02_15_2023 Convolutional layer.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "e286a399",
   "metadata": {},
   "source": [
    "## 2-D convolutional layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "90bacd7e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[[[ 0.,  5., -2.],\n",
      "          [ 1., -6.,  2.],\n",
      "          [-6.,  6., -2.]]]], grad_fn=<ConvolutionBackward0>)\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from torch.nn.parameter import Parameter\n",
    "\n",
    "# A is the input, 4x4 tensor\n",
    "A = torch.tensor([[[[1, 2, 0, 1],\n",
    "                    [-1, 0, 3, 2],\n",
    "                    [1, 3, 0, 1],\n",
    "                    [2, -2, 1, 0]]]]).float()\n",
    "\n",
    "# input_feature=1\n",
    "# output_feature=1\n",
    "# kernel_size=2\n",
    "conv2d = torch.nn.Conv2d(1, 1, kernel_size = 2,\n",
    "                         bias = False)\n",
    "\n",
    "# kernel: 2x2 tensor\n",
    "conv2d.weight = Parameter(torch.tensor([[[[1, -1], [-1, 1]]]])\\\n",
    "                          .float())\n",
    "\n",
    "output = conv2d(A)\n",
    "\n",
    "print(output)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "250a2a5d",
   "metadata": {},
   "source": [
    "## 1-D convolutional layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "fa187175",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[[-1.,  0., -1.,  0.],\n",
      "         [ 0.,  4.,  0.,  6.]]], grad_fn=<ConvolutionBackward0>)\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from torch.nn.parameter import Parameter\n",
    "\n",
    "A = torch.tensor([[[1, 0, 2, 0, 3, 0]]]).float()\n",
    "\n",
    "conv1d = torch.nn.Conv1d(1, out_channels = 2, kernel_size = 3,\n",
    "                         bias = False)\n",
    "conv1d.weight = Parameter(torch.tensor([[[1, 0, -1]], \\\n",
    "                                        [[0, 2, 0]]]).float())\n",
    "\n",
    "output = conv1d(A)\n",
    "\n",
    "print(output)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ee1aff59",
   "metadata": {},
   "source": [
    "## zero-padding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "4ed87ad4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[[-1.0, 0.0, -1.0, 1.0, 2.0, -1.0]]]\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from torch.nn.parameter import Parameter\n",
    "\n",
    "A = torch.tensor([[[1, 0, 2, -1]]]).float()\n",
    "\n",
    "conv1d = torch.nn.Conv1d(1, 1, kernel_size = 3,\n",
    "                         bias = False, padding = 2)\n",
    "conv1d.weight = Parameter(torch.tensor([[[1, 0, -1]]]).\\\n",
    "                          float())\n",
    "\n",
    "output = conv1d(A)\n",
    "print(output.tolist())\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "34c438ba",
   "metadata": {},
   "source": [
    "## stride"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "586e07c2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[[-2.0, -2.0]]]\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from torch.nn.parameter import Parameter\n",
    "\n",
    "A = torch.tensor([[[1, 2, 3, 4, 5]]]).float()\n",
    "\n",
    "conv1d = torch.nn.Conv1d(1, 1, kernel_size = 3, \n",
    "                         bias = False, stride = 2)\n",
    "conv1d.weight = Parameter(torch.tensor([[[1, 0, -1]]]).\\\n",
    "                          float())\n",
    "\n",
    "output = conv1d(A)\n",
    "print(output.tolist())\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "287d92ad",
   "metadata": {},
   "source": [
    "## max pooling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "8121e4e1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[[2.0, 1.0], [3.0, 5.0]]]\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "\n",
    "A = torch.tensor([[\n",
    "    [1, 2, -1, 1],\n",
    "    [0, 1, -2, -1],\n",
    "    [3, 0, 5, 0],\n",
    "    [0, 1, 4, -3]\n",
    "]]).float()\n",
    "\n",
    "max_pool = torch.nn.MaxPool2d(2)\n",
    "out = max_pool(A)\n",
    "\n",
    "print(out.tolist())\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2479c0ac",
   "metadata": {},
   "source": [
    "## average pooling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "57d79d11",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[[1.0, -0.75], [1.0, 1.5]]]\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "\n",
    "A = torch.tensor([[\n",
    "    [1, 2, -1, 1],\n",
    "    [0, 1, -2, -1],\n",
    "    [3, 0, 5, 0],\n",
    "    [0, 1, 4, -3]\n",
    "]]).float()\n",
    "\n",
    "avg_pool = torch.nn.AvgPool2d(2)\n",
    "out = avg_pool(A)\n",
    "\n",
    "print(out.tolist())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "240b4582",
   "metadata": {},
   "source": [
    "## dropout"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "f0c25fdc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Initial Tensor: tensor([5., 9., 4., 8., 3.])\n",
      "Dropout Train: tensor([10., 18.,  0., 16.,  6.])\n",
      "Dropout Eval: tensor([5., 9., 4., 8., 3.])\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from torch.nn import Dropout\n",
    "\n",
    "torch.manual_seed(1)\n",
    "\n",
    "t = torch.randint(10, (5,)).float()\n",
    "print(f'Initial Tensor: {t}')\n",
    "\n",
    "dropout = Dropout(p = .5)\n",
    "\n",
    "dropout.train()\n",
    "r = dropout(t)\n",
    "print(f'Dropout Train: {r}')\n",
    "\n",
    "dropout.eval()\n",
    "r = dropout(t)\n",
    "print(f'Dropout Eval: {r}')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "87c0daa6",
   "metadata": {},
   "source": [
    "## how to use the ReLU function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "4dd905af",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAh8AAAGzCAYAAACPa3XZAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAA9hAAAPYQGoP6dpAAA6pElEQVR4nO3dd3hUdd7+8TuFTEJIAiEkIRAg9BISUSQCFtAoIiAoYFmeXVddK3ZAwBUQLKFZVuVR12cX3J8VUBAbrA1ZpEgzQOi9JqEmISF1vr8/AlmRAEmYmTPl/bquuS5z5sw59+Ew5HY+Z2b8jDFGAAAALuJvdQAAAOBbKB8AAMClKB8AAMClKB8AAMClKB8AAMClKB8AAMClKB8AAMClKB8AAMClKB8AAMClKB8AAMClKB8ALmjGjBny8/OruAUGBqpRo0b685//rP3791d7ewsXLpSfn59mz559znX8/Pz0yCOPVHrf7Nmz5efnp4ULF1Z73wCsF2h1AACeY8KECUpISFBhYaGWLVumGTNmaPHixVq/fr2Cg4OtjgfAQ1A+AFRZ79691blzZ0nSX/7yF0VFRWnSpEmaN2+ebrvtNovTAfAUjF0A1NhVV10lSdq+fXvFsk2bNmnQoEGKjIxUcHCwOnfurHnz5lkVEYAbonwAqLFdu3ZJkurVqydJysjI0BVXXKGNGzdq1KhRevnllxUaGqoBAwZozpw5FiYF4E4YuwCospycHB0+fFiFhYVavny5xo8fL5vNpr59+0qSHn/8cTVp0kQrVqyQzWaTJD388MO68sorNXLkSN1yyy1WxgfgJnjlA0CVpaamqkGDBoqPj9egQYMUGhqqefPmqXHjxjp69Kh++OEH3XbbbcrLy9Phw4d1+PBhHTlyRL169dLWrVtr9M4YAN6HVz4AVNm0adPUunVr5eTk6J///KcWLVpU8QrHtm3bZIzRmDFjNGbMmEofn52drUaNGjksj5+fn8O2BcB1KB8AqqxLly4V73YZMGCArrzySv3hD3/Q5s2bZbfbJUnDhw9Xr169Kn18y5Ytq7wvm82mkydPVnpfQUGBJPH2XsBDUT4A1EhAQIDS0tLUs2dPvfnmm7rnnnskSbVq1VJqaupFb79p06bavHlzpfedXt60adOL3g8A1+OaDwA11qNHD3Xp0kWvvfaawsPD1aNHD73zzjs6ePDgWeseOnSoWtu+6aabtGzZMq1ateqM5cePH9cHH3ygSy65RLGxsReVH4A1eOUDwEUZMWKEBg8erBkzZmjatGm68sor1bFjR913331q3ry5srKytHTpUu3bt0/p6elnPPbTTz/Vpk2bztrmXXfdpVGjRmnWrFm6+uqr9cADD6ht27Y6cOCAZsyYoYMHD2r69OmuOkQADkb5AHBRbr31VrVo0UJTp07Vfffdp5UrV2r8+PGaMWOGjhw5oujoaHXq1Eljx44967Eff/xxpdvs0aOHrrzySi1fvlzPPfecZs6cqaysLIWHh6tbt2765JNPlJKS4uxDA+AkfsYYY3UIAADgO7jmAwAAuBTlAwAAuBTlAwAAuBTlAwAAuBTlAwAAuBTlAwAAuJTbfc6H3W7XgQMHFBYWxpdGAQDgIYwxysvLU1xcnPz9z//ahtuVjwMHDig+Pt7qGAAAoAb27t2rxo0bn3cdtysfYWFhksrDh4eHW5wGAABURW5uruLj4yt+j5+P25WP06OW8PBwygcAAB6mKpdMcMEpAABwKcoHAABwKcoHAABwKcoHAABwKcoHAABwKcoHAABwKcoHAABwKcoHAABwKcoHAABwKcoHAABwqWqXj0WLFqlfv36Ki4uTn5+f5s6de8b9xhiNHTtWDRs2VEhIiFJTU7V161ZH5QUAAB6u2uUjPz9fycnJmjZtWqX3T548Wa+//rrefvttLV++XKGhoerVq5cKCwsvOiwAAPB81f5iud69e6t3796V3meM0WuvvaZnn31W/fv3lyT961//UkxMjObOnas77rjjrMcUFRWpqKio4ufc3NzqRgIAAFVQWmbXg++v0m2d43VDh1jLcjj0mo+dO3cqMzNTqampFcsiIiKUkpKipUuXVvqYtLQ0RUREVNzi4+MdGQkAAJzyzqId+m5jtobPSldOQYllORxaPjIzMyVJMTExZyyPiYmpuO/3Ro8erZycnIrb3r17HRkJAABI2pyZp9e+2yJJGtevgyJq17IsS7XHLo5ms9lks9msjgEAgNcqKbNr+Kx0lZQZpbaL1q2XNrI0j0Nf+YiNLZ8fZWVlnbE8Kyur4j4AAOBaby/crnX7cxQRUksv3dJRfn5+luZxaPlISEhQbGysvv/++4plubm5Wr58ubp27erIXQEAgCrYeDBXr/9Q/pEX42/uoOjwYIsT1WDscuLECW3btq3i5507d+rXX39VZGSkmjRpoieeeEIvvPCCWrVqpYSEBI0ZM0ZxcXEaMGCAI3MDAIALKCmza9jM8nHL9e1j1P+SOKsjSapB+Vi5cqV69uxZ8fNTTz0lSbrrrrs0Y8YMPf3008rPz9f999+v48eP68orr9T8+fMVHGx90wIAwJdM+3GbNhzMVd3atfTiLYmWj1tO8zPGGKtD/FZubq4iIiKUk5Oj8PBwq+MAAOCR1u/P0YBpP6vUbvS3Oy5R/0uce5FpdX5/890uAAB4meLS8ne3lNqNbuwQq5uT3WPcchrlAwAAL/PmD1u1KTNPkaFBesGNxi2nUT4AAPAi6/blaNrC7ZKk5/snKqqO+32WFuUDAAAvUVRapmGzflWZ3ahPUkP1SWpodaRKUT4AAPASf/tuq7ZknVBUnSA93z/R6jjnRPkAAMALpO89rrd/Kh+3vDAgUZGhQRYnOjfKBwAAHq6wpEzDZqXLbqSbk+N0Y6J7jltOo3wAAODhXv1ui7Zln1BUHZvG39zB6jgXRPkAAMCDrd5zTO8u2iFJeumWRNVz43HLaZQPAAA8VGFJmYafGrfc2qmRbujgGd8gT/kAAMBDvfzvzdpxKF/RYTaN6+f+45bTKB8AAHiglbuO6v8W75Qkpd3aURG1a1mcqOooHwAAeJiTxeXjFmOkQZc11nXtYqyOVC2UDwAAPMyUBZu160iBYsODNaZve6vjVBvlAwAAD7J8xxFNX1I+bpk4sKMiQjxn3HIa5QMAAA9RUFyqEbPXyhjpjsvj1aNNtNWRaoTyAQCAh5j0zSbtOVqguIhg/bVPO6vj1BjlAwAAD7B0+xG9t3S3JGnSoCSFBXveuOU0ygcAAG4uv6hUI2anS5L+kNJEV7VqYHGii0P5AADAzaV9s1H7jp1Uo7oheuYmzx23nEb5AADAjS3eeljvL9sjSZoyKEl1bIEWJ7p4lA8AANxUXmGJRn66VpL0xyuaqlvLKIsTOQblAwAAN/XS15u0//hJxUeGaFTvtlbHcRjKBwAAbmjRlkP66JfyccvkgckK9YJxy2mUDwAA3Ezub8Ytf+7WTF1b1Lc4kWNRPgAAcDMvfLlBB3MK1bR+bT19Yxur4zgc5QMAADfy46ZszVy5T35+0pRByaod5D3jltMoHwAAuImcghKN+qx83HJP9wR1SYi0OJFzUD4AAHATE77coKzcIiVEhWr4Dd43bjmN8gEAgBv4fmOWPl1dPm6ZOjhJIUEBVkdyGsoHAAAWO15QrNGfrZMk3XdVc13W1DvHLadRPgAAsNj4LzYoO69IzRuE6qnrW1sdx+koHwAAWGhBRqbmrNkvfz/p5cHJCq7lveOW0ygfAABY5Gh+sf46p3zccv/VLdSpST2LE7kG5QMAAIuMm5ehwyeK1Sq6jp5IbWV1HJehfAAAYIGv1x3UF+kHFODvp6k+Mm45jfIBAICLHTlRpDFz10uSHrqmhZLj61obyMUoHwAAuNjYzzN0JL9YbWLC9Oh1La2O43KUDwAAXOjLtQf01bqDCvD308u3JcsW6DvjltMoHwAAuMihvP+OW4b2bKnERhEWJ7IG5QMAABcwxujZuet0rKBE7RqG65GevjduOY3yAQCAC8xLP6AFGVkK9PfTy4OTFRTou7+CfffIAQBwkezcQo2blyFJeuy6VmofF25xImtRPgAAcCJjjJ6Zs07HC0qU2ChcD/VoYXUky1E+AABwojlr9uu7jdmqFVD+YWK1AvjVy58AAABOkpVbqOdOjVueSG2ttrG+PW45jfIBAIATGGM0+rN1yi0sVVLjCD1wdXOrI7kNygcAAE4we9U+/bApW0EB/po6OFmBjFsq8CcBAICDHcw5qQlfbJAkPXl9a7WOCbM4kXuhfAAA4EDGGI38dJ3yikp1SXxd3XdVgtWR3A7lAwAAB5q5cq8WbTmkoEDGLefCnwgAAA6y//hJPf/lRknS8Btaq2V0HYsTuSfKBwAADmCM0cjZa3WiqFSXNqmre6/k3S3nQvkAAMABPvxljxZvOyzbqXFLgL+f1ZHcFuUDAICLtPdogV76qnzcMqJXGzVvwLjlfCgfAABcBLvdaOSna5VfXKbLm9XT3d15d8uFOLx8lJWVacyYMUpISFBISIhatGih559/XsYYR+8KAADLvb98t5ZsP6LgWv6aMohxS1UEOnqDkyZN0ltvvaX33ntPHTp00MqVK3X33XcrIiJCjz32mKN3BwCAZfYcKVDa15skSaNubKtmUaEWJ/IMDi8fS5YsUf/+/dWnTx9JUrNmzfTRRx/pl19+cfSuAACwjN1uNGJ2uk6WlCklIVJ/6trM6kgew+Fjl27duun777/Xli1bJEnp6elavHixevfuXen6RUVFys3NPeMGAIC7+9fSXVq+86hqBwVoyqBk+TNuqTKHv/IxatQo5ebmqm3btgoICFBZWZlefPFFDRkypNL109LSNH78eEfHAADAaXYeztfE+eXjltG926pJ/doWJ/IsDn/lY+bMmfrggw/04YcfavXq1Xrvvfc0depUvffee5WuP3r0aOXk5FTc9u7d6+hIAAA4TJnd6OnZ6Sossatbi/oaktLU6kgex+GvfIwYMUKjRo3SHXfcIUnq2LGjdu/erbS0NN11111nrW+z2WSz2RwdAwAAp5j+806t2HVMoUEBmjQwiXFLDTj8lY+CggL5+5+52YCAANntdkfvCgAAl9px6ISmLNgsSfprn/aKj2TcUhMOf+WjX79+evHFF9WkSRN16NBBa9as0SuvvKJ77rnH0bsCAMBlyuxGw2elq6jUrqtaRenOLvFWR/JYDi8fb7zxhsaMGaOHH35Y2dnZiouL0wMPPKCxY8c6elcAALjMPxbv0Oo9xxVmC9TEgUny82PcUlN+xs0+ejQ3N1cRERHKyclReHi41XEAANC27Dzd9PpiFZfaNWlgR91+eROrI7md6vz+5rtdAAA4j9Iyu4bNWqviUruuad1At3Vm3HKxKB8AAJzHu//ZqfS9xxUWHKiJAzsybnEAygcAAOewJStPr35b/ond4/p1UMOIEIsTeQfKBwAAlSgps2vYzHQVl9l1bdtoDby0kdWRvAblAwCASrzz03at25+j8OBApd3KuMWRKB8AAPzOpsxc/e37rZKk8f07KCY82OJE