bzACwQI8DZvs4Y1SEhczuV8l+9yOZdH26MuENaubWHNGqci1Hv2uElJ0fE//5PKzp1tfPCBA4fD38lEICNDS36+OUKoTdhsPnbvrmHr1iq2bq3m5ZdvJC7OxPz5n/Gb3yzHZjMQE6PH5wtQX9+mTBEzmbQUFKR0CHQaw4enMXRoctRmshcTMtcIzeMOjQoNCXVoGEpsrPjZaj2zaexvFRlLkqQF3gVeP5EQX+ioJBWJ5kQkSeK2227jtddeQyVZqW4uY1ROAcXjirn55pk899yzgIrdu0vPiYXiww8/zMsvv0xixyiYJ554gunTp3c65t5772XNmjV4vV7Ky8vJzs5WrrOnLk7nk0jh06HDhIkaagARRWeRhQcPDTRQQQUaNCSQQDzx6DjBZP0upJHGf/Pf3MM9rGAFE5mIGjV72au0W21gAytYQSqpSrQc2escpfdRXBxDcXG4hSkQkKmr85GSIt4XarXEypVO9u1zK4IcCEBVlZcvvmhWImlJgoED9Qwdmk5e3kCWLWsjL09m5sw8bDYjW7dWsWVLFaWldej1GjZsuJuSkloWLNjMgQMNbNtWzYsvbui4TRU5OQkUFqZRWJiq7AkJ5nP75ETpllBbVYguRnlKhHyuzSJOeXOSWCT5P2CPLMt/OPt36fxiNpvZvG0zbW1tmEwmNq4qISElAX/Qz7x5/8G8ed8BvPTtO4rly5d3slDMz8/nzTffVMT4rbfeOm5QRgiHw8EjjzzSyULx+uuvP85CEeD+++/ngQceOOF9fuGFFwCoqKhgxowZJxxj6ff70ZwPO5LTwIyZAQxQxLeBBjx4yCGHZJKVfuWajs2KlXjiiSW2R9GyDh1TCTthGTCwhz04cJBCCldzNfXU8zzP8xf+wgQmcDM3M4lJUUvHCwC1WlKEGOCvf+0PCAvK0lI3u3a1YbVquPZaO253AKt1Iz6fWH8uK/Nw6JCXDz5wKOlKtRoGDTIzdOgwvvOdcQwapMVi8ZCXl8rw4emsXl3Bhg1HFTtJjUZF//5x9OsXx8qV5bz+evh/MT09lsLCVIYNS2XYsBSGDUtl4MD4aJq7l3G+HKx68sl8GfB9YIckSaF31i9BlKPKsvyiJEkpwCYgFghKkvTfQK4sy86zcJ/POmMmjuHf7/2bebfN45P3P+G735vJ+rWbOy41Ax7g+PT+2bBQ/LaMHz+eiRMnsmrVKm666SY2bdrErFmzmDlzJgAWi4XWDtfuJ598kvfee4/29nZmzZrF/Pnzz+h9OR1Ca8gppODDh4REgABHOUoMMWSRhRs3Dhw004wOHfEd2+mI5n/wH/yAH7CCFSxkIUtZyjSmsYxlvMu7vMd7rGQl8cRzPddzMzdHR29egOj1qo5e5nB0ajSqaWoa1SHSbnbubGPXLlE8Nm5cDCtXOvmv/6pg3752Dh708M47DuVvNZpqBg0ykJeXx09+MhK7vR2fr5na2nqsVj2//rU4Ic/MfIojR5qx242oVBIbNhxh8eJ9BIPi88Nk0jJ0aHIngS4oSIkaZlyCnFKMZVlezSnKh2VZrgEyztSdAvjv/y5n27a2Ux94GhQWmnj22X6nPO66m6/jr8/8ldk3z2bHjh3cPvdGtqzfgtvnxqg1AkaEGHf23TpbForPP/88//znPykqKuKZZ57pNno+GU6nk6+++gqA2267rdtjPv30Uw4fPsz69euRZZnp06ezdu1aiouLT+u2zjQqVErvsYyMHTsOHDTSiAEDSSQpa83VHZsVKwkkEEPMaUXLU5lKJZV48dKHPlzHdSxkIVdxFS5cvMZrLGABhRRyEzcxnemYiaYeL2RMJnWHbeTxvcZZWXqsVg27dgmR3rmzjUOHvNx7bzJWq4bVq50sWtTIO++ET8xVKjuDBhkoKdlLbq6RW2+9Are7ierqenbsqKGy0smddxZx991j2LKlkl//+guOHm1m585aXn45bEjTt6+dgoKUTns0ir646d05y/PEFWOv4FeHf8XLr77M9OnTkSQTKknC5W3AqM0gXE19/MnCmbZQvPvuu3nooYeQJImHHnqIn/zkJ/yjO6PnkzB79uxTHrNs2TIWL17M8OHDAWhtbWXfvn3nXYwj0aAhgwzSSKORRuqpp4oqsskmgQTcuGmggUYaO0XLccT1eG05nXTl+yBBiihiOcvx4SOXXDLIYD/7mc98fstvmcY0ZjKTUYyK2jpeZFitGn7wg862PS5XoGOSmJqNG1v5zW+OsmtXG+XlHqUwKClJy65donhMjHq0IEkW+vXLZto0HXq9hpISDRkZmWRl2dm1q5a2trAQz5iRQ0yMjq1bq/j441IlijYYNAwdmkx+fgr5+eJrQUFKdGjJRUKvFeOeRLBnC6vBypVXX8n8X85n5YqVNDocBGWI0avw+j3oNHpEssAD+CAiLXqmLRSTQ3YqwH/+538yo8NcsycWiiHM5nD0ptFoFFOIQCCAv6MpT5ZlfvWrX3HnnXf24Bk6v6hQKSnpdtqVqVrHOEYbbd1Gy7HEkkBCj9eWAQYxiOd4DgcOFrGId3mXlR1bOeW8zussYxkf8AFppHEDNzCTmdGBIhcxkXO0R42ysGhRDgBud4C9e9uVNHdsrIY//rGKn/70EH6/KAg6eNDDkSNePvtMxu9v7LiWApKTNQwdKmO3tyFJTq69djAzZ/Zn587DXHWVOPHWalVYrQYqK50cPOhgwYLNyv1ISjJ3CHRYpHNzkzCbe3YCGqV30GvF+HwiSRL3/fA+0pPSKcgvYOXKlUho0Gs0ONrqiNNkIMRYBbQCnf22zqSFYnV1Nampov/1/fffZ+jQoUDPLBS7o2/fvmzevJmbbrqJ999/n0BHiem0adN47LHHmD17NmazmaNHj2IwGJQCtd5K5HjLGGJop50qqpCQsGEjiyzaaaeBBg5yEA0aRch7OnozjjjmMpfbuZ1KKrFipZBCHudx7NgpphgHDl7kRf7KXxnJSG7gBq7hmqjf8iWC0ahWHKpC3H9/Gvfem8KBA+3s2eNmzx43e/e6+b//G0BFhYf77ivns8+aqa3109AAwaCeYDCRjz5q5P/9v81YrTL5+dOw292Ak5aWRo4cqefTT+fSr5+dP/xhNc8//zVqtYodO2pZtaoCr1f8P0uSRP/+doYOTe60Z2cnotVG2/Z6I1ExPgHZA7L5+QM/V36WJDXO9gDVLQ3oNaE1WyMiMvZ2+tszaaH405/+lG3btiFJEn379uVvf/vbt3pcP/zhD7nhhhv47LPPmDp1KvoOr7Tp06dTWlrK2LFjAYiJieGNN97o9WIcSVzHFhom0kijktpOIUUp9qrt2GKIIY44bNh6lGKWkMjoKI2Qkfk+3+cDPuALvkBGJp98BjCA7WxnPvN5gie4iquYwQyKKY5WY1+C6HQqcnNN5OZ2HhQ0eLCRBQsGsH17myLUpaVuWlsD/O53WezZ4+aZZ6rYtUtCpdLj91sBUWNy+eVHyc52kJBgZtCg/rS0NFFZ2aAI8d//fiNVVS0sWbKP1asPsWhRqbIsptWqyc5OIC8viaFDk8nLEyLdv39cdD36PHPJuzadDFmWqXJWoVVrSbIk4Q/62VmzE6PGyODEwR2N4KEKyziiYzLD9IbXMtixadDQQgsHOEAMMVix4sePAwdevKhQYcdOPPGYMJ32QJFqqvmYj/mAD5jHPG7mZtawhtd4jW1sU6aAXcM1XMd1FFBw2rcR5dLjrbfqWbeuRan2rqz0Mniwgeuus7N3bzvLlzfjcoXmdstAG3FxbQwfPoAhQ0zs2bOZ5cu3KmvOarVEfLyZoqI0du+uo6KiUbktg0HDkCFJnUQ6Ly+JrCwbqgvFDukCIWqh+A3ZV7ePNl8b+Sn5qFVq6l3HABd6jZUYfRzQDjiBGOjhuMZLgd72WnrxKn3LPnyoUWPHrngtN9GEjIwBgxJhn24kKyMTJIgaNW/wBr/hN8QTTz75tNPOZjbjw0cmmVzHdcxgBn3pe3YecJSLjtbWAA6Hn8xMkc267bb9bNvmYt++dsWSsl8/PQkJGkpL3bS0BBHz9F1ota3ExbmxWmVuvXUS2dlGfve7hWzdegSNRoXNZkCjUeF2+2lubldu02TSKiKdm5sUFekzQFSMvyGtnlZK60pJj00nNTYVWQ7iD9biDwbRa1JRSRLCRCKIiI6jb1Dona8lCMFsoYUGGmijjVxykZBw4sSNm2aacSHmFMcSSxxxWLGedqW0Fy8rWclHfMQKVigiPI95LGUp61mvpLav5Vqu5mqSST71FUeJ0oVAQKaiwkNpqZv4eA1jx8bQ2Ohj0KBtNDSE2y8lCex2NY2NgY6hJk6gGbPZhV7vwut10qdPPI88cj3Qyv33v4vXG0CtlnC5fLS0eJTrMpm05OQkkpub1GmPprtPTVSMvwX76/fT6mklPzUfjUpDm9eBSefH2S4Ta0hGrBk3IQaCRPtOofe+lpHIyEhIyMjsYhc+fFixYsGCDx+NNCpRtA0bccRhxnzaKWYnTj7jM/axj1/wCwAe5VFqqeUIR9jPfiQkiijiWq5lKlOxc3q95FGidIfD4WPv3nb27hXFY1Om2Bg3zsKiRY3Mnr1fOS40dSoQCCICChlJKsVkaiUYbMHtFieoxcUDuOmmXA4fruedd3bS2urF6QyLtF6vITs7gSFDhFAPGSJEetCgeHS6aIkSRMX4W9HmbWP3sd2kxqSSbk1HlmXcvmp0aglIQKPWIsTYR9REQtBbX8sTEZrm5cCBHz9q1KSTjhatUvgVJIgOnZLG7mk1dlcCBLiVWymhBBUq8snHjp1yyjnEIdSoKaaY6UznSq4khphTX2mUKKdBICBz6JCHvXvd7NvX3vHVzf33p2Kzafj3vxv485/FfHiNRsLv9wEtiJrfGMxmHz7fKrzesNuVSiUxfnxfYmJ07NxZy6FDTcplarWKAQPiOgQ6LNTZ2QlYLN/s/+hCJSrG35Kalhpi9bGYdKIqst3Xil7jotUbJEafipjG5QAMiKmglza9+bU8GTIyTpw4cCgzr9tpx4EDNWpaOjYQc7TtHds38UQuo4xP+IRP+ZRDHOK/+C8mM5mP+IjFLKaaanTomMAEruEaJjMZU9S+M8o5oKbGy6pVLezbFxbr0lI3Tz+dhdsdZOHCBlavbkGlChAItCBaPFuAJBISkklJaWXnzrDlvcWix2jUYDZrOXrUqbheAfTpY2XIECHSOTmJDBmSyJAhSSQmmi9K/+ioGJ8F6lqPUN3SwOCEHAxaA+LN6Eb0HV/aDfcX2mt5Muqp5wjC4N6MmVhikZFpool2RLHLt1lflpHZwx4SSCCJJJaxjAd4gAIKMGJkD3tooAEDBiYxiWu4hsu5vFOPdZQo5wJZlpEkifXrW1i0qJH9+9sVwXa7g/zmN304fNjD0qUOKisbI4RaiLXNNpwhQ1Lx+Q6yadMG9HoNJpMOlQpaWjxKexYI/+jsbCHQkXv//nY0mgu3V/pbWSheSkRaKIJwOkpNTWXMmDHMvHEmf3j2D2hUGkr3lDI4ezC+oI9JV01i3PCxzJt3B59//i5XXjkRiOP99z+IWiheBIQmd4VmYldTjQoVQxmKBw8OHDTRRAUVqFBhxUoccT2ejS0hkUuu8vNABnIrt7KEJdRSix49IxlJX/qynOUsYQkmTFzJlVzDNVzGZT0e9xklyrchFKmOGRPDmDHh5RNZlqmp8ZGaKt6Hr70Wy/vvx7J3r5uysnY8HhmDQeKmmxIoK/OwfXsMkIvH04rHExLrAEOHTiElRUVV1X4OH66itLSekpJqXK7wuFCtVs3AgXFkZyeSnZ1ATk6i8n1c3IWbOYpGxl2wWCwMGjSItWvXYjQaWbx4Mb/4xS/IyMhg0aJF7KrdhSRJzBg7g02bNuHXezHrArz5r494/rkXGT16JH//+1OAme997w727t3brSg6HA6Kioo6WShu3rz5OBOIhx9+GIvFclILxRAhC8WdO3d2e/m5tFDsDa/l2UBGxt2xxRMPwH72o0WLAQMePDTTTIAAGjTYsGHH/o0Kv4IE2cIWlrKUPezhNV4jQICneIp97KOUUpw4iSGGyUxmGtO4jMu+8Vp2lChng2BQpqrKS02Nj6IiMZHu5z8/xMcfN1JW1k57u9Agq9VPYaGdsjIPR49WAEcQIh32ABg1qhit1s2RI0doamrF5fIofdQAiYnm4wQ6OzuRfv3svWbyWDQyPg2uueYaPvnkE2bNmsWbb77JLbfcwqpVq1CpVGTaMtnfsJ+ALNIpieZkPIEa1JKX8ePHs3r1anw+FR7PMQ4c2B+1ULzIkJAwdWwgxFmPniaaaKRRqbzWo6eNNhpooJ56tGiV9WUjxh4JswoVRR1bCDVq1rOe/exHj57hDEeNmuUsZxGLMGFiMpOZylQmMAFjtPc9ynlGpZLIyNCTkRE+SXzyySyefDILWZaprvaxf78bn0/mqqvEaOGrrjKwdm0Gbne4Vzo21kN9fRqHDnkIBhsRLaVhIdbp9KSk9KGmxsmWLSW0tYUnI4Z8pkPinJ2dwODBCWRnJ5CUZOkVa9O9WownTdp13O+++9147rknhba2ANOnlx53+dy5icydm0R9vY9Zs/Z1umzFirwe3e7s2bN59NFHmTFjBiUlJdxxxx2sWrUKAKvRit1oJxAM4PF7UKvU+Dx61CoVQTwdFopraG6u4vrrp1FeXt3tbUQtFC8OJCQyySSDDFpoUVLZ6aTTj35KGtuFi2Mdmx69Isynu+4rIfE+7ysR8zKWUUcdt3ALV3AFS1jC53zOJ3yCESMTmcgUpjCRiVG7xyi9DkmSSEvTkZbWeZnl88/zkGWZY8d8HDjQTlmZh4QEDdOn23G7A2RmaqivH4oYuuQCWtFoYN++fng8MrCSyDHFsqyipsaD03mMJUv24/OF16ZjY/UMHpxw3D5oUPw59ZXu1WJ8vigoKKCiooI333zzuPVZgD5WIaLVzmrSU9KJNcTT5vWhlmS++91ZPP/8X2hudvDMMw/xxBN/6fY2ohaKFxehtWIrVgIElMjXiZMaatChI5FEVKhopZWajs2IEVvH1lNhVqNmVMf2C37BNrYRRxz96IcJEx/yIXnkYcTIBjawhCXo0FFMMVOYwmQmR/uYo/R6JEkiOVlHcrKOyy4L/95oVFNXN4rmZj9lZe0cPOihrKydwkIzU6ZY2by5ldGjA4j0tliPDgRcuN1mnM5BiAFNSwn50btcQXbsqGfXrnra2z1EfjSnpMRQUvIjEhPPvuFLrxbjk0WyJpP6pJcnJGh7HAl3x/XXX88DDzzAihUraGho6HSZTqNDo9KQbhXet5IkoZJMyED+iP4RForZiN7jYNRC8RJCTXhtyo4dFSoaaaSOOgD06Mkll2aaaaJJsXn8psI8kpHKz0kkMZe5LGMZu9iFChU55JBHHmtZywpWKGJ+VccWnfwV5ULEatUwYoSFESM6C+WoUTG0txdz6JCHgweFWB882M4NN8SRn2/inXfq+c//HEMoog4EXAQCLiQpBVkeCDQDawCJpiYjweC5MXjp1WJ8PrnjjjuwWq3k5+d3Es0QKkmFVq1FlmUx01hrxO0LsLeukl8/+mtsMTZQhjW0Ry0UL1EiLRv9+JV2KD16kkjCi1cxp3Dh+lbCDJBBBj/hJ/wP/8M+9rGMZaxlLfOZjxYtz/Ec29hGJZU81rEVUKAIcz/On494lChnCr1exeDBRgYPPr5m4s47k7n++qmUlbVTXh4W63vvTcFu1/CXvxzlmWecgIvs7BwSEs5Np0JUjE9ARkYGP/7xj096TFAOUlpXikUnzsyMWjNqlYbsMYPJScpGPL1qwjaL4Rc1aqF46aFBQwLh51NGpp12ZYiIAQOJJJ4RYZaQyO7YfsSPlN8f4QjrWQ9AX/qSTDL11POHjm0AAxRhziMv6i4V5aJDkiSSkrQkJWkZN+746XZPPNGfu+5Kp7zcw/DhZtTqc/M/EG1t+pZUOCqob6snNykXk85Es7sJtaoFnVqHTpOEqPZzIOwVLx2bxQvxtTxf+PApzlGttJJKKimk0E479dTjwkVbR3uHAUMnYf4mYllFFZ/zOZ/xGZvZzDSm8SAP8iVf8iEfspvdBAmSQgpXdmxFFEX9mKNEOQNEJ3CdJfwBPztrd2LQGMhOzEaSJI61HiHJosfr16LT2AEPYh3i0jGSuBBfy96ADx8SEho0NNFEOeXo0BFDDCpUtNGmuErp0SvC3NN2qa6ExD6LLA5zmGlMw4aNAQzAh4+97MWDh1hiuZzLuYIrmMAELJz9gpYoUS5Gon3GZwmNWkOGNYOKxgoa2hpIMCcQZ0qlsa0Sm1EmKHtRSXpAjygY0BN92qOciMjo04KFPvShmWYcOJCR0aAhm2zaaKOJJmo7Ni1abNgU16meCnNCxwaQSCLP8ixf8AUrWEELLRgwcB/3UUUVy1nOx3yMFi2jGc3kji2NtLPyXESJcikRVYUzQLwpnnpXPXWuOuJN8WhUGtQqK75AG5LUiEqdBFgQ68ZOwM6lkq6O8s0JrTEnkECAAM0000orRoyYMOHBgwYNatR48VJPPXXUoUaNFSs2bEpE3ROMGJnWsXnxsolNfM7n3Mqt2LHzGq/xDu9gxUoFFUoB2BCGcAVXMIlJ5JJ72rO5o0SJEhXjM4IkSfSP649GpVF6hWMNNmpanJh1AXwBFyadBVFd7UT0vkVt8aL0HDVqxboxRMhhKkAAFSpiiMGAAR8+JZpWoSK2Y7Ni7bG7VKgvuZhwn3kssahRs5GNAKR3bF68/IW/8AIvkEgikzq2cYyLTgCLEqWHRMX4DKHTiErpQDCAy+si1hBLgjmNXbW7UEvN5CblolIZEJXVbkRldXSGcJRvTgYZpJFGK61Kz7IOHX3pS4AAVVQRIEALLTQhvGXNmJXhJKc7/euGji2Usv6SLwkS5E3exIGDR3mUGmr4hE9YyEL06BnDGCYzmUlMIoWUs/E0RIlyURAV4zPM0eaj1LvqGZI0BJPORF9bX8oby2jz1WLRJyPS1T5EhBwH9I7h5VEuTCIj3wwyCCIGunjwUE89INLPVqwAuHBR1bHp0SvCfDpGFmmkMadjC92eBQub2Uw99WjQkEMOZszsZz9f8RWP8Ag55DCxYyugoNNwlChRLnWiiztdkCSJ73//+8rPfr+fxMREZsyYwYIFCygsLKSwsBCdTkd+fj6FhYX8/Oc/55VXXkGSJEo3lqJRaTjoOMi7776LzWRj4+frMetU+AJNiLXiWABeffVFBg0axKBBg3j11Ve7vT8PP/ww6enpyu2eaNrW3Llz6devn3Lcn/70p5M+zkmTJtG1mj3KhY2EpAicCRNDGEIaaahQ0dCxpZNOHnmkkooWLXXUsZ/97GAHFVTgwIG/Y0xgTwitD+vQsYIVvM7rzGPe/2fvvOOjqtI//NxpyaQnpEFCSYAkpANBEAygNFFEpONaQP0p9sK6uiorYttVsawNURBQlmoFVAQEReklQBJKILSEQEJ6TyZzf3+cmUsCCQQIkITznM/9MDO3ztwh33nPec/7xYqVbWzjTu5kGct4nMexYuVzPudO7uQGbuAf/IPlLNeidonkWkZGxmfg7OxMYmIipaWlmM1mVq5cSUCAKHs5YcIEJkyYAIhKVmvWrNGKYsyePZuoqCgWL1rMu73fZf+p/Xz59ZfExMTg5uhFVnExvi4uWNUydIojOTmVvPLKW2zd+geK4kbXrl0ZOnRorSYQTz/9dL0sFN9++23pXyzRcLQ1P/ywYKGAAlxwQYcOK1aKKcYZZ0yYtO7sXERFOGectYi7vtOm9OjpYmvP8AzppGPGjBdeBBPMfvZjxkwQQejR8zu/s5Sl6NDRmc70pjc3cAOd6CSLjUiuOWRkXAt2C0VAs1CsD/Hx8WzevBmz3oyb4sbBgwfpFNkJvU6Pg6EFJRUVqGo+YGHFit8ZMKAfXl5mPD3NmoViQ/Pwww8TFxdHREQEL7/88lnrq6qqGD9+PJGRkURFRfHee+8BcPDgQW6++Wa6du1KfHw8e/ee7ZAlaToYMOCFlxbJeuCBDz5YsJBDDvnkY8RIRzrijz9VVJFBBvvYRxJJHOUoeeRRRdV5znSaAAK0hLM+9OFTPmUYwyiggN3spoAC3uM9HuRB7fEIRtCb3vyTf/ITP8moWXLN0Ggj46eeWkZCQu32gxdLbGxL3n9/yHm3O5eF4rlQFMVmobiCvLw8Bt86mKx0YQ7g7uhOen4+vi5WrFW5NgvFIMQtKCQwsNUlWyg+++yzvPbaawB89YqaAwwAACAASURBVNVXREVF8frrr+Pl5UVVVRX9+vVj165dREdHa/skJCSQnp5OYmIiAHl54o/fgw8+yPTpoht906ZNPPLII/z222/n/QwkTQO7J3MAAZRTrmVlu9haIYW44YYBA5VUkksu2QjDFBdctKi5vlXAzJi1LGsVlcMcZjObGcQgbuZm0knnGMdoS1v06FnFKr7ne3ToiCaaG7iBeOKJIEKONUuaJTIyroXzWSiei7Fjx7JgwQIWLlzIxAkT0ev0qKqKVbXS0i2QtPwi9mQep8JSgRg/drftWU5t/tYPP/wwBw8eJCEhgZYtWzJp0qQ6z/3222+TkJBAQkICUVFRACxatIguXbrQuXNnkpKSSE5OrrFPcHAwqampPP744/zyyy+4ublRVFTE+vXrGTVqFLGxsTz00ENkZDTsDyNJ48EBB3zw0bKdVVQccaSUUnLIoZBCHHDAz9bsmdp72Vsjaq7vWLOCQhBBjGGMJuTDGMZwhlNKKckkU0QRkUQykYlYsfIRHzGGMfSiF0/zNItZzHGOX7bPRCK50pw3MlYUpTUwF/BHGEHOUFX1gzO2UYAPgFsQJpLjVVXdfikXVp8I9nJyLgvFc3HddddVs1AMAeBE4QmWrl7Ky39/GVVVmfD0BIzuBlJ27ENkU7vZLBQ7nHW8S7FQPHToEO+88w5btmzB09OT8ePHU1ZWVmMbT09Pdu7cyYoVK/j4449ZtGgR77//Ph4eHjVcpiTXDgoKbWiDikoppRTYmgMOmgPVMY5hwKA5UdmjZmecccUVN9w0N6r6UH1O81GO8id/YsDAaEbzCI8QTzyeeOKII5vYxC+IIZ1ggulFL3rSk250w/kaKTcraX7Up5vaAkxSVXW7oiiuwDZFUVaqqlo9xBoMdLQt3YFPbf82Wc5noXgu3nzzTRwdT8/hNOgNBIQFsHbDWjzMHhSWFXIoI5lP3vqInJzjKIqZX3/9gzfffBEog2rzPy/FQrGgoABnZ2fc3d05efIkP//881meyadOncJkMjFixAjat2/P+PHjcXNzIygoiMWLFzNq1ChUVWXXrl3ExMRc0OcgadooKFp3dvU5whVUUE655jblgAMuuGDCRDHFnLA1PXpcbc0NN0zUz4quDW24kzu152WUMYIRrGc9yYg/Oy640Ic+5JPPQhbyFV9hxEgssfSkJ9dzPZFEyi5tSZPhvGKsqmoGkGF7XKgoyh4gAKguxrcDc1XhOrFRURQPRVFa2vZtktTHQrEuBg8eXON5C6cWOBmdOJx7mHBjOK6Orvh5t+Pvzz/Oddf1AnQ2C0UfoJAHHniEiRMfuWQLxZiYGDp37kxERATBwcH06tXrrG3S09OZMGECVquYL2r3VJ43bx4PP/wwr732GpWVlYwdO1aKsQQAV1wJJ1wbay60tXDCMWIkm2zyyceKlSKKtCQsBxw0cXbBpd7VwJxxZpKt5ZDDBjbwF38xnOHEEcd61vMCL+CFF8c5zge25oYb3emuiXMb2sgsbUmj5YJcmxRFaQf8AUSqqlpQ7fVlwL9VVf3T9nw18JyqqlvP2P9B4EGANm3adD1y5EiN4zdnp5+yyjKSM5MxG8yau9OR3EP4uzpgMhjRKS1sW+ba/vWkKRcEac73UnI2KqomdGmkkYVIXNShwwknDBg0cbYXCnHCSRNnZ5wvuqb1DnYwnelsYQullKKg0JKWRBPNTnaSIWIJAgigBz3obmu++DbAO5dILoxLdm1SFMUF+AZ4qroQ21fXsstZKq+q6gxgBggLxfqeuzngaHQkyCuIo7lHKa8qx2w009qjLYdyUgjy8sCq5qHTeQEeCEHOQwiyzLGTNH6qR5yBBOKPvxYxF1KIikoIIaioHOMYFixUUKG5TikoWia3K64XNN7cmc58xmdUUEEiiWxkIwkk8BZvYcDAZCazgQ0YMPAzP/MN3wBivNkuzNdxHZ7UPktBIrkS1EuMFUUxIoR4nqqq39aySRrQutrzQJCpjmfiafbEzcENvU5EvHqdnkD3dhzKSUWnGGjn6YlOZ0BkWOchPJA9kA5PkqaGAQOetgbUmJ9cQAGVVAKi69oBB3ToKKOMDFvTodOE2W6AcT5xNmHSio5UJ4ooDnGIXezCggUDBoIIwh9/fuAH5jMfgDDCuM7W4ojTSohKJFeC+mRTK8BMYI+qqu/WsdmPwGOKoixAJG7lN+Xx4suJfapTekE6TkYnvJy88HZuxYHsAyjKEdp5BqIoJkTJzALb4oYUZElTxp5IpaAQQQSllGpRcxFF+OFHEEGUUUY66aiolFFGAaITzoBBi5xdcKn3/GaAMbZWTDHb2c5GNmLAwNM8TSWVDGEICgollDCf+cxlLgoKYYRpUXMccbhKpzXJZaQ+kXEv4G5gt6Io9rkuLwBtAFRVnQ78hJjWdAAxtWlCw19q80FFpai8iJOFJ3HQO+Bh9qCVayvKLHkoSi7CXtGMmElWhLRclDQnqmdp++GHFSuqbVTLgkXr1gZRLMSAAQWFYoq1ZLCLEWdnnIm3terEE89GNnKQg4AoIxpHHBVUMI95zGY2OnR0opMmzF3oggceDf3RSK5hLiiBqyGJi4tTzzQquJaSfiqrKtmbuReraqWTbyeMeiMHsw/g42zCzdGMorghLBaLEJaLLoDTVb3mC+FaupeShsVeN7uIIgoppIQSQgnFEUeyySaPPC1ythcaMWDAGWdNnOtbT7s6WWSx1dbiiacvfUkiiXGMwwcfVFQyydS63EMIIc7WutJVJoRJ6sUlJ3BJGhaj3kgH7w7szdxLSnYKYT5hBHkFs//UfnQ6HS6mAhTFHSHC9ghZBxfoQSuRNDV06LSx4pa0pIoqdOhQULBgoYgiLXJ2xBEjRgwYKKaYfPK1Y1SPnM2Yz5ut7YMPg23Njjvu/I2/sZWtJJOMFSs6dAxjGCc4wXd8x//4HyDmR9vFuQtd5FQqyQUhU3XP4FItFFevXq3t+91336EoCkuWLKn1XIv+t4iRfUYyuMdgPp/1OXqdng4tOnA0L5+SSmEqMWXKywQEhBMb24/Y2Dh++umHWo8lLRQlzRU9ek3U/PEnmmg60AF//LXa2e1oRwQRtKAFLrjghBNllHGc4+xnP7vZTQopZJBBIYX1NrwIJJDneI7FLGYTm/iCL3iIh3iWZ5nJTB4UMzXxwIMKKviZn3mBF7iZm+lNb57iKeYylySSLsiaUnLtISPjM7hUC8X58+fTr18/ABYsWFBnoYycnBxeeeUVtm7disVqocd1PRgzYgyenp50aNGRlFP7cDE5UGVVbRaKzyCmPFkRRdHOvnXSQlFyLVA9cga0KNn+uJRSTWztY8tGjBRRxAlOaNs64aR1bTvjjBHjOc/rggu9bM3OAAZgwsR2trONbZRSihEj/+Af7GIX61nPClZo54slVsv4jiZalu+UaEgxrgW7heLIkSM1C8X6uDbFx8ezbt06KisrKS8v58CBA8TGxta67YoVKxgwYABeXsJibsCAASz5YQl3jLoDb2dvgr06sDdrL9kl2bi6OgFVnJ7ylIuYg3z+2/fwww+zZcsWSktLGTlyJK+88kqN9VVVVdx///1s3boVRVG47777ePrppzl48CCPPvooWVlZODk58fnnnxMWFnbe80kkV5rqXcFtaauNJxfZmhkz/vijopJIIkaM6NBRRRVZtgZimlV1cXbA4bzdzO1t7T7uQ0UllVQOc5h+iB/kYxlLNtmYMOGKK/vYx3rWAyLiDyWUrnSlM53pQhf88DvX6STNmEYtxn37fn7Wa6NHR/HIIz0oKangllvmnLV+/PgujB/flVOnihk58n811q1d+3/1Om9DWCjm5+czdOhQDh06VOu2wkLx9NTsgIAADhw+wOHcw+gUHV5OXnRo0YEqq4WPP/6Er7/6iri465g27W08PRVqE2RpoSiRCHE225oPPtrrVVThjjvFtmbHF18MGCikkDzyyCEHEGJZXZydcDrnuLOCoomznf/yX7axTYuc97KXeOK5m7vZznZ+5EcWsICv+AqAVrSiC12ItbVQQutdNlTStJF3uRYu1ULxv//9L/n5+UybNo033nij1u3OzGJXFAVvF29cTC6k5qQC4OXkxTOPT+LhSQ8R7tuSKS+/zaRJzzJr1hecjpA9wNa9Vls39aJFi5gxYwYWi4WMjAySk5NriHF1C8Vbb72VgQMH1rBQtFNeXn5Bn4NE0tgwYKCNmJGJBYsmyh544IQTjjhSSCEGDJqphd21Ck6LvHO1ZsR4zujZF98aSWFFFJFPPgEE0IUuzGe+VgDFEUcqqGAd61jGMkB0bUcRRSyxdKYzMcTIKVXNlEYtxueKZJ2cTOdc7+3tXO9IuDYa0kIRYNOmTTz00EMATJ06lcDAwBpuUMJCsS8dvTuScipFE+TQoFAyizJJyc7g/gfu5Pah9wAKEyZMYseObbRq5c9PP/1c67VIC0WJpHYMGHC3NTtmzAQQoIm0XSQ70AErVnLJpYSSGl3bRoxa1Fyf6Nme3Q1i3vN61nOIQ+xgBwm2Np7x9KQnK1jBp3zKXvaymc3a2HgwwUTbWgwxhBAio+dmgLyDddCQFooA3bt3ryFuOTk5vPDCC+TmCmOIX3/9lTfffBO9Tq8JcmllqWahaLFa+GDGe0REhAKlfPnlHEQiV55tsZ51HdJCUSKpPyZMNeYKV1BBMcW44IKCQgEFlCN6iOzTqYAaxUiqR892gTZhqjN61qHTurZHInq17KYbvejFNraxk52aEOvR44Yb61jH93wPiLHuSCKJsbVoomtYXkqaBlKM66AhLRRrw8vLi8mTJ9OtWzcAm4WiSOZ66MGHePChBwmNC+Wex+7RLBT9Avx4/s1nSc/Pp5WbM4qiR3RT5wEVcMbUCWmhKJFcPCZbs9OKVnjgQQklFFNMCSUARBJJJZWkkUY55VRRxSlOaQJqwKAJs12kz+WzbBfuEEL4iI9QUTnBCXazm53s5D7uwwsv3ud9ZjADBYUUUtjBDs0Ryw8/LXKOJpoIInBqQkWDrkVkBa5GTmllKfuz9tPaozWeZk+O5B7hVMkpWrn60dLNjKI4I8aM8xBGWafHkK8m8l5KrgXsxhMAhzhEAQWaINrtI02YKKZYi6pBRLPVu7brU5TkTI5whPWsZ6etHeYwevQ8wzPsYQ8b2cgpTmnXEkpoDYEOIuiibSslF4+swNVEMelNOBgcSM1JJdgrmLaebdHr9JwoysTTqRVmYyHgzOkIOY/GIsgSSXOn+lhtEEHatCp75GzCpE2r2sUudOi0ffLJJ9fmX66g4IijJs72hLJzJYe1tbVxjAMgjzwOcpCudAXgER5hDWu04x/jGAc4wEIWAmL8OpJIookmiiiiiZYlPa8iUowbOdXHkO2CHOgeiE7RkXwynRCflrg6FCPGjKsLsr22tUQiuVJUn1ZVHRUVf/wpsbUKKgBRgtMFFy3LOpvsGtGsGbMmzk44nXPuswcemhADfMzHpJHG7mrNCy+e4Al2s5t3eZfd7GYTm7QudXv3tl2cI4jQEs4klxcpxk2AMwU5xDuEAPcA9Do9+7LSCPbyw8sJRHVTT4QPcj6n3Z8kEsnVRIeuRkEPCxYtcnbEET16LUMbRIKYHj1WrDXGn/XozxLouhLEFBRa29ot1Jyi2Z727LK1/ezXSnU64sg+9rGSlTW2jbK1aKIJIaTGWLqkYZBi3ESwC3JGYQYuJvFL1d/VH52iIzXnKOWWFvi6eKHX2ZO6CoBCROUuZ6QfskTSeDBgwA037bkLLkQQoUXO9taBDpgxk0kmWWShQ0c55RRRpO17IQJtR0HhZV4GoJxy9rGPRBIJIojruZ5EEhmFqDOQRhrHOKZlbxsx0olORBKpibQcf750pBg3IfQ6PYHugQBYqixkFmfS0rUlekXPodxD5JWV0dG7AwZdCUKAdQh76SpEt7UUZImkMaKgaNnb9qIe9mjYPp7siCMllNQwuWhFK8opp5hiMsnUXrcLtF2kzZjrHIN2wEGbt2wnhBC+4isSSdS6uI9xjPGMR4eOTWxiIQs1xyonnIi0tWiiiSSSVrSSrlUXgBTjJkpeWR7HC45TXFFMsFcw7ZX2pOakcjjnIO1beKMouYhuaj1QjBhHdkcadUkkTYPqQmYvUKKiUkEFpZRSRhm++KKgcJjDlCEK+uhtzS7S1UW9ujjbk8Rqi2hNmDQ7SDu55GpZ4GtYwyu8wklOAlBCCTvZyXa2a13ennhqAm1vMkGsbqQYn4GiKNx111189ZWoFWuxWGjZsiXdu3dnxIgRfPDBBwAkJycTGhqKXq/n5ptvJiwsjAkTJrBq1SrNtem7775j+PDhLF68uFY3pTlz5mi1pF966SXuvffes7YZM2YM+/btA0TdaHtlrKKsIuI7xdM6uDU6RUfvXr15+4O3OZh9kD2ZJwj1acmqlT/y3HOvA3DgwEECAlpiNrsQHR3N3LlzG/7Dk0gklxUFBQdbq04AAXjhRQkllFJKCSXo0RNBBGWUcZSjVNpaNtk1nK4ccawh0mbMtVb08sRTe3yjrWWRRZKtJZLIG7xBGmn8l//yJ3+yiU2s43Rdfx98iCKKCCKIJFKzvZRIMT6Lq2GhqCgKXbt2ZejQoXh6etbYbuHChdrjSZMm4e5+unxf+/bt+XPLn6TmpGLUGXHQOxDqHcqB7APsPnGUvv36k5BwI+BI37638c47LxEXFwtn1La1WCwYDPKrIJE0VYy2Vn0c2oq1RkWwQgq16BlE+U833CiltMY0KxCRsfmMVts4tA8+9LU1O5548iiPEk20JtLZZOOAA93pTjLJ/MZp0xl//DVxtgt0deG/VpB/gWvhalko/vLLL4wbN67W7VVVZdGiRWc5J3mYPQjxDuFo3lF0ig5HB0fCfMNsgnyIEO82uDk6IcaLXW3/5vLFF0tYteoPioqKKC8v57nnnuOjjz7i++9FksbEiRO54YYbuOuuu9iyZQt///vfKSoqwtfXl9mzZ+PnJ63eJJLGTPXu50BErokVK2WUUUopevR44IEVKzvZqW1rwICKSjHF5JNf43hnCrQ9E/xM7K5TIMa+M8kkjTRt6tXd3M1WRNGnE5wgm2xWs1rbvyUtNWEOJ5wIIvDCqwE/ncZHoxXjNNIopbRBj2nGrH0pz8XVsFAMDAwkPT29zmOvW7cOPz8/OnbsqL126NAhOnfujJubG6+++irhvcOxqlbKLeWE+YSRmpPK/lNH8XWx/xouA5wQpTPL2LBhPQkJCXh6erFq1apaz1teXs6TTz7Jjz/+iLe3N/PmzWPy5MnMmDHjvJ+HRCJpXNirglUvjamgEE44pbZm7+r2ww9vvCmggCMc0cahSyip0c3tgIPW1V1bFK2g4GdrduYylxOcIJlkkkgimWR88WUIQ0gkkQ/5kDWsqTHFygcfookm3NYiiKhhkdnUabRifDW5GhaKIMS8LuwRup2WLVty9OhRWrRowbZt2xg2bBhJSUmUKqUcyz+Gr4svwV7BHC84zsmik5RWlqKqVUARYu6xiYEDeyN6xS21nxRR1jIpKYn+/fsDUFVVRWDg+X/QSCSSpkH1cejq9ox2wwozZjzw0JLG7ELsiy969BTZ2plRdHWBtj+2j0UrKLS0tX70q3E9XehCKaXsYQ/JJJNBBiCmf6WSWiOC9sBD6+LuZGuBBDbJaVaNVozrE8FeTq6WhWJtWCwWvv32W7Zt26a95uDggIODSOLo2rUr7du3Z//+/Rw5eoR/vfwvKq2VvP7u6wztNxRHgyMWq4XUnFN0sZrQ60qBKpydPRDTnnIxGKyaWQSgWS2qqkp0dHS9egYkEknzwR7ZOuBAW9oCaOU+yyjDBReMGMkmm0IKa+xnF8M88sjm9N9PI8azRPrMjG4DBh7lUe15IYXsYx+eeNKe9iSTzFjGUkkleeSxnvX8xV/a9s4405GORBFFGGGEEkoHOpyV9NbYaLRifLW5WhaKtbFq1SrCwsJqRKRZWVl4eXmh1+tJTU0lJSWF4OBg4uLiGDF8BLkluRzOPUzyyWSCWwTjZHTCUmVh94nDhPm0Q5TPtCAqdhXQtq0nSUmJVFSUU1xcwm+//Ub//v0JDw8nPT2dzZs3c91111FRUUFKSgoREREX9JlIJJKmT23lPlvQokbkbP83mGB06EgnXStYoqJSSmkN8QYh+NXF2YxZK/3pimuNKVbhhLODHRzmMHvZyx72kESSVmVsNav5nd9rWE8qKLSlLRFEEFqt2aeGNQakGNfB1bRQfOCBB5g4cSJxceILuGDBgrMSu/744w/+9a9/YTAY0Ov1TJ8+XdsfwNPJE7PJzOGcwxh1RvQ6PW292qJX9CSdPEhhmRVVNSK+Ah4EBYUybNggoqIiCQkJpUuXLoCIwJcsWcITTzxBYWEhFouFSZMmSTGWSCQaevS42NqZeOGFEWMNodahI4QQyigjiyzKKaeQQs0XGk53n9vF2S7UDjigR6/5QN/KrTXO141uxBLLHltLJx0rVnzwYQc7WM5ybVsnnAgmmEgiCSOMjrbmiuvl+7DqQFooNnNUVdXGok8WnsTVwZWMwgxyS3Nxc3AjyCsIo74CkdTlgBhT1iEqdl18/Vl5LyUSSW2oqFRSqdW3Ps5x8smvMe3KiBFPPDUHrOpVx6qLdPXmgEOtY8UVVHCIQwQTjBEjc5nLTGbWqFh2Jh540IEORBHFRCbWmDJ2qdRloSjF+BqhwlJB4slEdIqOdp7tqKyq5GjeUfQ6PSHebXAyWRAi7MzpEprOiOzrC+/GkfdSIpFcCFaslFOuibJ9rnESSZrLFVDDhrL660ANka7+uLbpV6WUkkoqBzhAOunczu2kkMK7vEsKKYD4UbCFLQ063iz9jK9xTAYT4b7hHMw5yIHsA/g4+xDmG8bhnMMkZ6YS6O6Pn4sZRbH7I1sQZTQrEFHy2V9miUQiaSiqz2OuTjjhmkjbmxkzfvhRRRW72KXtX0XVWZndcDpx7Eyhtk+RshNAAN3prol0N7pdscQvKcbXEI5GRzr5dCK9IJ2TRScpqSghzDeMtPw00vJPkFfqTEdvP/S6YkRilwnRbZ2DKBjieK7DSyQSSYNT3SjjTHToCCa4hlCXUUZLWuKBB0UUkUYaVqyUUEIRRTXmSOvRnxVF1ybSVwIpxtcYOp2O1h6tcXd0p8pahV6np41HG1wdXDmSe4SdGYcJ9mqDu6MBRTEiuqhLEJaMlYAL0v1JIpE0BhQUzUTDjmprOltrQQvKKKOccm3s2RdfjBi1OdJnFjIBNGFuTetaa3U3NFKMr1HcHE8nJGQVZ3Gq+BQdWnQgvSCdA9mH8XDMo51nAAZ9ISJCNgOliG5rd+RXRyKRNEYUWwNRY7s1pysdVlFFOeWYMGHAgBkz5bZWHQ88tDnVtY03Xw7kX1QJRr2RiqoKUrJTCHALwMPsQXp+Ookni+joHYCTscKWke2EKKmZg4iQzcgoWSKRNBX06GuUAnXFlU500jK87RG0J55XJBquznlrhimKMktRlExFURLrWO+pKMp3iqLsUhRls6IokQ1/mVcORVG4++67tecWiwUfHx+GDBnCl19+SWxsLLGxsZhMJqKiooiNjeX5559n9uzZKIrC6tWnS7V99913KIrCkiVLaj3XnDlz6NixIx07dmTOnDm1bjNmzBjtnO3atdOMJw4fPozZbNbWTZw48ax9V6xYoa13cXEhNDSU2NhY7rnnnhrbeZo9CfcLx8XkwrH8YxSUFdDRuyOOBkf2ZB7haF4hVlWH6K42cnosOZdzldKUSCSSpoCCggkTbrjhg88VF2KoX2Q8G/gIqMsA9wUgQVXVOxRFCQM+hjOKjTYhmpqFYvWqXmcyaNAgBg0aBEDfvn155513tEIi1bFYLJgMJjp6dySrOIu0/DQAQn1CySzKJL0gndzSAjp6B+JkNKEojkA5UIiIki9+CpREIpFI6hEZq6r6B+Ivbl2Eg6jcrarqXqCdoihN2l/PbqEIZxs0nIv4+Hg2b95MZWUlRUVF9bZQ9PT01CwU68JuoVjfazkfX3zxBWPHjmXIkCEMHjyYVatWcccdd+Dr4ku0fzT/eOofzJs3D4Ds1Gz+b+SDxMX1pk//QRxLTwNUTkfJxYivSGWDXJtEIpFcazSEtcVOYDiAoijXAW2hYVweUmppWWQBYoJ4bevtRcktWM5aV1/Gjh3LggULKCsrY9euXXTv3r1e+1W3UPzhhx8YOnRonds2pIVinz59LsrIYcOGDXz11VesXLmyxusGvegwsVqtpGWnMenpSUyfM53f/vyNgcMG8uQ/nqSgrABVtVfuckTUus5FdF9fnUIyEolE0lRpiI7xfwMfKIqSAOwGdlDHQKKiKA8CDwK0adOmAU59eWjKFopubvUv2zZw4MCzusWro9PpIAcOpxzmzmF3itfQ4ePvw/5Tx3B3cCHIyweDvgxRFESPGFcuR44lSyQSSf25ZDFWVbUAmACgCDU5ZFtq23YGMANEOczzHbsjHetcp0N3zvUGDOdcfz6aqoXisWPHeOWVVwDRFV3bGLEdZ2dn7bHBYKjVQlGn6IiJjuHX337lSO4RiiuKifCLoKSyhGP5x0jIOEQbDz98nB1RFHuGdSFiGtSvQC/EmLJEIpFI6uKSxVhRFA+gRBV9lg8Af9gEuknTlC0U77jjjgu6XoC2bduSlJRERUUFxcXFZ1ko7t6xm27dupFfnM/B/QeJiIigoqqCcks5x/IyySjQ08oNvJ29URQz4quVgvhd1h2IQpbUlEgkkto5rxgrijIf6At4K4qSBryMyNxBVdXpQCdgrqIoVUAycP9lu9orSFO3ULxQgoKCGDZsGFFRUYSEhJzXQrF9SHvS89PRKTp8XXwprijmSN4RThWfIsTHB9FN7Y5I8PoT8dXoAwRc9DVKJBJJc0W6NkkumtLKUtLy08gvy8dB74CH2YPskmwsVgvWUxV0id6HQVcMtEB0yks72wAAIABJREFUW5cAIUBPqMX3VCKRSJo7dbk2NUQ2teQaxWw009G7Ix29O6IoCtkl2YT7huPr4ktOSQEtpjzBsj35WNVi4FagG3AAmAdsRyZ5SSQSiUCWw5RcMu6O7rj6uVJWWYbJYKK1e2uOOhzlxuB+3PblS/i7+vBsnwIeuf4RHI3ZwClgPWJWXBxiqrocT5ZIJNcuMjKWNAg6RYeTSdR8Laksoai8iJ/2/cTIqJGEeHdi0rJJdHonjJSsY6hqIad/B/4OfA3sQcxVlkgkkmsPKcaSBsfZ5EyAewAPXPcAPyT/wMajG7k9/HY8HD0JeXsiw+cuJD3fAVUtQZTQVBBF3P6HyMCWRUMkEsm1hRRjyWXBoDPwyR2fkPKPFO7tei+pOalsfnwzC/+2kN0Zxwl8/T6Gz/2WI7nuqOpwYDAiMl4BLEBMiZKiLJFIrg2kGEsuK2092zJj5Ay2PbkNo8HI4NDBVFRVMKDjALYc20e7N++m96eDWXPgKGCfmpUHLAcWIhK+pChLJJLmjRTjM2hOFooAa9euxd3dXduuf//+53z/s2fP5rHHHjvnNheDUW8EoNRSyo0dbmT1gdVkl2RzY/CNpJxK4aYZNzF45gcczA4EzW80F/gF0X29F6hq8OuSSCSSxoDMpj6D5mShaCc+Pp5ly5ZdwKdw+fB18WXOmDlM7jeZN357g7nb52LQGZjcfzIzNs6gw3/uYHDozXw47FHat8gCWgNZwCpgE9AVCEN+dSUSSXNCRsa1cC1YKC5dupTu3bvTuXNn+vfvz8mTJ8/aZvHixURGRhITE0Pv3r0BqKqq4tlnn6Vbt25ER0fz2WefXdT5O3h3YNboWaT8I4XXB73O1IFTOfj8QcbFjmPDkY10+M9t3DprAWsOKKjqGCAGMS95LcJaewfSslEikTQXGm148QZvsJe9DXrMMMJ4gRfOu93YsWOZOnUqQ4YMYdeuXdx33331siisbqGYn5/P0KFDOXSoVs+MBrVQdHNz47XXXiM+Pr7Ofe0/CkaNGsWLL77IDTfcwMaNG1EUhS+++IK33nqLadOm1dhv6tSprFixgoCAAPLy8gCYOXMm7u7ubNmyhfLycnr16sXAgQMJCgo67+dTG0FeQUzqMwkQPziW7VlGaWUpcYFxbDq6jZtm9Kdb627MGvUoEX7lKIoe8RvyL2AbQqSjEDaOEolE0jRptGJ8NWluFoq1dVOnpaUxZswYMjIyqKioqFVMe/Xqxfjx4xk9ejTDhw8HhKHFrl27tHHw/Px8UlJSLlqMq+Pi4ELypGTe/v1tZmyaQZmljE6+ncgoyCDq3fHc1CGOGSPuo32LIkTNaxdE1/V2IBKIRTpESSSSpkijFeP6RLCXk+ZioVgXjz/+OM888wxDhw5l7dq1TJky5axtpk+fzqZNm1i+fDmxsbEkJCSgqioffvghgwYNqu9HckEEegTywe0fMLn/ZD7f9DmfbvyU1Q+uJuF4AlNXTqXDfx7h5tAufHrHvfi59MRs8gc2I7qtdyF8SzojTCokEomkadBoxfhq01wsFOu69vz8fC0xra5M7oMHD9K9e3e6d+/O0qVLOXbsGIMGDeLTTz/lpptuwmg0sn//fgICAmp4IzcE3s7e/POmf/Jc3+fQ6XSE+IQwb8c8DuUeIvFEJkH/fhJXB1fu7XovLw+4GW9nBREVJ9mWjkAXwLtBr0sikUguB1KM66C5WyhOmTKFUaNGERAQQI8ePWod23722WdJSUlBVVX69etHTEyM1oXfpUsXVFXFx8eH77//vt7nvVB0utM5hi/c9ALOJmcW71qMgoK7ozvTN07n28SvmDbkXkZEhWDUqwhRPgjsBwIR48rtEJW+JBKJpPEhLRQll4XLeS/T89P5dMOnfL75c8Z3HY+72Z1P1n9Cfvkp/t77Fp6K74vZEIDJ0A7YDRQhuq2jEd3YpstyXRKJRHI+6rJQlGIsuSxciXtZYamgoqoCFwcXFiYs5J6F99DCqQWZxSdp6erJTR1u5fGedxHXeh9gREyFMgARCGGW48oSieTKIv2MJc0Ok8GEi4MLADGtYhgfN578snyqrFYqqvQs2rmIgV8M499r/uRkUbltLwvCuvErYBmiBrZ0i5JIJFcXKcaSZkGYbxifjfiM9JfS+WDoB3iaPQl0D2TakI9Yvicd/6lP0vm9//B94hFKK/VYrdHASUQN7FnABkRNbIlEIrnyyG5qyWXhat9LVVXJKMiglXsryirL8JvqB4DFaqGksoRWbq24r9t4JvfrgMlQXG3Plog5y+2R+Y0SiaShkd3UkmsKRVFo5d4KAJ2iY/rw6Vzf9nrKLGUAlFaW8vrqN2n75lPM3ppMaaU90/oEsBIRLa9FRM/SNUoikVxe5E9/SbPHZDAxrvM4xnUeR3p+Ol9v/5o52+YwZ/Qcdp7YyUsrpvN/Sz5jXGwPXuo3FA/HEHxdXYBkIBHwBMKBUE47SkkkEknDISPjM5AWipfHQrGxEOAewHM3PkfSpCSGhA/hpX4vMSx8GBarlSW7dxD+zov4vTqcuA/+ybpUe0Sci6iFPQv4ATGHWdo5SiSShkNGxmcgLRSvDarXAX9nyDv0ad+H2Vtn88s+4Zy1/9R++n3+JAM6hvNEr9vo0z4QR0MVcMy2OAIdENOkfK78G5BIJM0KGRnXgrRQFFxOC8XGhKPRkVHRo1h+33LSXkzj7Vvf5sWbXmTDo5sI8uzN7XOmYX7hEQbMmMkPSXmk5cUCAYgu7IWIiHkjUHo134ZEImnCNOrI+B7uOeu1m7mZO7mTUkp5iIfOWn+HreWSy5PULGc5l7n1Oq+0UBRcCQvFxkZLt5b8vc/ftec+zj4s3rWYTEsmv6cmsfrATlRUerTpwdu33sP1bc3odUXAVtviA3QD2gL6q/IeJBJJ06NRi/HVQlooCq60hWJjpI1nG9JfSmdlykq+3v413yZ+S5mljPSCdOI/fQSDzsBdnW/n2b4DCPPVoVPygJ8AB8APiENMl5J1sSUSSd00ajE+VyRrxnzO9Z541jsSrg1poXh1LBQbIwa9gcFhgxkcNpjCskK+S/qO0dGjSTyRyBM/PMHXO35g9rZvMOgMDAm7hcdvGEbvICcM+qPAUcR/s9YIw4pWyNEhiURyJo1ajK8m0kLx6looNlZcHV25p6sYPolrHcczvZ/hs42fsSZ1DRarhaV7lvN98o84Gh158cY7+b/ucfi6VKEohxClNx0RBUXaI8adZVe2RCKRYlwn0kKxcVgoNnZGRo9kZPRIcktyWbZnGd8mfktxRTERfhFM37SYyb/OwsnowL8G3MU9XeJo4eSLybAf4bmsB9ognKRaI8wsJBLJtYgshym5LFzL91JVVRRFIbckl6B/B5Fflq+t0yt6BocO4ItRQ20Rs32NgujCjkB4L0ubR4mkOVJXOUwZGUskDYw9Ec/TyZNTL59i3aF1fJf4HYt2LeJk0Un+OrIJ/1d/oY1HC57rO5wxMbF4OVlRlJNAOkKYvRERc0fAfNXei0QiuTKcNzJWFGUWMATIVFU1spb17sDXiP42A/COqqpfnu/EMjJu3sh7eTaqqrIjfQdtPNpwIPsAk1dMZtWBVQB4mp25MfgG7o0bzsBQfxwNR6vt6Q6EIEpyul6FK5dIJA1FXZFxfcS4N1AEzK1DjF8A3FVVfU5RFB9gH+CvqmrFuY4rxbh5I+/l+TmSe4T5CfOZvXU2+7L2aa+b9AYe6zmCh3rE08HbgE6prLaXL6Ibuz3Q4spesEQiuWQuuptaVdU/FEVpd65NAFdF9M25ADkIB3eJRHIO2nq25fkbn+f5G59nf9Z+FiQsYEvaFmJaxrBk9xLeXSdKoQ6PuJG/97mV4Bah+LqUoCibgc2IceUAIBoIRM5llkiaLvVK4LKJ8bI6ImNX4EcgDNGHNkZV1eXnO6aMjJs38l5eGpYqC53e6cSB7AM1Xvd39effg+9nWEQ73BwrUBSrbY0e4SoVhMjMlukgEklj5HImcA0CEoCbEH1nKxVFWaeqakEtF/Eg8CBAmzZtGuDUEknzxKA3sP8f+9mZsZPVKav5ed/P/HX4L9wc3Jj47TTGLyrDZDDxSPchPBl/E4EeZgy6FITto4IYZ+6IyM52uZpvRSKR1IOGKAU0AfhWFRxAVDYIq21DVVVnqKoap6pqnI9P43S6aWwWijt37uT6668nKiqK2267jYIC8RunPhaKK1as0Na7uLgQGhpKbGws99xzds1vSeNDURRiW8Uyqc8kVj24isJXC9n+1Hayp2Tz1i1vYbVaef+vbwn692MYn59AxDvv8b/t+VRWGYE8YAswG5gJrALSkNaPEkkjRVXV8y6IjJHEOtZ9CkyxPfZDzM3wPt8xu3btqp5JcnLyWa9daZydndXY2Fi1pKREVVVV/emnn9SYmBj11ltvrbFd27Zt1aysLO35l19+qUZFRan333+/9tro0aPVmJgYdfHixWedJzs7Ww0KClKzs7PVnJwcNSgoSM3JyTlru7i4OHXt2rWqqqrqzJkz1ZdeeklVVVU9dOiQGhERUe/31adPH3XLli21rqusrKz3cepLY7iXzZ0KS4X6x8E/1Lvn3616/stT5Vm0Je79TuqPiS+qReXTVav1Q1VVP1JV9UNVVT9RVXWWqqo/q6qaqqpq1VV8BxLJtQewVa1FE88bGSuKMh/YAIQqipKmKMr9iqJMVBTFHoq9CvRUFGU3sBp4TlXVUw37k+HK0pgsFPft26fZFw4YMIBvvvnmIt9VTb744gvGjh3LkCFDGDx4MKtWrWLYsGHa+okTJ/L1118DsGXLFvr06UPXrl0ZPHhwrXaLkiuPUW8kPjieuWPnkj0lm93P7ObIP4/wyR2fkFFUwNA5r+Py0kQ8Xn6eUXN/YnWKI+UWP6AEOAAsBz5BRM8JQNnVezMSyTVOfbKpz6lEqqoeBwY22BXZeOrHp0g4nnD+DS+A2FaxvD/0/fNu15gsFCMjI/nxxx+5/fbbWbx4MceOHdPW1ddCsS42bNhAQkICnp6erFq1qtZtysvLefLJJ/nxxx/x9vZm3rx5TJ48mRkzZlzQuSSXF0VRiPQX+ZUPX/8w/Tr04387/sfXO77mYPZBliT+xJLEn1AUhevbXM/dnW9kVEwUXk5lKEoh8CfwF+ABeHLa1EJmaEskVwKZclkLjclCcdasWTzxxBNMnTqVoUOHYjKJMokXYqFYFwMHDsTT0/Oc2+zZs4ekpCT69+8PQFVVVQ3DCknjJMQnhCkDpzBl4BROFJ5gzYE1HMg+QJW1iqXJS3n4+9d5+HswG8zEBXbl733u5pawcAz6PUAukIrI0PZFJIEFI0t0SiSXj0YrxvWJYC8njcVCMSwsjF9//RWA/fv3a93n9bVQtJtN1EZ1pyWDwYDVatWel5WJLktVVYmOjq5Xz4CkceLv6s+4zqc7uF7s9yIzNs5g4c6FbD++nXWH/2Td4T8x6ozc0ukWHuo+mBvbe+FozAYybIsO4cvsjZhC5YOMmiWShqPRivHVprFYKGZmZuLr64vVauW1117Tsqbra6FYX9q2bUtSUhIVFRUUFxfz22+/0b9/f8LDw0lPT2fz5s1cd911VFRUkJKSQkRExAWfQ9I4MOqNPNrrUR7t9SgA+zL38d6698guyWbTsU3cMusHAPxc/Hj0+hHc0/Vm2nh6oSiHgZ22RY8Q5o6IyROOtZ1KIpHUEynGddBYLBTnz5/Pxx9/DMDw4cOZMGECcOkWimcSFBTEsGHDiIqKIiQkhC5dugAiAl+yZAlPPPEEhYWFWCwWJk2aJMW4GRHqG8r0EdMB0RPyv4T/8eqqV9mftZ9/rfyEf638BEeDI0PDb+HRnv2JC3THbCy0GVucRIw1+9sWb0SXtrSDlEguBGmhKLksyHvZ9CksK2ThroXM3DyT3Sd24+bgRkZhBgDuju6MjR7Mo7360s4rAFeHUoQw23FCdGuHICqCyfFmiQQuwSjiciHFuHkj72Xzwv53IuVUCuPmjWP78e011ge4BfBYz3Hc1SWKVu5WdErRGUfwRUTO7RD1tPWX/6IlkkaI9DOWSCQXjT3TP8QnhG1PbeNY3jGW71nOol2L+OvwX1SpVbyy6iP++UsZRp2RHm3ieLr3KPp3aIeroz+iFtAu26IgytgHIjyb/ZHJYJJrHRkZSy4L8l5eO5RbysktzcXNwY3vE7/n7oV3Y1VPZ+a3cGrBLWG38FjPIXQOcMCoz0YUHrHjirBDd0AYXUhxljRfZDe15Ioi7+W1S3FFMX8d+otvE7/ll/2/cDT3KHqdHovVgk7R4e/qz03t+/DEDbcR08oHk74cOMZp51UFcEMUHemEGHuW4ixpHshuaolEckVwNjkzMHQgA0NFYb7CskIqqirYmbGTj9Z/xHeJ3/H1jvl8vWM+CgqB7oGMjR3N+K69CfXVoddlAfm2ZQ/gjIiWjUA4QqQlkuaFjIwllwV5LyV1UVBWwKqUVcxPmM9fh//iROEJAFRUHPQOBLgHMKBDP57ufQchPt4oSg4icrbXztYhLCLbIMp21r/qnERytakrMm4IC8VmhbRQlEguL26ObgyPGs7iuxdzfPJxil4rIueVHL675zvCfMNIzUnls82fE/bOLZj+2ZPIaZP4covCycI4xBxmB0TJzp3AXOAbYC2wCcgGrk6AIZFcErVZOV2JRVooSgtFiaQ2isqL1Hnb5qlDZg1Rfab4qKbnTZo1pNMLTmrIf0LUh7/5P3XPiW9Uq/VPVVUXqMIe0r58rKrqV6qqrlZVtezqvRGJpBa4WAvFaxFpoSgtFCVXD2eTM3d2uZOlE5aS+XImpa+XsvPpnXww9APcHN1IOZXCpxs/p9O0ERie703Me2/w+UYrafntsFoDEGPLeUAyMBP4FhE9LwUSEVG1jJ4ljYtGncDVd3rfs14bHT2aR3o+QklFCbfMOttRaXzceMbHjedU8SlGfjWyxrq1E9fW67zSQvE00kJRcrXR6XREt4wmumU0T9zwBJWWSpbtXcainYtYf2Q96fnpPPitKF2roNDSrSUDO/ZmYo8hdGvdBp0uA8i0He2I7V89Iku7L2L82YosRCK5mjRqMb5aSAvF00gLRUljw2gwckfkHdwRedoQ5VjeMRbvWsynGz7lUO4hZm9bwOxtC1BQ6Nu+LxPi/kbf9qG0cqtAr8tAZGqnAV8jxqArEWYX3ogKYb62x+Yr/fYk1yiNWozPFck6mZzOud7b2bvekXBtSAtFaaEoaTq09mjNM72f4Znez6CqKn8e/pM5W+ew9uBadmXs4p6FawAROfu5+nFd6zju7Xo7t4Z1wcF4ChExlwBHbQuI6VTdEKK8HWiBEOkWyHnPkoamUYvx1URaKEoLRUnTRFEU4oPiiQ8SwzaqqpJ8MplZW2axdM9SDucc5sfkZfyYvAyALgFd6NWuF90CoxgYEoWfaxmifOcJxDjzmTgizC+6IHydJZJLR4pxHUgLRWmhKGkeKIpChH8E026bxrTbpgGQkJ7A3G1z2XhsIw56B2ZunsmHf4kSnUadEX9XfzoHRHJX5/6MjO6JohxHiLMVMd85BShCzHXWI5LC2iG6uKW3s+TCkUU/JJcFeS8lTQlLlYVpf0xjZcpK9mTuIbMoE4tVlOf0MHvQs21PUrNT6BrYjhFRvYgLbE1LNz0GXWEtR3MEvICbEVaSVcjkMIkdWZtackWR91LSlFFVlZRTKazcv5KdGTv58/Cf7Mncc9Z2w8Jv4dPhr+DnaqWkYj9OpmIUpbLaFi0QNbdLbI9b2hZ/hFBLrjVkbWqJRCKpJ4qiEOITQojP6STMU8Wn2HBkA3+k/sGqlFUknUzi++Sf+D75J3ycfcgqzsLR4MigkBsYH3cLvdp1wNtZtXVxWxDd3CeAHYjouR9ClE8CLohoWpZ+uFaRYiyRSCT1wNvZm9vCb+O28NsA0bWdeDKRDUc28Hvq76w5uIbMokx+SF7FD8li3v7AkIHc3+0+olsG4eVUia9LFaLOdh6w3HZkBVGERI9ICAsE2iIiaMm1ghRjiUQiuQgMegOxrWKJbRXLw9c/DEBWURYbjmzg530/89uB3/gj9Q9+3f+rto+zyZmOLTpyQ7sejIntS482bTDo7UVJqkfP+xEOVd5AEkKYfW2L6Uq+TckVQo4ZSy4L8l5KJFBZVcnO4ztZumcpy5KXsSdrD6WVpdr6Tr6d6NGmBzoUAjycGBoeRydfd5xMhYjCJLXhBsQDQUAFIqvbHZkk1jSQCVySK4q8lxJJ7ZwsPMmag2tYsX8FJwtPsiVtC6eKT9XYxt3RnRvb9+TD21+ilZsBne44YmzZnhxmL+fpABxEjDV7IJLEvIBOiHFoSWNDWijWk+ZkoQiwdu1a3N3dte3sZS3rYvbs2Tz22GPn3EYikVw8fq5+jI0dy5ejv+Sn+38i81+ZpD6fytwxcxkXO45gr2CKyov4PulnWr/RC68pA/Gb+hDR737MK7+msjezDVa1E1COEGIQ858LgcMIK8mTiClV+4AfgPWIudF5SJOMxokcMz4DZ2dnEhMTKS0txWw2s3LlSgICAgCYMGGCVnSjXbt2rFmzBm9vb0CIWFRUFPPnz6dfv34ALFiwgJiYmFrPk5OTwyuvvMLWrVtRFIWuXbsydOjQs2pFP/DAA7zzzjv06dOHWbNm8fbbb/Pqq68C0L59+xpVveoiPj6eZcuWXdwHIpFILiuKohDkFUSQVxB3dxWBgKXKwp7MPWxJ28KmI5tYtGsRu0/sZveJ3UxZ9R4KCp18O/Foz4e4qUMEBt0pWnsYcTDkIKLnnxGxlpPteRqnRdgETEC4W2Ujomx3ZInPq4uMjGvhWrBQXLp0Kd27d6dz587079+/VlvExYsXExkZSUxMjHYNVVVVPPvss3Tr1o3o6Gg+++yzBrkeiURyGoPeQFTLKO7rdh+fjfyM3Km5FL9WzLxx8xgbM5YO3h3IKMzg0e+fpNM7/en41lgcXxiBy4tPcdusRXy24QgZBX6cni5VPRpWgN8Q9bZ/Q5hlfA58j4iga3eak1xeGnFkvA7IauBj+iASH85Nc7NQXLdunfajYNSoUbz44ovccMMNbNy4EUVR+OKLL3jrrbeYNm1ajf2mTp3KihUrCAgIIC8vD4CZM2fi7u7Oli1bKC8vp1evXgwcOJCgoKDzfj4SieTicTI5cWfnO7mz853aa8fyjvF76u98s/sbtqVvIy0/jWV717Fs7zomLXemR5seBLoFkpy5haHhPejbPpIQHxdaOKWj16VUO7qKyOhOR4ixD2LM+Q9EJG3P5Jbj0JeLRizGV4/mZqFYWzd1WloaY8aMISMjg4qKilrFtFevXowfP57Ro0czfPhwQBha7Nq1SxsHz8/PJyUlRYqxRHIVaO3Rmru63MVdXe4CoLSylB3pO9iatpWE4wnsytjF76m/Y7Fa2JKWXGPfF2+axF2dB+HpVIWiZOHtpKLTVSDqbM9GdHFbEBnbdpyAGKArQsCPAs62xRHZ1X3xnFeMFUWZBQwBMlVVjaxl/bPA36odrxPgo6pqzqVd2vkj2MtJc7FQrIvHH3+cZ555hqFDh7J27VqmTJly1jbTp09n06ZNLF++nNjYWBISElBVlQ8//JBBgwbV+zORSCRXBrPRzP+3d+/BVdZnAse/zzlJTCCGSy4QEkJUIGC4t5sAVleDbbDVplWctrNMWWdnHGedWXtzh91xRt3B2e50uq477so4rQO1VGusYOPuH4qWgTJbJCEXCHc1kHtC5CK3JCTP/vF7SQ4hwAke8p5zeD4zZ07e3/vmvD+e8ObJ731/l6X5S1mav3SgrLevl/0d+6luqWZ7w3b+cuQvHDx2kBc++iUvfPRLBEFRBOG2ibmUFd7FAwVfoWT6HQQDXVyajHsZ7L19KxD6aC2IS8pfxY2RvoCb4CTdO9YS9dWE0zJeB7wM/Ga4nar6C+AXACLyEPDjL5+I/RcvSyheqe4nT54c6Jh2pZ7cn3zyCcXFxRQXF1NRUUFjYyOlpaW88sorlJSUkJiYyMGDB8nJyblkbWRjTPRIDCYyN3suc7Pn8sOv/BBwd+YajjdQ01LD9s+2s+2zbezr3Mennzfy4rY3eXHbm+RPyKdoahHHz7aRmnyOr8+Yx19NncaMjG7GJXeEnCEJ12JOwiXcXlxP7i4GZxlLwk1gkg7M8d5NqGsmY1XdKiL5YX7eD4A3vkyFokW8L6H43HPP8eijj5KTk8PixYuHfbb99NNPc+jQIVSVZcuWMX/+/IFb+IsWLUJVyczMZNOmTWGf1xjjv9Ae3N+dM/jH+7Ezx6hudi3o+vZ6Pm78mIbjDQBs3LOTYCBIX38f09NzeGvlK8zMTKfh80oyU7vJGBsgIIobVrUdN+45D5dm+oBzwH7gDu9snwF/xnUyS2dwjPQEbsa+xWFN+uEl4/eGu00dcswYXP/56eG0jG3Sj/hmP0tj4sOxM8eoaqqisqmSyqZKdjTuoPVU68D+pGASPX09JAQCzJmUy4Ozl/DI3HtYkDMV1zo+HfJpybikm4G7rd3p7Q8d/7wSl8iP4hL7BO81nniYZWw0Vm16CNh+tUQsIo8DjwPk5eVF8NTGGGNuhIyxGZQWlFJaMNhPpP2Ldmpba6ltqaW2tZadjTs51HWImtaj1LQe5edb/kDR1CIWTFnA23UbuOf2OymZPp+FOXncPrGbjLGtBKTf+zTBJdo03NjnE957E274FUOO+z6DiTzBK4v959GRbBlvBMpV9XfhnNhaxvHNfpbG3FzO955nb/tealprqG6upqbFvZ/pPQOAeAlTUcoKH+K5+39Czrhb2PrZO8zLnkLuuBRSEi+EfGIKrgWdinvmrLhOYSXe/vdwM44l4YZiTcJNERrdIztuaMtYRMYBf427v2CMMeYmk5yYzKLcRSzKXQSuKwz9/f0c7jpMdXM1u1p2UdVURVVTFe/WV/BufQU7IoE2AAAI9klEQVSCkBBIoLffzbk9PiWFe2+fw78+8FNmZaXT199GQJoQudhoHANU4MY8TwOmAKdwY6RrcCteXUzGf2ZwjHQmrqd39ApnaNMbwL1Ahog0Ac/i7iGgqmu9w74LvK+qZ25QPY0xxsSYQCDAzMyZzMycyfcWfA9wPbkbTzRS3VJNdXM11S3VVDVX0XyymRPnzrGpfif/d+QpiqYWEZAA7x/6X+7On81DdxazZNodTE/vIS35aEiCTsQ9hy7A3epuxnUEawJCF+AYixsjvcjbPotL7tEhnN7U15wLUlXX4YZAGWOMMVckIuRNyCNvQh5lhWUD5V1nutyt7ZAkvb9jP4ry/qE6PjxcT7/2oyj//Z3/5BszF9J2ei/nepsoyDzPpNQOkhL6Q840FshlMOF2M7gW9FngNVyHslu9VyowA3eruw84733v6DyPthm4jDHG+C59bDrLZixj2YxlA2Vne86yu233wDPoquYq6lrr+PtN/wBAQAIEJMCFfveseUraOIqm3sHbK9cSDJ7g5PlPSU3qJTgwUqoRqMN1+prG4HPoE96+TFwyPgaUA6W4BH3jWTIeQkRYuXIlr7/+OuCWUMzOzqa4uJhHHnmEl156CYC9e/dSUFBAMBhk+fLlzJo1i8cee4zNmzcPrNq0ceNGHn74YcrLy1mxYsVl51q/fj1r1qwB4JlnnmHVqlWXHVNbW8sTTzzB6dOnyc/PZ8OGDaSlpdHQ0MDs2bMpKCgAYPHixaxdu/ay79+yZQtlZWUD01VmZGSwefPmK/77161bR2VlJS+//PJIwmaMMRE3JmkMxXnFFOcVD5Rd6LvAgc4D1LTUsKd9D7tbd1PXWkfjyUZaTp1kU/0uxj9bwpzJczhy/AidZ9qZmZnF/Oxc7rmtkKX5BczL7sUtOemoBhG5OJzqDK7z2FdxyXl0WDIewpZQNMaY6JUQTKBwciGFkwsvKT91/hT17fXsbt09kKR7+3q50N/P3vY29ra38XZdLRPHTKSssIyFOXP4n32/IXNsAnMmT2HhlGnMysoiO21wMQzVXYj8LaPxbNmS8TAuLqG4YsWKgSUUw1m16e6772bbtm309vbS3d0d9hKKwMASikOXaxy6hGJpaelAMv4yKioqWLNmDT09PaSnp7NhwwYmTZp0yTHl5eU8//zzBINBxo0bx9atW+nr62P16tVs2bKF7u5unnzyyYE5t40xxi9pyWksmbaEJdOWXFLecbqD+rZ66tsHX+V15by649VLvjftljQSAgl8Lb+In39zNdm33sKRE7XcNjFlVOof5cn4nWHKZgBzcfOfVgyzf7b3OodbYDvUw2Gd1ZZQdGwJRWNMrMtKzSJrehb3Tb9voExVafuijbrWOmpba6lpqaGmpYYDnQf4bXUDv61+i7Rb0pg3eR7vrLqfzNQbf7s6ypOxP2wJRceWUDTGxCMRITstm+y07EtmFjvbc5Y9bXtccm6tob6tngkpE67ySZET5cn4ai3ZxGvsT7nG/quzJRRtCUVjzM1lTNIYivKKKMorGvVzR3ky9o8toWhLKBpjzGixZHwFtoSiLaFojDGjJayFIm4EWygivtnP0hhjLnelhSJuvhWcjTHGmChjydgYY4zxmSVjY4wxxmdRl4z9eoZtIsd+hsYYMzJRlYyTk5Pp6uqyX+YxTFXp6uq6bGiXMcaYK4uqoU25ubk0NTXR2dnpd1XMl5CcnExubq7f1TDGmJgRVck4MTHRplU0xhhz04mq29TGGGPMzciSsTHGGOMzS8bGGGOMz3ybDlNEOoEjEfzIDOBYBD/vZmaxjByLZeRYLCPHYhk5I43lNFW9bIFk35JxpIlI5XDzfZqRs1hGjsUyciyWkWOxjJxIxdJuUxtjjDE+s2RsjDHG+CyekvGrflcgjlgsI8diGTkWy8ixWEZORGIZN8+MjTHGmFgVTy1jY4wxJibFRTIWkeUickBEDovIar/rE0tE5DUR6RCRPSFlE0XkAxE55L1P8LOOsUJEporIn0Rkn4jUi8hTXrnFcwREJFlEPhaRWi+Oz3vlt4nIDi+OvxeRJL/rGitEJCgi1SLynrdtsbwOItIgIrtFpEZEKr2yiFzfMZ+MRSQI/BfwAHAn8AMRudPfWsWUdcDyIWWrgQ9VdQbwobdtru0C8FNVnQ0sBp70/i9aPEemGyhR1fnAAmC5iCwG/g140YvjceDvfKxjrHkK2BeybbG8fvep6oKQ4UwRub5jPhkDRcBhVf1UVXuAN4Eyn+sUM1R1K/D5kOIyYL339XrgO6NaqRilqq2qusv7+gvcL78cLJ4jos5pbzPReylQArztlVscwyQiucC3gF9524LFMpIicn3HQzLOARpDtpu8MnP9JqlqK7gEA2T5XJ+YIyL5wEJgBxbPEfNuq9YAHcAHwCfACVW94B1i13n4/gP4R6Df207HYnm9FHhfRKpE5HGvLCLXd1QtoXidZJgy6yJufCMiqcAfgB+p6inXEDEjoap9wAIRGQ9sBGYPd9jo1ir2iMiDQIeqVonIvReLhznUYhmeu1S1RUSygA9EZH+kPjgeWsZNwNSQ7Vygxae6xIt2EckG8N47fK5PzBCRRFwi3qCq73jFFs/rpKongC24Z/DjReRiA8Ku8/DcBXxbRBpwj/BKcC1li+V1UNUW770D90diERG6vuMhGe8EZni9A5OA7wN/9LlOse6PwCrv61XAuz7WJWZ4z+J+DexT1X8P2WXxHAERyfRaxIhICnA/7vn7n4AV3mEWxzCo6j+paq6q5uN+N36kqn+DxXLERGSsiNx68WvgG8AeInR9x8WkHyLyTdxfe0HgNVV9wecqxQwReQO4F7fySDvwLLAJeAvIA44Cj6rq0E5eZggR+RqwDdjN4PO5f8Y9N7Z4hklE5uE6wgRxDYa3VPVfROR2XOtuIlANrFTVbv9qGlu829Q/U9UHLZYj58Vso7eZAPxOVV8QkXQicH3HRTI2xhhjYlk83KY2xhhjYpolY2OMMcZnloyNMcYYn1kyNsYYY3xmydgYY4zxmSVjY4wxxmeWjI0xxhifWTI2xhhjfPb/ufv6V6ahphsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\"\"\"Code Here\n",
    "將結果繪出\n",
    "\"\"\"\n",
    "# color_bar = [\"r\", \"g\", \"b\", \"y\", \"m\", \"k\", \"r\", \"g\", \"b\", \"y\", \"m\", \"k\"]\n",
    "color_bar = ['#DDDDFF','#7D7DFF','#0000C6','#000079','#CEFFCE','#28FF28','#007500','#FFFF93']\n",
    "plt.figure(figsize=(8,6))\n",
    "for i, cond in enumerate(results.keys()):\n",
    "    plt.plot(range(len(results[cond]['train-loss'])),results[cond]['train-loss'], '-', label=cond, color=color_bar[i])\n",
    "    plt.plot(range(len(results[cond]['valid-loss'])),results[cond]['valid-loss'], '--', label=cond, color=color_bar[i])\n",
    "plt.title(\"Loss\")\n",
    "plt.legend()\n",
    "plt.show()\n",
    "\n",
    "plt.figure(figsize=(8,6))\n",
    "for i, cond in enumerate(results.keys()):\n",
    "    plt.plot(range(len(results[cond]['train-acc'])),results[cond]['train-acc'], '-', label=cond, color=color_bar[i])\n",
    "    plt.plot(range(len(results[cond]['valid-acc'])),results[cond]['valid-acc'], '--', label=cond, color=color_bar[i])\n",
    "plt.title(\"Accuracy\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在有開啟NESTEROV下 準確度較快提升，動量較高者也比較有機會加速找到最佳成果"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}