3oXyAQDAb5wet5SUGV3fPkYDLmHc4miUDwAAfuN/f9yujAO5qlu7ll68JZFxixNQPgAAOCXjQI7e+OHUuOXmDooOY9ziDJQPAAAkFZfaNXzWWpXajW7sEKubk+OsjuS1KB8AAEh688dt2ngwV5GhQXqBcYtTUT4AAD5v/f4cTftxmyRpQv8OiqpjsziRd6N8AAB8WlFpmYbNTFeZ3ahPx4bqm8S4xdkoHwAAn/bG99u0OStP9UODNKF/B6vj+ATKBwDAZ6XvPa63ftouSXphQKLqM25xCcoHAMAnFZaUafis8nFLv+Q49e7Y0OpIPoPyAQDwSa99t1Vbs08oqo5NE25m3OJKlA8AgM9ZveeY/r6ofNzy0i2JqhcaZHEi30L5AAD4lNPjFruRbunUSDd0iLU6ks+hfAAAfMor327RjkP5ahBm07h+7a2O45MoHwAAn7Fq91G9+58dkqS0Wzqqbm3GLVagfAAAfMLJ4jINn7VWxki3XtpIqe1jrI7ksygfAACfMGXBZu08nK+YcJvG9ePdLVaifAAAvN4vO49q+pKdkqSJA5MUEVLL4kS+jfIBAPBqBcWlGjE7XcZIt3VurJ5toq2O5PMoHwAArzZ5/mbtPlKghhHBerYv725xB5QPAIDXWrr9iGYs2SVJmjQwSeHBjFvcAeUDAOCV8otK9fSn6ZKkO7vE6+rWDSxOhNMoHwAArzTxm03ae/SkGtUN0V/7MG5xJ5QPAIDX+XnbYf2/ZbslSZMHJamOLdDiRPgtygcAwKucKCrV07PXSpL+54om6t4yyuJE+D3KBwDAq7z09UbtP35SjeuFaHTvdlbHQSUoHwAAr7FoyyF9uHyPJGnKoGSFMm5xS5QPAIBXyC0s0ahPy8ctd3Vtqq4t6lucCOdC+QAAeIUXv9yoAzmFahJZWyN7t7U6Ds6D8gEA8HgLN2frk5V75ecnTR2crNpBjFvcGeUDAODRck6WaNSn6yRJd3dLUJeESIsT4UIoHwAAj/b8lxuUmVuohKhQjejVxuo4qALKBwDAY/2wKUuzV+2Tn580ZVCSQoICrI6EKqB8AAA8Uk7Bf8ctf7kyQZ2bMW7xFJQPAIBHGv9FhrLzitS8QaiG3cC4xZNQPgAAHuffGZn6bM1++Z96d0twLcYtnoTyAQDwKMfyi/XMnPWSpPuubq5Lm9SzOBGqi/IBAPAo4+Zl6PCJIrWMrqMnU1tbHQc1QPkAAHiM+esPal76AQX4++llxi0ei/IBAPAIR04U6a+nxi0PXtNcyfF1rQ2EGqN8AAA8wth5GTqSX6w2MWF67LpWVsfBRaB8AADc3ldrD+qrtQfLxy23JcsWyLjFk1E+AABu7VBekZ6dW/5hYkN7tFBiowiLE+FiUT4AAG7LGKMxc9frWEGJ2saG6ZFrGbd4A8oHAMBtfbH2oOZnZCrw1LglKJBfW96AswgAcEvZeYUa+3n5u1sevbaVOsQxbvEWlA8AgNsxxuivc9breEGJOsSF6+GeLayOBAeifAAA3M7cX/fr2w1ZqhXgp6mDk1UrgF9X3oSzCQBwK1m5hXpu3gZJ0uPXtVK7huEWJ4KjUT4AAG7DGKNnPlunnJMl6tgoQg9cw7jFGzmlfOzfv1//8z//o/r16yskJEQdO3bUypUrnbErAIAX+XT1fn2/KVtBAf56+TbGLd4q0NEbPHbsmLp3766ePXvqm2++UYMGDbR161bVq8dXHgMAzi0zp1Djv8iQJD1xfSu1jgmzOBGcxeHlY9KkSYqPj9f06dMrliUkJDh6NwAAL2KM0ajP1iqvsFTJ8XV1/1XNrY4EJ3L461nz5s1T586dNXjwYEVHR6tTp0569913z7l+UVGRcnNzz7gBAHzLrJX7tHDzIQUF+uvlwUkKZNzi1Rx+dnfs2KG33npLrVq10oIFC/TQQw/pscce03vvvVfp+mlpaYqIiKi4xcfHOzoSAMCNHTh+Us9/Wf7ulmHXt1bLaMYt3s7PGGMcucGgoCB17txZS5YsqVj22GOPacWKFVq6dOlZ6xcVFamoqKji59zcXMXHxysnJ0fh4by9CgC8mTFGf/rnL/rP1sPq1KSuZj/YTQH+flbHQg3k5uYqIiKiSr+/Hf7KR8OGDdW+ffszlrVr10579uypdH2bzabw8PAzbgAA3/Dxir36z9bDsgX6a+rgZIqHj3B4+ejevbs2b958xrItW7aoadOmjt4VAMCD7TtWoBdOjVtG9GqjFg3qWJwIruLw8vHkk09q2bJleumll7Rt2zZ9+OGH+vvf/66hQ4c6elcAAA9ljNHIT9cqv7hMlzerp7u7865IX+Lw8nH55Zdrzpw5+uijj5SYmKjnn39er732moYMGeLoXQEAPNQHy/fo521HFFzLX1MGMW7xNQ7/nA9J6tu3r/r27euMTQMAPNzeowV66euNkqSRN7ZVs6hQixPB1XgjNQDAZex2oxGz01VQXKYuCZG6q2szqyPBApQPAIDL/L9lu7Vsx1HVDgrQ1EHJ8mfc4pMoHwAAl9h9JF8Tv9kkSRrVu62a1K9tcSJYhfIBAHA6u91oxKy1OllSpq7N6+t/Uvj4BV9G+QAAON2MJbv0y66jCg0K0ORBSYxbfBzlAwDgVDsP52vygvJxyzN92ik+knGLr6N8AACcpsxuNGJWugpL7LqyZZT+0KWJ1ZHgBigfAACnmf7zTq3cfUx1bIGaOLCj/PwYt4DyAQBwkm3ZJzRlQfl3fT3bp50a12PcgnKUDwCAw5XZjYbPSldRqV1Xt26g2y+PtzoS3AjlAwDgcO/+Z4d+3XtcYbZATWLcgt+hfAAAHGprVp5e+fcWSdKYfu3VMCLE4kRwN5QPAIDDlJbZNXxWuorL7OrZpoEGX9bY6khwQ5QPAIDDvLNoh9L35Sg8OFBptyYxbkGlKB8AAIfYnJmnv323VZL03M0dFBsRbHEiuCvKBwDgopWU2TVs1q8qLrMrtV20bunUyOpIcGOUDwDARXtr4Xat35+riJBaeukW3t2C86N8AAAuSsaBHL3+ffm4ZUL/DooOZ9yC86N8AABqrLjUruGz1qrUbtSrQ4xuTo6zOhI8AOUDAFBj037cpo0Hc1Wvdi29MIBxC6qG8gEAqJH1+3M07cdtkqQJ/RPVIMxmcSJ4CsoHAKDaikrLNHxWukrtRr0TY9U3qaHVkeBBKB8AgGp74/tt2pSZp/qhQXphQCLjFlQL5QMAUC1r9x3XWz9tlyS9MCBR9eswbkH1UD4AAFVWVFqmYTPTVWY36pvUUL07Mm5B9VE+AABV9tp3W7U1+4Si6gRpQv9Eq+PAQ1E+AABVsmbPMb1TMW7pqMjQIIsTwVNRPgAAF1RYUv7uFruRBlwSpxsTY62OBA9G+QAAXNCr327R9kP5ahBm03M3d7A6Djwc5QMAcF6rdh/V3/+zQ5KUdktH1a3NuAUXh/IBADink8VlGj5rrYyRbr20kVLbx1gdCV6A8gEAOKep/96snYfzFRNu07i+jFvgGJQPAEClVuw6qn/+vFOSNPHWJEXUrmVxIngLygcA4CwFxaUaMStdxkiDL2usnm2jrY4EL0L5AACcZfL8zdp1pEANI4L1bN/2VseBl6F8AADOsGzHEc1YskuSNHFgkiJCGLfAsSgfAIAK+UWlenr2WknSnV3idU3rBhYngjeifAAAKkyav0l7jhaoUd0QPXNTO6vjwEtRPgAAkqQl2w7rX0t3S5ImDUxSWDDjFjgH5QMAoBNFpXr60/Jxy5CUJrqyVZTFieDNKB8AAKV9vVH7jp1U43ohGs24BU5G+QAAH7d462F9sHyPJGnyoCTVsQVanAjejvIBAD4sr7BET89OlyT98Yqm6taCcQucj/IBAD7sxa826kBOoZpE1tao3m2tjgMfQfkAAB+1cHO2Pl6xV5I0ZVCSQhm3wEUoHwDgg3JOlmjUp+skSXd3b6aU5vUtTgRfQvkAAB/0wpcblJlbqGb1a+vpXoxb4FqUDwDwMT9sytKsVfvk5ydNHZyskKAAqyPBx1A+AMCH5BT8d9xyb/cEdW4WaXEi+CLKBwD4kPFfZCg7r0jNo0I1vFcbq+PAR1E+AMBHfLshS5+t2S9/P2nqbckKrsW4BdagfACADziWX6xn5pSPW+67urkubVLP4kTwZZQPAPABz32RoUN5RWoZXUdPpra2Og58HOUDALzc/PWZ+vzXA+XjlsGMW2A9ygcAeLGj+cV6dm75uOXBa1rokvi61gYCRPkAAK829vP1OnyiWK1j6ujx1FZWxwEkuaB8TJw4UX5+fnriiSecvSsAwG98ve6gvlx7UAH+fpo6OFm2QMYtcA9OLR8rVqzQO++8o6SkJGfuBgDwO4dPFOnZueslSQ/3aKGkxnWtDQT8htPKx4kTJzRkyBC9++67qlePt3QBgKsYYzRm7nodzS9W29gwPXot4xa4F6eVj6FDh6pPnz5KTU0973pFRUXKzc094wYAqLkv1x7UN+szFXhq3BIUyOV9cC+Bztjoxx9/rNWrV2vFihUXXDctLU3jx493RgwA8DmH8oo09vPyccvQni2V2CjC4kTA2Rxeh/fu3avHH39cH3zwgYKDgy+4/ujRo5WTk1Nx27t3r6MjAYBPMMbo2bnrdKygRO0bhmtoz5ZWRwIq5fBXPlatWqXs7GxdeumlFcvKysq0aNEivfnmmyoqKlJAwH+vuLbZbLLZbI6OAQA+Z176AS3IyFKtAMYtcG8OLx/XXXed1q1bd8ayu+++W23bttXIkSPPKB4AAMfIzi3U2M8zJEmPXdtK7ePCLU4EnJvDy0dYWJgSExPPWBYaGqr69euftRwAcPGMMXpmzjrlnCxRx0YRerBHC6sjAefFa3IA4OE+W71f323MVlCAv6YOTlatAP5ph3tzyrtdfm/hwoWu2A0A+JzMnEI990X5uOXx1FZqExtmcSLgwqjHAOChjDEa/dla5RWWKrlxhB64urnVkYAqoXwAgIeatWqfftx8SEGB5eOWQMYt8BD8TQUAD3Tg+Ek9/8UGSdKw61urVQzjFngOygcAeBhjjEZ9tk55RaXq1KSu/nIV4xZ4FsoHAHiYT1bs1aIth2Q7NW4J8PezOhJQLZQPAPAg+44V6IWvNkqSRvRqoxYN6licCKg+ygcAeAhjjEZ+ulYnikrVuWk93d09wepIQI1QPgDAQ3ywfI9+3nZEwbX8NYVxCzwY5QMAPMDeowV66evyccvTvdoqISrU4kRAzVE+AMDN2e1GT89eq4LiMnVpFqk/d2tmdSTgolA+AMDNvb98t5buOKKQWgGaMjhJ/oxb4OEoHwDgxnYfyVfa15skSaN6t1XT+oxb4PkoHwDgpux2oxGz1+pkSZmuaB6pP17R1OpIgENQPgDATb23dJd+2XlUtYMCNGVQMuMWeA3KBwC4oZ2H8zVpfvm45Zmb2ik+srbFiQDHoXwAgJspsxuNmJWuwhK7uresryEpTayOBDgU5QMA3Mz0n3dq5e5jCg0K0KSBSfLzY9wC70L5AAA3sv3QCU1ZsFmS9Gzf9mpcj3ELvA/lAwDcRJndaPisdBWV2nVVqyjdcXm81ZEAp6B8AICb+L//7NCaPccVZgtk3AKvRvkAADewNStPL3+7RZI0pm97xdUNsTgR4DyUDwCwWGmZXcNnpau41K4ebRpocOfGVkcCnIryAQAWe2fRDqXvy1FYcKAm3sq4Bd6P8gEAFtqcmafXvisftzzXr4NiI4ItTgQ4H+UDACxScmrcUlJmdF3baN16aSOrIwEuQfkAAIu8vXC71u3PUURILb10a0fGLfAZlA8AsMCGA7l6/YetkqTxN3dQTDjjFvgOygcAuNhvxy03tI9R/0virI4EuBTlAwBcbNqP27ThYK7q1a6lF29h3ALfQ/kAABdavz9Hb/6wTZI0vn+iGoTZLE4EuB7lAwBcpLi0fNxSajfqnRirfkkNrY4EWILyAQAu8sYPW7UpM0+RoUF6fkAi4xb4LMoHALjAun05+t+F2yVJz/dPVFQdxi3wXZQPAHCyotIyDZv1q8rsRn2TGqoP4xb4OMoHADjZ377bqi1ZJxRVJ0gT+idaHQewHOUDAJwofe9xvf1T+bjlhQEdFRkaZHEiwHqUDwBwksKSMg2blS67kW5OjtONibFWRwLcAuUDAJzk1e+2aFv2CTUIs2n8zR2sjgO4DcoHADjBqt3H9O6iHZKkl27pqHqMW4AKlA8AcLDCkjKNODVuubVTI13fPsbqSIBboXwAgIO9/O/N2nE4X9FhNo3rx7gF+D3KBwA40MpdR/V/i3dKkiYO7KiI2rUsTgS4H8oHADjIyeIyjZi9VsZIgy5rrGvbMm4BKkP5AAAHmbxgk3YezldseLDG9G1vdRzAbVE+AMABlu84ohlLdkk6NW4JYdwCnAvlAwAuUkFxacW45Y7L49WjTbTVkQC3RvkAgIs06ZtN2nO0QHERwfprn3ZWxwHcHuUDAC7Cku2H9d7S3ZKkSYOSFBbMuAW4EMoHANTQiaJSPT17rSTpDylNdFWrBhYnAjwD5QMAaijt643ad+ykGtUN0TM3MW4BqoryAQA1sHjrYX2wfI8kacqgJNWxBVqcCPAclA8AqKa8whKN/LR83PKnrk3VrWWUxYkAz0L5AIBqeunrjdp//KTiI0M08sa2VscBPA7lAwCq4acth/TRL3slSVMGJSuUcQtQbZQPAKii3MISjTo1bvlzt2a6onl9ixMBnonyAQBV9MKXG3Qwp1BN69fW0ze2sToO4LEcXj7S0tJ0+eWXKywsTNHR0RowYIA2b97s6N0AgEv9uClbM1fuk59f+bildhDjFqCmHF4+fvrpJw0dOlTLli3Tt99+q5KSEt1www3Kz8939K4AwCVyCko06rPyccs93RPUJSHS4kSAZ3N4dZ8/f/4ZP8+YMUPR0dFatWqVrr76akfvDgCcbsKXG5SVW6TmUaEafgPjFuBiOf11w5ycHElSZGTl/6dQVFSkoqKiip9zc3OdHQkAquy7DVn6dPWpccvgJIUEBVgdCfB4Tr3g1G6364knnlD37t2VmJhY6TppaWmKiIiouMXHxzszEgBU2fGCYo2es06SdN9VzXVZU8YtgCM4tXwMHTpU69ev18cff3zOdUaPHq2cnJyK2969e50ZCQCq7Ll5GTqUV6QWDUL11PWtrY4DeA2njV0eeeQRffnll1q0aJEaN258zvVsNptsNpuzYgBAjSzIyNTcXw/I30+aOjhZwbUYtwCO4vDyYYzRo48+qjlz5mjhwoVKSEhw9C4AwKmO5hfrr6fGLfdf3UKdmtSzOBHgXRxePoYOHaoPP/xQn3/+ucLCwpSZmSlJioiIUEhIiKN3BwAON25ehg6fKFar6Dp6IrWV1XEAr+Pwaz7eeust5eTkqEePHmrYsGHF7ZNPPnH0rgDA4b5ed1BfpB9QgL8f4xbASZwydgEAT3TkRJHGzF0vSXrwmuZKjq9rbSDAS/HdLgBwytjPM3Qkv1htYsL02HWMWwBnoXwAgKQv1x7QV+sOKsDfTy/flixbIOMWwFkoHwB83qG8/45bhvZsqcRGERYnArwb5QOATzPG6Nm563SsoETtGobrkZ4trY4EeD3KBwCfNi/9gBZkZCnQ308vD05WUCD/LALOxrMMgM/Kzi3U2M8zJEmPXttK7ePCLU4E+AbKBwCfZIzRM3PWK+dkiTrEhevhni2sjgT4DMoHAJ80Z81+fbcxS7UCyt/dUiuAfw4BV+HZBsDnZOUW6rl55eOWJ1Jbq20s4xbAlSgfAHyKMUajP1un3MJSJTWO0ANXN7c6EuBzKB8AfMrsVfv0w6ZsBQX46+XByQpk3AK4HM86AD7jYM5JTfhigyTpyetbq1VMmMWJAN9E+QDgE4wxGvXpOuUVleqS+Lq676oEqyMBPovyAcAnzFy5Vz9tOaSgQH9NZdwCWIpnHwCvt//4ST3/5UZJ0ogb2qhldB2LEwG+jfIBwKuVj1vW6kRRqS5rWk/3XMm4BbAa5QOAV/vwlz36z9bDsgX6a8qgJAX4+1kdCfB5lA8AXmvv0QK99FX5uOXpG9uqeQPGLYA7oHwA8Ep2u9HIT9cqv7hMXZpF6u5uzayOBOAUygcAr/TB8t1asv2IQmoFaPKgJPkzbgHcBuUDgNfZc6RAL329SZI0qndbNYsKtTgRgN+ifADwKna70fDZ6TpZUqYrmkfqj1c0tToSgN+hfADwKv9auku/7Dyq2kEBmjwwmXEL4IYoHwC8xq7D+Zo4v3zcMvqmdmpSv7bFiQBUhvIBwCuU2Y2Gz0pXYYld3VvW15AuTayOBOAcKB8AvML0n3dq5e5jCg0K0KSBvLsFcGeUDwAeb8ehE5qyYLMk6a992qtxPcYtgDujfADwaKfHLUWldl3VKkp3dom3OhKAC6B8APBo/1i8Q6v3HFeYLVCTBibJz49xC+DuKB8APNa27DxN/fcWSdKzfdsprm6IxYkAVAXlA4BHKi2za9istSouteua1g10W2fGLYCnoHwA8Eh//88Ope89rrDgQE0c2JFxC+BBKB8APM6WrDy99u1WSdK4fh3UMIJxC+BJKB8APEpJmV3DZqaruMyua9tGa+CljayOBKCaKB8APMo7P23Xuv05Cg8OVNqtjFsAT0T5AOAxNh7M1d++Lx+3jO/fQTHhwRYnAlATlA8AHqGkzK7hs9JVUmZ0ffsYDbiEcQvgqSgfADzC//64XRkHclW3di29eEsi4xbAg1E+ALi9jAM5euOHU+OWmzsoOoxxC+DJKB8A3Fpxafm7W0rtRjd2iNXNyXFWRwJwkSgfANzamz9s1abMPEWGBukFxi2AV6B8AHBb6/fnaNrC7ZKk5/snKqqOzeJEAByB8gHALRWVlmnYzHSV2Y36dGyoPkkNrY4EwEEoHwDc0uvfb9XmrDzVDw3ShP4drI4DwIEoHwDcTvre43rr1LjlhQGJqs+4BfAqlA8AbqWwpEzDZqXLbqSbk+PUuyPjFsDbUD4AuJXXvtuqbdknFFXHpvE3M24BvBHlA4DbWL3nmP6+qHzc8tItiaoXGmRxIgDOQPkA4BYKS8o0/NS45ZZOjXRDh1irIwFwEsoHALfw8r83a8ehfEWH2TSuX3ur4wBwIsoHAMut2n1U/7d4pyQp7daOqlubcQvgzSgfACx1srhMw2etlTHSoMsa67p2MVZHAuBklA8AlpqyYLN2Hs5XbHiwxvRl3AL4AsoHAMv8svOopi85NW4Z2FERIbUsTgTAFSgfACxRUFyqEbPTZYx0e+d49WwTbXUkAC5C+QBgicnzN2v3kQLFRQTrr33bWR0HgAsFWh0AgG8xxmjmyr2asWSXJGniwCSFBzNuAXyJ0175mDZtmpo1a6bg4GClpKTol19+cdauAHiI/KJSPTUzXSM/XSdJ+uMVTXV16wYWpwLgak4pH5988omeeuopjRs3TqtXr1ZycrJ69eql7OxsZ+wOgAfYeDBX/d5YrDlr9ivA308jerXhu1sAH+VnjDGO3mhKSoouv/xyvfnmm5Iku92u+Ph4Pfrooxo1atQZ6xYVFamoqKji59zcXMXHxysnJ0fh4eEOy3T4RJGm/bjNYdsDUHVFpXbNXrVPxaV2xYYH640/dNLlzSKtjgXAgXJzcxUREVGl398Ov+ajuLhYq1at0ujRoyuW+fv7KzU1VUuXLj1r/bS0NI0fP97RMc6Se7JE03/e5fT9ADi3a9tGa+rgZEXyhXGAT3N4+Th8+LDKysoUE3PmpxTGxMRo06ZNZ60/evRoPfXUUxU/n37lw9Hq1g7S0J4tHL5dAFXTKjpMNyfHyd/fz+ooACxm+btdbDabbDab0/cTGRqkEb3aOn0/AADg/Bx+wWlUVJQCAgKUlZV1xvKsrCzFxvIV2QAA+DqHl4+goCBddtll+v777yuW2e12ff/99+rataujdwcAADyMU8YuTz31lO666y517txZXbp00Wuvvab8/HzdfffdztgdAADwIE4pH7fffrsOHTqksWPHKjMzU5dcconmz59/1kWoAADA9zjlcz4uRnXeJwwAANxDdX5/88VyAADApSgfAADApSgfAADApSgfAADApSgfAADApSgfAADApSgfAADApSgfAADApSz/VtvfO/2ZZ7m5uRYnAQAAVXX693ZVPrvU7cpHXl6eJCk+Pt7iJAAAoLry8vIUERFx3nXc7uPV7Xa7Dhw4oLCwMPn5+Tl027m5uYqPj9fevXu98qPbvf34JO8/Ro7P83n7MXJ8ns9Zx2iMUV5enuLi4uTvf/6rOtzulQ9/f381btzYqfsIDw/32r9Ukvcfn+T9x8jxeT5vP0aOz/M54xgv9IrHaVxwCgAAXIryAQAAXMqnyofNZtO4ceNks9msjuIU3n58kvcfI8fn+bz9GDk+z+cOx+h2F5wCAADv5lOvfAAAAOtRPgAAgEtRPgAAgEtRPgAAgEtRPgAAgEt5Vfl48cUX1a1bN9WuXVt169atdJ09e/aoT58+ql27tqKjozVixAiVlpaed7tHjx7VkCFDFB4errp16+ree+/ViRMnnHAE1bNw4UL5+flVeluxYsU5H9ejR4+z1n/wwQddmLzqmjVrdlbWiRMnnvcxhYWFGjp0qOrXr686depo4MCBysrKclHi6tm1a5fuvfdeJSQkKCQkRC1atNC4ceNUXFx83se58zmcNm2amjVrpuDgYKWkpOiXX3457/qzZs1S27ZtFRwcrI4dO+rrr792UdLqS0tL0+WXX66wsDBFR0drwIAB2rx583kfM2PGjLPOVXBwsIsSV89zzz13Vta2bdue9zGedP6kyv9N8fPz09ChQytd393P36JFi9SvXz/FxcXJz89Pc+fOPeN+Y4zGjh2rhg0bKiQkRKmpqdq6desFt1vd53F1eVX5KC4u1uDBg/XQQw9Ven9ZWZn69Omj4uJiLVmyRO+9955mzJihsWPHnne7Q4YMUUZGhr799lt9+eWXWrRoke6//35nHEK1dOvWTQcPHjzj9pe//EUJCQnq3LnzeR973333nfG4yZMnuyh19U2YMOGMrI8++uh513/yySf1xRdfaNasWfrpp5904MAB3XrrrS5KWz2bNm2S3W7XO++8o4yMDL366qt6++239cwzz1zwse54Dj/55BM99dRTGjdunFavXq3k5GT16tVL2dnZla6/ZMkS3Xnnnbr33nu1Zs0aDRgwQAMGDND69etdnLxqfvrpJw0dOlTLli3Tt99+q5KSEt1www3Kz88/7+PCw8PPOFe7d+92UeLq69ChwxlZFy9efM51Pe38SdKKFSvOOL5vv/1WkjR48OBzPsadz19+fr6Sk5M1bdq0Su+fPHmyXn/9db399ttavny5QkND1atXLxUWFp5zm9V9HteI8ULTp083ERERZy3/+uuvjb+/v8nMzKxY9tZbb5nw8HBTVFRU6bY2bNhgJJkVK1ZULPvmm2+Mn5+f2b9/v8OzX4zi4mLToEEDM2HChPOud80115jHH3/cNaEuUtOmTc2rr75a5fWPHz9uatWqZWbNmlWxbOPGjUaSWbp0qRMSOt7kyZNNQkLCeddx13PYpUsXM3To0Iqfy8rKTFxcnElLS6t0/dtuu8306dPnjGUpKSnmgQcecGpOR8nOzjaSzE8//XTOdc7175E7GjdunElOTq7y+p5+/owx5vHHHzctWrQwdru90vs96fxJMnPmzKn42W63m9jYWDNlypSKZcePHzc2m8189NFH59xOdZ/HNeFVr3xcyNKlS9WxY0fFxMRULOvVq5dyc3OVkZFxzsfUrVv3jFcSUlNT5e/vr+XLlzs9c3XMmzdPR44c0d13333BdT/44ANFRUUpMTFRo0ePVkFBgQsS1szEiRNVv359derUSVOmTDnvmGzVqlUqKSlRampqxbK2bduqSZMmWrp0qSviXrScnBxFRkZecD13O4fFxcVatWrVGX/2/v7+Sk1NPeef/dKlS89YXyp/TnrSuZJ0wfN14sQJNW3aVPHx8erfv/85/71xB1u3blVcXJyaN2+uIUOGaM+ePedc19PPX3Fxsd5//33dc8895/0WdU86f7+1c+dOZWZmnnGOIiIilJKScs5zVJPncU243bfaOlNmZuYZxUNSxc+ZmZnnfEx0dPQZywIDAxUZGXnOx1jlH//4h3r16nXBbwX+wx/+oKZNmyouLk5r167VyJEjtXnzZn322WcuSlp1jz32mC699FJFRkZqyZIlGj16tA4ePKhXXnml0vUzMzMVFBR01jU/MTExbne+KrNt2za98cYbmjp16nnXc8dzePjwYZWVlVX6HNu0aVOljznXc9ITzpXdbtcTTzyh7t27KzEx8ZzrtWnTRv/85z+VlJSknJwcTZ06Vd26dVNGRobTv8G7ulJSUjRjxgy1adNGBw8e1Pjx43XVVVdp/fr1CgsLO2t9Tz5/kjR37lwdP35cf/7zn8+5jiedv987fR6qc45q8jyuCbcvH6NGjdKkSZPOu87GjRsveFGUJ6nJMe/bt08LFizQzJkzL7j9316v0rFjRzVs2FDXXXedtm/frhYtWtQ8eBVV5/ieeuqpimVJSUkKCgrSAw88oLS0NLf+7oWanMP9+/frxhtv1ODBg3Xfffed97FWn0NIQ4cO1fr16897TYQkde3aVV27dq34uVu3bmrXrp3eeecdPf/8886OWS29e/eu+O+kpCSlpKSoadOmmjlzpu69914LkznHP/7xD/Xu3VtxcXHnXMeTzp8ncfvyMWzYsPO2Uklq3rx5lbYVGxt71hW7p98FERsbe87H/P4im9LSUh09evScj7lYNTnm6dOnq379+rr55purvb+UlBRJ5f/X7YpfXBdzTlNSUlRaWqpdu3apTZs2Z90fGxur4uJiHT9+/IxXP7Kyspx2vipT3WM8cOCAevbsqW7duunvf/97tffn6nNYmaioKAUEBJz1zqLz/dnHxsZWa3138cgjj1RcfF7d//utVauWOnXqpG3btjkpnePUrVtXrVu3PmdWTz1/krR7925999131X610JPO3+nzkJWVpYYNG1Ysz8rK0iWXXFLpY2ryPK4Rh1094kYudMFpVlZWxbJ33nnHhIeHm8LCwkq3dfqC05UrV1YsW7BggVtdcGq3201CQoIZNmxYjR6/ePFiI8mkp6c7OJnjvf/++8bf398cPXq00vtPX3A6e/bsimWbNm1y6wtO9+3bZ1q1amXuuOMOU1paWqNtuMs57NKli3nkkUcqfi4rKzONGjU67wWnffv2PWNZ165d3faCRbvdboYOHWri4uLMli1barSN0tJS06ZNG/Pkk086OJ3j5eXlmXr16pm//e1vld7vaefvt8aNG2diY2NNSUlJtR7nzudP57jgdOrUqRXLcnJyqnTBaXWexzXK6rAtuYHdu3ebNWvWmPHjx5s6deqYNWvWmDVr1pi8vDxjTPlfmsTERHPDDTeYX3/91cyfP980aNDAjB49umIby5cvN23atDH79u2rWHbjjTeaTp06meXLl5vFixebVq1amTvvvNPlx3cu3333nZFkNm7ceNZ9+/btM23atDHLly83xhizbds2M2HCBLNy5Uqzc+dO8/nnn5vmzZubq6++2tWxL2jJkiXm1VdfNb/++qvZvn27ef/9902DBg3Mn/70p4p1fn98xhjz4IMPmiZNmpgffvjBrFy50nTt2tV07drVikO4oH379pmWLVua6667zuzbt88cPHiw4vbbdTzlHH788cfGZrOZGTNmmA0bNpj777/f1K1bt+IdZn/84x/NqFGjKtb/+eefTWBgoJk6darZuHGjGTdunKlVq5ZZt26dVYdwXg899JCJiIgwCxcuPONcFRQUVKzz+2McP368WbBggdm+fbtZtWqVueOOO0xwcLDJyMiw4hDOa9iwYWbhwoVm586d5ueffzapqakmKirKZGdnG2M8//ydVlZWZpo0aWJGjhx51n2edv7y8vIqftdJMq+88opZs2aN2b17tzHGmIkTJ5q6deuazz//3Kxdu9b079/fJCQkmJMnT1Zs49prrzVvvPFGxc8Xeh47gleVj7vuustIOuv2448/Vqyza9cu07t3bxMSEmKioqLMsGHDzmi+P/74o5Fkdu7cWbHsyJEj5s477zR16tQx4eHh5u67764oNO7gzjvvNN26dav0vp07d57xZ7Bnzx5z9dVXm8jISGOz2UzLli3NiBEjTE5OjgsTV82qVatMSkqKiYiIMMHBwaZdu3bmpZdeOuNVqt8fnzHGnDx50jz88MOmXr16pnbt2uaWW24545e5O5k+fXqlf2d/+6Kkp53DN954wzRp0sQEBQWZLl26mGXLllXcd80115i77rrrjPVnzpxpWrdubYKCgkyHDh3MV1995eLEVXeuczV9+vSKdX5/jE888UTFn0dMTIy56aabzOrVq10fvgpuv/1207BhQxMUFGQaNWpkbr/9drNt27aK+z39/J22YMECI8ls3rz5rPs87fyd/p31+9vpY7Db7WbMmDEmJibG2Gw2c91115113E2bNjXjxo07Y9n5nseO4GeMMY4b4gAAAJyfT33OBwAAsB7lAwAAuBTlAwAAuBTlAwAAuBTlAwAAuBTlAwAAuBTlAwAAuBTlAwAAuBTlAwAAuBTlAwAAuBTlAwAAuNT/B4rRb/hB5a0mAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 700x100 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import torch\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "x = torch.linspace(-10, 10,100)\n",
    "relu = torch.nn.ReLU()\n",
    "y = relu(x)\n",
    "\n",
    "plt.title('ReLU')\n",
    "plt.plot(x.tolist(), y.tolist())\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "940251d6",
   "metadata": {},
   "source": [
    "## sigmoid function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "acf8bde2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiMAAAGzCAYAAAD9pBdvAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAA9hAAAPYQGoP6dpAABDHElEQVR4nO3deXxU5b3H8e/MJJksZCEEEhJCArLvypIGRUSjFBC1t1qKtiB1qZZaFdsqVkGtNS5U7VWvqFfF1npFrbULFFQEFYnKqrLvWyAJBLKQkEwy89w/koxEkpAJSU5m8nm/XvMi88xzJr/DYZIv53nOc2zGGCMAAACL2K0uAAAAtG+EEQAAYCnCCAAAsBRhBAAAWIowAgAALEUYAQAAliKMAAAASxFGAACApQgjAADAUoQRAI2Smpqq66+/3uoyGrRgwQLZbDbt3bv3jH39YX+A9oIwAkDffPONrr76aqWkpCg0NFRJSUm69NJL9cwzz1hdGoB2wMa9aYD2bdWqVRo3bpy6d++u6dOnKyEhQQcOHNDnn3+uXbt2aefOnZKk8vJy2e12BQcHW1xx/dxutyoqKuR0OmWz2Rrsm5qaqosuukgLFixoneIA1CvI6gIAWOsPf/iDoqOjtXr1asXExNR6LS8vz/u10+ls5cp853A45HA4rC4DgI8YpgHauV27dmngwIGnBRFJ6tKli/fruuZYfP311xo7dqzCwsLUrVs3Pfzww3r11VdPm7eRmpqqyy+/XCtWrNCIESMUFhamwYMHa8WKFZKkd999V4MHD1ZoaKiGDx+u9evXn1bLRx99pDFjxigiIkIxMTG68sortWXLllp96pozYozRww8/rG7duik8PFzjxo3Tpk2bfP57AtByODMCtHMpKSnKysrSxo0bNWjQoEZvl52drXHjxslms2n27NmKiIjQ//7v/9Z7BmXnzp269tpr9fOf/1w/+clPNG/ePE2ePFnz58/Xvffeq1/84heSpMzMTP3oRz/Stm3bZLdX/X/pww8/1IQJE9SzZ0898MADOnnypJ555hmdf/75WrdunVJTU+utc86cOXr44Yc1ceJETZw4UevWrdNll10ml8vV+L8kAC3LAGjX3n//feNwOIzD4TDp6enmt7/9rVm6dKlxuVy1+qWkpJjp06d7n992223GZrOZ9evXe9vy8/NNbGyskWT27NlTa1tJZtWqVd62pUuXGkkmLCzM7Nu3z9v+wgsvGElm+fLl3rZhw4aZLl26mPz8fG/bV199Zex2u5k2bZq37dVXX631vfPy8kxISIiZNGmS8Xg83n733nuvkVRrfwBYh2EaoJ279NJLlZWVpSuuuEJfffWVHn/8cY0fP15JSUn65z//We92S5YsUXp6uoYNG+Zti42N1XXXXVdn/wEDBig9Pd37PC0tTZJ08cUXq3v37qe17969W5J0+PBhbdiwQddff71iY2O9/YYMGaJLL71UixcvrrfGDz/8UC6XS7fddlutCa133HFHvdsAaH2EEQAaOXKk3n33XR0/flxffvmlZs+ereLiYl199dXavHlzndvs27dPvXr1Oq29rjZJtQKHJEVHR0uSkpOT62w/fvy49/tIUt++fU97z/79++vo0aMqKSmpt0ZJ6t27d632zp07q2PHjnVuA6D1EUYAeIWEhGjkyJF65JFH9Pzzz6uiokJvv/12s7x3fVe51NduWHUAaDcIIwDqNGLECElVwyR1SUlJ8a5Bcqq62s5GSkqKJGnbtm2nvbZ161bFxcUpIiKiwW137NhRq/3IkSPeMy8ArEcYAdq55cuX13kWomYuRl3DI5I0fvx4ZWVlacOGDd62Y8eO6a9//Wuz1te1a1cNGzZMr732mgoKCrztGzdu1Pvvv6+JEyfWu21GRoaCg4P1zDPP1NrHp59+ullrBHB2uLQXaOduu+02lZaW6gc/+IH69esnl8ulVatWaeHChUpNTdWMGTPq3O63v/2tXn/9dV166aW67bbbvJf2du/eXceOHTvjCqi+eOKJJzRhwgSlp6frhhtu8F7aGx0drQceeKDe7Tp37qxf//rXyszM1OWXX66JEydq/fr1+s9//qO4uLhmqw/A2SGMAO3cvHnz9Pbbb2vx4sV68cUX5XK51L17d/3iF7/QfffdV+diaFLVxNPly5frV7/6lR555BF17txZM2fOVEREhH71q18pNDS02WrMyMjQkiVLNHfuXM2ZM0fBwcEaO3asHnvsMfXo0aPBbR9++GGFhoZq/vz5Wr58udLS0vT+++9r0qRJzVYfgLPDvWkANKs77rhDL7zwgk6cOMHS7AAahTkjAJrs5MmTtZ7n5+frL3/5iy644AKCCIBGY5gGQJOlp6froosuUv/+/ZWbm6uXX35ZRUVFuv/++60uDYAfIYwAaLKJEyfqnXfe0YsvviibzabzzjtPL7/8si688EKrSwPgR5gzAgAALMWcEQAAYCnCCAAAsJRfzBnxeDw6dOiQIiMjm3UhJQAA0HKMMSouLlZiYqLs9vrPf/hFGDl06NBpd/YEAAD+4cCBA+rWrVu9r/tFGImMjJRUtTNRUVEWVwMAABqjqKhIycnJ3t/j9fGLMFIzNBMVFUUYAQDAz5xpigUTWAEAgKUIIwAAwFKEEQAAYCnCCAAAsBRhBAAAWIowAgAALEUYAQAAliKMAAAASxFGAACApQgjAADAUj6HkU8++USTJ09WYmKibDab3nvvvTNus2LFCp133nlyOp3q1auXFixY0IRSAQBAIPI5jJSUlGjo0KF67rnnGtV/z549mjRpksaNG6cNGzbojjvu0I033qilS5f6XCwAAAg8Pt8ob8KECZowYUKj+8+fP189evTQH//4R0lS//79tXLlSj311FMaP358nduUl5ervLzc+7yoqMjXMgEA7YDbY1Th9lQ/qr52VXrk9hhVeqraqr42cns8qjz1uTFyu6v+9FQ/9xhVfe0x8hgjYyRPTbsxMqd87TGSqe5jdGqfqtq+fU21+tR0OOVLGZ2y3XfadUp7zfvqlD6nqqPptH7mlF6nvnbDBT2UHBveiL/15tfid+3NyspSRkZGrbbx48frjjvuqHebzMxMPfjggy1cGQCgNbg9RsVlFSo8WaGik5UqKqtQcVmFTpS7VVJeqRPllSopr1Spy62TLrdKK9w66arUyQq3yio8Kq+s+rOswi1XpUfllVWBw+WuCh1oHlcMSwzcMJKTk6P4+PhabfHx8SoqKtLJkycVFhZ22jazZ8/WrFmzvM+LioqUnJzc0qUCABrJVelRblGZDheW6UhxuY4UlymvuFxHist1rMSlY6UuHS9x6ViJS0Vlla1Wl8NuU7DDpiC7XUE1f9pt1V/b5PA+qtrtdpsctqrt7Dab98+adrvNJpvNJrtNslX3q3puk01VbVV9VKvNpqo2m01Szdfffa26ZltVJ+9r3379reou3/b1vnBKn1O2/a7vNtXVJz4qtDF/xS2ixcNIUzidTjmdTqvLAIB2yxij3KJy7c0v0b78Eu3LL9W+/FIdLDipwwUndeREeZ3DBA0JC3YoKixI0WHB6uAMUoQzSJGhQYoIqfo6PMSh8BCHwkKqvg4Ldig02C5nsEOhQQ45g+1yBtU8HAoJsivYYa/+06Zgu112ex2/ZdHmtXgYSUhIUG5ubq223NxcRUVF1XlWBADQuorKKrTxYKG25hRre26xtuUWa0fuCZ0ob/iMRkiQXV2jQxUfGarOkU7vo1NEiDpGhKhjeIhiI4IVEx6iqNBghQSxmgTq1uJhJD09XYsXL67V9sEHHyg9Pb2lvzUA4DvcHqNNhwq1dt9xfX2wUF8dKNDuoyV19nXYberWMUwpnSKU2ilc3WPD1a1juJJiwpQYE6rYiBDvsAFwNnwOIydOnNDOnTu9z/fs2aMNGzYoNjZW3bt31+zZs5Wdna0///nPkqRbbrlFzz77rH7729/qZz/7mT766CO99dZbWrRoUfPtBQCgTsYYbT5cpKxd+crala8v9xxTcR1nPJJjwzSga5T6xkeqT0Kk+sZHKjUuQsEOzmag5fkcRtasWaNx48Z5n9dMNJ0+fboWLFigw4cPa//+/d7Xe/TooUWLFunOO+/Un/70J3Xr1k3/+7//W+9lvQCAs1Ne6VbWrnx9sDlXH27JVW5Rea3XI0ODNCKlo4Yld9SQ5GgN7Raj2IgQi6oFJJsxvk5Ban1FRUWKjo5WYWGhoqKirC4HANoct8fo0x1H9Ld12Vq+Na/WfI/wEIfSesQq/ZxOSu8ZpwGJUXIw0ROtoLG/v9vk1TQAgMbZkVusd9Yd1N/XZSuv+NszIF0incoYEK9LB8QrvWcnhQY7LKwSaBhhBAD8jMdjtHxbnl76dLc+333M294xPFhXDkvSlcMSNbRbDJe5wm8QRgDAT5RVuPXe+my99Olu7TpSdQWMw27TuL5ddPXwbrq4Xxcun4VfIowAQBtX6fZo4ZoDevrDHTpSPRQT6QzStWnddf35qeoazZpN8G+EEQBoo4wxen9zrh5bslW7q8+EJEaH6mcX9NCUkcmKDA22uEKgeRBGAKAN+vpggR7812at3XdckhQbEaJfXdxL16alMBSDgEMYAYA2pKzCrac+3K6XPtktj5FCg+268YKe+vnYnpwJQcAijABAG7F67zHd/c7X3uXZrxiaqHsn9ldCtHV3UwVaA2EEACxWVuHWo//Zqtey9sqYqjVC/vCDwbp0QLzVpQGtgjACABY6eLxUt76+Tt9kF0qSfjSim343aYCiwxiSQftBGAEAi3y644h+9X/rdby0Qh3Dg/XklGEa17eL1WUBrY4wAgCtzBij5z/epXlLt8ljpMFJ0Xr+J+epW8dwq0sDLEEYAYBW5Kr0aNZbG/Tvrw9LqhqWeejKQdw7Bu0aYQQAWslJl1u3vL5WH28/omCHTQ9eMUhTRyXLZuMeMmjfCCMA0AqKyip0w4LVWr33uMKCHXpx2nCN6d3Z6rKANoEwAgAtLP9Euaa98qU2HSpSZGiQFswYqeEpsVaXBbQZhBEAaEF5RWWa+tLn2nWkRJ0iQvTnG0ZpYGK01WUBbQphBABaSFFZhaa/ulq7jpQoMTpUf7kxTed07mB1WUCbQxgBgBZQXunWz/+8VlsOFymug1Nv3pyu7p24dBeoC7d+BIBm5vEY3fXWV8rana8Ozqo5IgQRoH6EEQBoRsYY/X7RZv3768MKdtg0/yfDNSiJOSJAQwgjANCMXvp0t179bK8kad41Q3VB7zhrCwL8AGEEAJrJyh1HlfmfrZKk+yb115XDkiyuCPAPhBEAaAY5hWW6/c31Mkb68chk3Timp9UlAX6DMAIAZ6nC7dFt/7dO+SUuDegapQeuGGh1SYBfIYwAwFma9/42rd57XB2cQfqf687jpneAjwgjAHAWPticqxc+3i1JeuLqIUqNi7C4IsD/EEYAoIkOHCvVXW9tkCTNOD9VEwZ3tbYgwE8RRgCgCYwx+u07X6uorFLDkmM0e0J/q0sC/BZhBACaYOHqA8rana/QYLv+9ONhCgnixynQVHx6AMBHOYVl+sOiLZKkX1/WVymdmCcCnA3CCAD4wBij+97bqOLySg1NjtGM83tYXRLg9wgjAOCDRd8c1odbchXssOnxHw6Rw26zuiTA7xFGAKCRjpe4NPcfmyRJv7iol/omRFpcERAYCCMA0Ei///dm5Ze41Ce+g34x7hyrywECBmEEABph9d5jend9tmw26bEfDpEziFVWgeZCGAGAMzDG6OHqq2d+PDJZ53bvaHFFQGAhjADAGfzr68P66kCBIkIcuvPSPlaXAwQcwggANKCswq3H/rNVknTL2HPUJTLU4oqAwEMYAYAGLFi1V9kFJ5UQFaobx/S0uhwgIBFGAKAe+SfK9dxHOyVJvxnfV2EhTFoFWgJhBADq8adlO1RcXqmBiVH6wblJVpcDBCzCCADUYWfeCf31i/2SpN9N6i87K60CLYYwAgB1ePKDbXJ7jDL6d9Hoc+KsLgcIaIQRAPiOHbnF+s/GHEnSb8b3s7gaIPARRgDgO/5nxS4ZI40fGM/9Z4BWQBgBgFPsyy/RPzZkS5J+Oa63xdUA7QNhBABO8fyKXfIY6aK+nTW4W7TV5QDtAmEEAKplF5zU39YdlCTddnEvi6sB2g/CCABUe+HjXapwG40+p5OGp8RaXQ7QbhBGAEBSXlGZ3lx9QJL0S86KAK2KMAIAkl76dLdclR4NT+mo9J6drC4HaFcIIwDaveMlLr3+edVqq7+8uJdsNlZbBVoTYQRAu/fm6gM6WeHWgK5RuqhPZ6vLAdodwgiAdq3S7dHrn++TJF1/fipnRQALEEYAtGsfbslTdsFJdQwP1hVDE60uB2iXmhRGnnvuOaWmpio0NFRpaWn68ssvG+z/9NNPq2/fvgoLC1NycrLuvPNOlZWVNalgAGhOr63aK0maOqq7QoMd1hYDtFM+h5GFCxdq1qxZmjt3rtatW6ehQ4dq/PjxysvLq7P/G2+8oXvuuUdz587Vli1b9PLLL2vhwoW69957z7p4ADgbW3OKlLU7Xw67TT/5XorV5QDtls9h5Mknn9RNN92kGTNmaMCAAZo/f77Cw8P1yiuv1Nl/1apVOv/883XttdcqNTVVl112maZOnXrGsykA0NJeW1U1V+SyAfFKjAmzuBqg/fIpjLhcLq1du1YZGRnfvoHdroyMDGVlZdW5zejRo7V27Vpv+Ni9e7cWL16siRMn1vt9ysvLVVRUVOsBAM2psLRC762vuiHe9NGp1hYDtHNBvnQ+evSo3G634uPja7XHx8dr69atdW5z7bXX6ujRo7rgggtkjFFlZaVuueWWBodpMjMz9eCDD/pSGgD45K01VZfz9kuIVFoPln4HrNTiV9OsWLFCjzzyiP7nf/5H69at07vvvqtFixbp97//fb3bzJ49W4WFhd7HgQMHWrpMAO2I22P058/3SpKuH83lvIDVfDozEhcXJ4fDodzc3Frtubm5SkhIqHOb+++/Xz/96U914403SpIGDx6skpIS3Xzzzfrd734nu/30POR0OuV0On0pDQAabfnWPB04dlLRYcG6cliS1eUA7Z5PZ0ZCQkI0fPhwLVu2zNvm8Xi0bNkypaen17lNaWnpaYHD4ai6fM4Y42u9AHDW/lK9yNmPRyYrLITLeQGr+XRmRJJmzZql6dOna8SIERo1apSefvpplZSUaMaMGZKkadOmKSkpSZmZmZKkyZMn68knn9S5556rtLQ07dy5U/fff78mT57sDSUA0FoOFZzUJzuOSJKuTetucTUApCaEkSlTpujIkSOaM2eOcnJyNGzYMC1ZssQ7qXX//v21zoTcd999stlsuu+++5Sdna3OnTtr8uTJ+sMf/tB8ewEAjfTuuoMyRhrVI1YpnSKsLgeAJJvxg7GSoqIiRUdHq7CwUFFRUVaXA8BPGWM0bt4K7c0v1RNXD9E1I5KtLgkIaI39/c29aQC0G6v3Htfe/FJFhDg0cXBXq8sBUI0wAqDdeHtN1TIBk4Z0VYTT51FqAC2EMAKgXSgpr9Sibw5LEsMzQBtDGAHQLiz+5rBKXW71iIvQiJSOVpcD4BSEEQDtwttrDkqSrh7ejRVXgTaGMAIg4O09WqIv9x6T3Sb913msuAq0NYQRAAHvnbVVZ0Uu6N1ZXaPDLK4GwHcRRgAENLfH6G/rqsLIj0Z0s7gaAHUhjAAIaFm78nW4sEzRYcHK6B9vdTkA6kAYARDQ/vXVIUnSxMFdFRrM/bCAtogwAiBguSo9+s/GqrVFrhiaaHE1AOpDGAEQsD7dcURFZZXqEunUqB6xVpcDoB6EEQAB69QhGoedtUWAtoowAiAglVW49cHmXEnSZIZogDaNMAIgIC3fmqcSl1tJMWE6r3uM1eUAaABhBEBA+tfXVUM0lw/tyvLvQBtHGAEQcE6UV2rZljxJ0uQhDNEAbR1hBEDA+XBzrsorPeoRF6GBiVFWlwPgDAgjAAJOzVU0k4cwRAP4A8IIgIBSWFqhT3YckcRVNIC/IIwACChLN+Wowm3ULyFSveMjrS4HQCMQRgAElJqraDgrAvgPwgiAgFFYWqGsXfmSqlZdBeAfCCMAAsZH23JV6THqE99BPeIirC4HQCMRRgAEjKUbq5Z/Hz8wweJKAPiCMAIgIJRVuPXx9qqraAgjgH8hjAAICJ/uOKqTFVX3omGhM8C/EEYABISlm3IkSZcOiGehM8DPEEYA+L1Kt0fLtlTNF7lsYLzF1QDwFWEEgN9bvfe4jpdWqGN4sEalxlpdDgAfEUYA+L33N1cN0VzSP15BDn6sAf6GTy0Av2aM0fubqodoBjBEA/gjwggAv7bpUJGyC04qLNihC/t0trocAE1AGAHg196vvormwj5xCg12WFwNgKYgjADwa+9vZtVVwN8RRgD4rX35JdqaUyyH3aaL+3WxuhwATUQYAeC3Pqg+K5LWI1Yx4SEWVwOgqQgjAPzWR1vzJFVd0gvAfxFGAPil4rIKfbnnmCTpEoZoAL9GGAHgl1buOKpKj1HPuAilxkVYXQ6As0AYAeCXllUP0YzjrAjg9wgjAPyOx2O0Ylv1fBHCCOD3CCMA/M432YU6esKlDs4gjeDGeIDfI4wA8Ds1V9GM6R2nkCB+jAH+jk8xAL/zEfNFgIBCGAHgV/KKyvRNdqEkaVxfwggQCAgjAPzKim1HJElDu0Wrc6TT4moANAfCCAC/whANEHgIIwD8RnmlW5/uqDozckk/loAHAgVhBIDfWL3nuEpcbnWOdGpgYpTV5QBoJoQRAH7DO0TTt7PsdpvF1QBoLoQRAH5jefWqqxczXwQIKIQRAH5hX36J9hwtUZDdpvN7xVldDoBmRBgB4Bc+2V41cXV4SkdFhgZbXA2A5kQYAeAXPt5+VJI0tm9niysB0NwIIwDaPFelR6t2VYWRC3sTRoBA06Qw8txzzyk1NVWhoaFKS0vTl19+2WD/goICzZw5U127dpXT6VSfPn20ePHiJhUMoP1Zs++YSl1uxXVwakBXLukFAk2QrxssXLhQs2bN0vz585WWlqann35a48eP17Zt29Sly+kz3F0uly699FJ16dJF77zzjpKSkrRv3z7FxMQ0R/0A2oFPqodoLuwTxyW9QADyOYw8+eSTuummmzRjxgxJ0vz587Vo0SK98soruueee07r/8orr+jYsWNatWqVgoOrJp2lpqaeXdUA2pWPqyevju3DEA0QiHwapnG5XFq7dq0yMjK+fQO7XRkZGcrKyqpzm3/+859KT0/XzJkzFR8fr0GDBumRRx6R2+2u9/uUl5erqKio1gNA+5RXVKYth4tks0kXcEkvEJB8CiNHjx6V2+1WfHzte0LEx8crJyenzm12796td955R263W4sXL9b999+vP/7xj3r44Yfr/T6ZmZmKjo72PpKTk30pE0AA+WRH1RDN4KRoderAXXqBQNTiV9N4PB516dJFL774ooYPH64pU6bod7/7nebPn1/vNrNnz1ZhYaH3ceDAgZYuE0AbxRANEPh8mjMSFxcnh8Oh3NzcWu25ublKSEioc5uuXbsqODhYDofD29a/f3/l5OTI5XIpJCTktG2cTqecTv4HBLR3bo/Ryuq79F5IGAEClk9nRkJCQjR8+HAtW7bM2+bxeLRs2TKlp6fXuc3555+vnTt3yuPxeNu2b9+url271hlEAKDGN9mFOl5aocjQIJ2bHGN1OQBaiM/DNLNmzdJLL72k1157TVu2bNGtt96qkpIS79U106ZN0+zZs739b731Vh07dky33367tm/frkWLFumRRx7RzJkzm28vAASkj7dVnRW5oFecghys0QgEKp8v7Z0yZYqOHDmiOXPmKCcnR8OGDdOSJUu8k1r3798vu/3bHxrJyclaunSp7rzzTg0ZMkRJSUm6/fbbdffddzffXgAISJ8wRAO0CzZjjLG6iDMpKipSdHS0CgsLFRXF6otAe1BYWqFzf/++PEb67J6LlRQTZnVJAHzU2N/fnPcE0CZ9tuuoPEbq1aUDQQQIcIQRAG3SJ9WX9HJjPCDwEUYAtDnGGH1avdjZmD6sugoEOsIIgDZnz9ESZRecVIjDrrQesVaXA6CFEUYAtDkrd1adFRme0lHhIT5f9AfAzxBGALQ5n2xniAZoTwgjANqUCrdHn+/OlySN6cXkVaA9IIwAaFM2HCjQifJKdQwP1sBE1hUC2gPCCIA2peYqmvN7xclut1lcDYDWQBgB0KZ8uoP1RYD2hjACoM0oPFmhrw4USJIu6M3kVaC9IIwAaDOyqpeAP6dzhBJZAh5oNwgjANoM76qrDNEA7QphBECb8W0YYYgGaE8IIwDahH35Jdp/rFRBdpvSenayuhwArYgwAqBNqDkrcl5KR3VwsgQ80J4QRgC0CStrhmh6MUQDtDeEEQCWq3R79NmumvvRMHkVaG8IIwAs93V2oYrLKhUVGqTBSdFWlwOglRFGAFhu5SlLwDtYAh5odwgjACx3ahgB0P4QRgBYqqS8Uuv2H5fE+iJAe0UYAWCpL/bkq9JjlBwbppROEVaXA8AChBEAlqpZX+SCXlxFA7RXhBEAllrJEvBAu0cYAWCZ3KIy7cg7IZtNSmcJeKDdIowAsEzNWZHBSdHqGBFicTUArEIYAWCZlTtr5oswRAO0Z4QRAJYwxnwbRpgvArRrhBEAltiee0JHissVGmzX8JSOVpcDwEKEEQCW+HTHEUnSqB6d5AxyWFwNACsRRgBYomaIZgzzRYB2jzACoNWVV7r1xe5jkpgvAoAwAsAC6/cX6GSFW3EdQtQ3PtLqcgBYjDACoNWdepdeu91mcTUArEYYAdDqPmV9EQCnIIwAaFUFpS59fbBAkjSmNzfHA0AYAdDKVu3KlzFS7y4dlBAdanU5ANoAwgiAVlWzvghnRQDUIIwAaDXGGH2yvXp9ES7pBVCNMAKg1ezNL1V2wUkFO2xK6xlrdTkA2gjCCIBWs7J6iGZ4SkeFhwRZXA2AtoIwAqDVfLqjZoiG+SIAvkUYAdAqKt0eZe3Kl8R8EQC1EUYAtIqvDhaouLxSMeHBGpgYbXU5ANoQwgiAVlFzFc35veLkYAl4AKcgjABoFSurl4AfwxLwAL6DMAKgxRWVVWjDgQJJ0gXMFwHwHYQRAC0ua1e+3B6jnnER6tYx3OpyALQxhBEALW5l9SW9nBUBUBfCCIAWx/1oADSEMAKgRR04Vqq9+aVy2G36HkvAA6gDYQRAi/p4e9VZkfO6xygyNNjiagC0RYQRAC3qk+owMrYPQzQA6kYYAdBiXJUerapeAn5sny4WVwOgrSKMAGgx6/Yf14nySnWKCNHAxCirywHQRjUpjDz33HNKTU1VaGio0tLS9OWXXzZquzfffFM2m01XXXVVU74tAD9TM0Qzpnec7CwBD6AePoeRhQsXatasWZo7d67WrVunoUOHavz48crLy2twu7179+rXv/61xowZ0+RiAfiXmsmrY/syXwRA/XwOI08++aRuuukmzZgxQwMGDND8+fMVHh6uV155pd5t3G63rrvuOj344IPq2bPnWRUMwD8cKS7XpkNFklhfBEDDfAojLpdLa9euVUZGxrdvYLcrIyNDWVlZ9W730EMPqUuXLrrhhhsa9X3Ky8tVVFRU6wHAv9QsdDYoKUpxHZwWVwOgLfMpjBw9elRut1vx8fG12uPj45WTk1PnNitXrtTLL7+sl156qdHfJzMzU9HR0d5HcnKyL2UCaAM+5pJeAI3UolfTFBcX66c//aleeuklxcU1/p4Us2fPVmFhofdx4MCBFqwSQHPzeIw+rb4fzYUM0QA4gyBfOsfFxcnhcCg3N7dWe25urhISEk7rv2vXLu3du1eTJ0/2tnk8nqpvHBSkbdu26ZxzzjltO6fTKaeT07qAv9p4qFDHSlzq4AzSeSkdrS4HQBvn05mRkJAQDR8+XMuWLfO2eTweLVu2TOnp6af179evn7755htt2LDB+7jiiis0btw4bdiwgeEXIEDVXNJ7fq9OCnawnBGAhvl0ZkSSZs2apenTp2vEiBEaNWqUnn76aZWUlGjGjBmSpGnTpikpKUmZmZkKDQ3VoEGDam0fExMjSae1AwgcNfNFLmS+CIBG8DmMTJkyRUeOHNGcOXOUk5OjYcOGacmSJd5Jrfv375fdzv+EgPaqqKxC6/YXSGK+CIDGsRljjNVFnElRUZGio6NVWFioqCiWlAbasiUbD+uW19epZ+cIfXTXRVaXA8BCjf39zSkMAM2KS3oB+IowAqDZGGO0YhthBIBvCCMAms2Ww8U6XFimsGCHvtezk9XlAPAThBEAzeajrVVrEJ3fK06hwQ6LqwHgLwgjAJrNR1ur7t59cb8uFlcCwJ8QRgA0i2MlLq0/UCBJGteP+SIAGo8wAqBZfLw9T8ZI/btGqWt0mNXlAPAjhBEAzWLZlqohmksYogHgI8IIgLNW6fZ470czjjACwEeEEQBnbe2+4yoqq1RsRIiGJcdYXQ4AP0MYAXDWaq6iGdunsxx2m8XVAPA3hBEAZ41LegGcDcIIgLNy4FipduSdkMNu04UsAQ+gCQgjAM7K8m1VZ0WGp3RUdFiwxdUA8EeEEQBnpeaSXoZoADQVYQRAk5W6KpW1O18SYQRA0xFGADTZZzvz5ar0KCkmTL27dLC6HAB+ijACoMmWbsqRJF02MF42G5f0AmgawgiAJql0e7RsS64k6bIBCRZXA8CfEUYANMnqvcd1vLRCHcODNTK1o9XlAPBjhBEATVIzRHNJ/3gFOfhRAqDp+AkCwGfGGH2wuWqIZvxAhmgAnB3CCACfbTpUpOyCkwoLdmhM7zirywHg5wgjAHxWM0Qztk9nhQY7LK4GgL8jjADw2fubqodoBsVbXAmAQEAYAeCTvUdLtC23WEF2my7uSxgBcPYIIwB88v7mqiGa7/XspOhwbowH4OwRRgD4ZGnNEM1AzooAaB6EEQCNlldcpnX7j0uSLmXVVQDNhDACoNE+3JwnY6ShyTFKiA61uhwAAYIwAqDRvDfGG8AQDYDmQxgB0CjHS1z6bOdRSdL3BzFEA6D5EEYANMqSTTmq9BgN6Bqlczp3sLocAAGEMAKgUf711SFJ0uShiRZXAiDQEEYAnFFecZk+350vSbp8SFeLqwEQaAgjAM7oP9/kyGOkc7vHKDk23OpyAAQYwgiAM6oZorl8CEM0AJofYQRAg7ILTmrNvuOy2aRJgxmiAdD8CCMAGrTo66qzIqNSY1noDECLIIwAaNC/vjosiatoALQcwgiAeu09WqJvsgvlsNs0gYXOALQQwgiAev27eohm9Dmd1KmD0+JqAAQqwgiAejFEA6A1EEYA1Gl7brG25RYr2GHT+IEM0QBoOYQRAHV6d122JGlsny6KDgu2uBoAgYwwAuA0lW6P/rbuoCTp6uFJFlcDINARRgCc5pMdR3SkuFyxESG6uF+81eUACHCEEQCneXtN1VmRq4YlKSSIHxMAWhY/ZQDUcqzEpQ+35EqSrhnRzeJqALQHhBEAtfxjQ7Yq3EaDkqLUv2uU1eUAaAcIIwBqqRmiuWZ4ssWVAGgvCCMAvDZmF2rz4SKFOOy6chgLnQFoHYQRAF7vrK06K3LpwHjFhIdYXA2A9oIwAkCSVF7p1nsbqhY6u2Y4E1cBtB7CCABJ0rIteSoorVBCVKjG9O5sdTkA2hHCCABJ0ttrDkiSfjg8SQ67zeJqALQnTQojzz33nFJTUxUaGqq0tDR9+eWX9fZ96aWXNGbMGHXs2FEdO3ZURkZGg/0BtL4Dx0q1YvsRSdLVXEUDoJX5HEYWLlyoWbNmae7cuVq3bp2GDh2q8ePHKy8vr87+K1as0NSpU7V8+XJlZWUpOTlZl112mbKzs8+6eADN4y+f75Mx0pjeceoRF2F1OQDaGZsxxviyQVpamkaOHKlnn31WkuTxeJScnKzbbrtN99xzzxm3d7vd6tixo5599llNmzatUd+zqKhI0dHRKiwsVFQUizABzemky63vZS5T4ckKvTx9hC7pz71oADSPxv7+9unMiMvl0tq1a5WRkfHtG9jtysjIUFZWVqPeo7S0VBUVFYqNja23T3l5uYqKimo9ALSM9zZkq/BkhbrHhuuivl2sLgdAO+RTGDl69Kjcbrfi42v/zyk+Pl45OTmNeo+7775biYmJtQLNd2VmZio6Otr7SE5mDBtoCcYYLfhsryRpWnoKE1cBWKJVr6Z59NFH9eabb+rvf/+7QkND6+03e/ZsFRYWeh8HDhxoxSqB9uPz3ce0LbdYYcEOXTOC0A/AGkG+dI6Li5PD4VBubm6t9tzcXCUkJDS47bx58/Too4/qww8/1JAhQxrs63Q65XQ6fSkNQBO8tmqvJOm/zktSdFiwtcUAaLd8OjMSEhKi4cOHa9myZd42j8ejZcuWKT09vd7tHn/8cf3+97/XkiVLNGLEiKZXC6DZZBec1Pubq4ZXp49OtbYYAO2aT2dGJGnWrFmaPn26RowYoVGjRunpp59WSUmJZsyYIUmaNm2akpKSlJmZKUl67LHHNGfOHL3xxhtKTU31zi3p0KGDOnTo0Iy7AsAXf8naJ4+RRp/TSX3iI60uB0A75nMYmTJlio4cOaI5c+YoJydHw4YN05IlS7yTWvfv3y+7/dsTLs8//7xcLpeuvvrqWu8zd+5cPfDAA2dXPYAmKatw683V+yVxVgSA9XxeZ8QKrDMCNK83v9yve979RkkxYfrkt+O4igZAi2iRdUYA+L9Kt0fzP94lSbp+dCpBBIDlCCNAO7Pom8Pam1+qjuHBujatu9XlAABhBGhPPB6jZz/aKUm64YIeinD6PG0MAJodYQRoR97fnKMdeScUGRqkaUxcBdBGEEaAdsIYo2eqz4pcPzpVUaEscgagbSCMAO3E8m152nSoSOEhDs04v4fV5QCAF2EEaAeMMfrvZVVnRX7yvRTFRoRYXBEAfIswArQDq3bla8OBAjmD7LpxDGdFALQthBGgHfjvZTskSVNHdVeXyPrvmA0AViCMAAFuxbY8fbHnmIIdNt18YU+rywGA0xBGgABW6fbokcVbJEnT0lOVGBNmcUUAcDrCCBDA3lpzUNtzTyg6LFi3XdzL6nIAoE6EESBAnSiv1JMfbJMk/eqS3ooJ5woaAG0TYQQIUC98vEtHT7iU2ilcP/1eitXlAEC9CCNAADpceFIvfbpbknTPhH4KCeKjDqDt4icUEICeWLpNZRUejUqN1fiBCVaXAwANIowAAWZjdqH+vj5bkvS7Sf1ls9ksrggAGkYYAQKI22P0u/c2yhjpymGJGpocY3VJAHBGhBEggLz62R59daBAkc4gzZ7Q3+pyAKBRCCNAgNifX6p571ddynvvpP5KiGbZdwD+gTACBABjjO5592uVVXiU3rOTfjwy2eqSAKDRCCNAAHhrzQGt2pWv0GC7Mv9rMJNWAfgVwgjg53KLyvTwoqr7z9x1aV+lxkVYXBEA+IYwAvgxY4zuf2+jissqNbRbtGacn2p1SQDgM8II4MfeXH1A72/OVZDdpseuHqIgBx9pAP6Hn1yAn9qYXai5/9wkSbrrsr7qlxBlcUUA0DSEEcAPFZVVaOYb6+Sq9OiSfl308wt7Wl0SADQZYQTwM8YY/ebtr7Qvv1RJMWH644+Gym7n6hkA/oswAviZVz7bq6WbchXssOl/rjtPMeEhVpcEAGeFMAL4kbX7jitzcdVlvPdNGsC9ZwAEBMII4Cf2Hi3RzX9eo0qP0aQhXTUtPcXqkgCgWRBGAD9w9ES5pr/6pfJLXBqYGKXHfjiEVVYBBAzCCNDGlZRXasarq7Uvv1TJsWF6dcZIdXAGWV0WADQbwgjQhlW4Pbr1r+v0TXahYiNC9NqMUeoSyd14AQQWwgjQRnk8Rnf/7Wt9sv2IwoIdeuX6kerZuYPVZQFAsyOMAG1QpdujX7/zld5dly2HveoS3mFcOQMgQDHwDLQx5ZVu3fbGer2/OVcOu01P/mioxvXrYnVZANBiCCNAG1LqqtTNf16rlTuPKsRh17PXnqvLBiZYXRYAtCjCCNBGFJZWaMaCL7Vuf4HCQxx6adoInd8rzuqyAKDFEUaANmBn3gnd8vpa7cw7oeiwYC2YMVLndu9odVkA0CoII4DFlmzM0a/f/konyisVH+XUghmj1L9rlNVlAUCrIYwAFnF7jOa9v03Pr9glSUrrEatnrz1PnSOdFlcGAK2LMAJYIK+oTLPe+kordx6VJN14QQ/dPaGfgh1cbQ+g/SGMAK3IGKO/rcvWQ//apKKySoUFO/T41UM0eWii1aUBgGUII0AryS44qXvf/UYfbz8iSRqcFK0//mio+sRHWlwZAFiLMAK0sAq3R298sV9PLN2mE+WVCgmya9alfXTjBT0UxLAMABBGgJZijNGSjTl6Yuk27T5aIkkantJRj189ROdwjxkA8CKMAC1g9d5jemTxFq3fXyBJiusQotsz+ujaUd3lsNusLQ4A2hjCCNBMPB6jFdvz9OInu/X57mOSpLBgh266sKduvrCnOjj5uAFAXfjpCJylsgq33lufrZc+3a1dR6qGY4LsNl0zIll3ZvRWl6hQiysEgLaNMAI0gTFGG7OL9M7aA/rnV4d0vLRCkhTpDNLUtO66fnSqEmPCLK4SAPwDYQTwwcHjpVr8zWH9bW22tuUWe9sTo0P1swt6aMrIZEWGBltYIQD4H8II0ICaMyAfbM7RB1vytOVwkfc1Z5Bdlw1M0NXDu+mCXnFMTAWAJiKMAKcwxmj/sVJl7cpX1u58Ze3KV15xufd1u00akRqrq4YladKQrooO4ywIAJwtwgjatVJXpTZmF+mrAwX66mCB1u07rkOFZbX6hIc4dGHvzsoYEK+L+3VRbESIRdUCQGAijKBdcHuMDhwr1bbcYm3PKdb2vBPallOknXkn5DG1+wY7bBraLUbp53RSes9OOi+lo0KDHdYUDgDtAGEEAcEYo+OlFTpUcFLZBSd1qOCk9uWXal9+ifbll+rA8VJVuE2d2yZEhWpIt2gNTY7R0G4xOi8lRuEhfDQAoLU06Sfuc889pyeeeEI5OTkaOnSonnnmGY0aNare/m+//bbuv/9+7d27V71799Zjjz2miRMnNrloBD5jjMoqPCoqq1BBaYWOlbh0vNSlYyVVj6MnynWkuPpxoly5RWUqq/A0+J7OILt6demgvvGR6pMQqT7xHTQwMVrxrAMCAJbyOYwsXLhQs2bN0vz585WWlqann35a48eP17Zt29SlS5fT+q9atUpTp05VZmamLr/8cr3xxhu66qqrtG7dOg0aNKhZdgKtxxijSo9RpdvI5faoovrhqqx6lFc/XJUelVW6VV7hVlmFR2UVbpVVuFVa4VaZy61SV9XXpeWVOlHu1onyCpWUu1VSXqmisgoVnqyo90xGQzpHOpUYHaqu0WFK6RSulE4RSu0UrpS4CCVEhXLFCwC0QTZjjE8/8dPS0jRy5Eg9++yzkiSPx6Pk5GTddtttuueee07rP2XKFJWUlOjf//63t+173/uehg0bpvnz59f5PcrLy1Ve/u0VDEVFRUpOTlZhYaGioqJ8KbdBL6/cowPHSs/Yr+avyHif19Gn+tVTXzO1nhvv18bU7n/q+xpVNZjq72tO6WOM8fYx5tv+npqvjZHHVD33VPf1nNJmjJHb8+3rHmPk9khuj8fbXunxyOOp+tPt+TZ4VLi/fd6aHHabosOC1TE8WLERIYoJD1FseIg6RzprPeIjQxUf7ZQziLkdANBWFBUVKTo6+oy/v306M+JyubR27VrNnj3b22a325WRkaGsrKw6t8nKytKsWbNqtY0fP17vvfdevd8nMzNTDz74oC+lNcmirw9pXfWNzNB0IQ67gh02hQTZv3047AoNdlQ/7AoNqvo6LMSh8BCHwqq/jggJUofQIEU4g9TB6VAHZ7CiwoIUFRqs6LBghYc4ZLNxNgMAAplPYeTo0aNyu92Kj4+v1R4fH6+tW7fWuU1OTk6d/XNycur9PrNnz64VYGrOjDS3Hw7vptHnxJ3WXtfvvtOaTulk+06T7ZTe37Z957mtdh+bbNV/1n5ew2479XWb7NUd7af0tdtqXquqwGGvabdVPyS7veprh/3b9iC7TXZ77T8ddpuC7PbqP20KclQ9D3LYFGy3KzjIpmCHXUF2G2EBAHBW2uQlA06nU06ns8W/z3VpKS3+PQAAQMPsvnSOi4uTw+FQbm5urfbc3FwlJCTUuU1CQoJP/QEAQPviUxgJCQnR8OHDtWzZMm+bx+PRsmXLlJ6eXuc26enptfpL0gcffFBvfwAA0L74PEwza9YsTZ8+XSNGjNCoUaP09NNPq6SkRDNmzJAkTZs2TUlJScrMzJQk3X777Ro7dqz++Mc/atKkSXrzzTe1Zs0avfjii827JwAAwC/5HEamTJmiI0eOaM6cOcrJydGwYcO0ZMkS7yTV/fv3y27/9oTL6NGj9cYbb+i+++7Tvffeq969e+u9995jjREAACCpCeuMWKGx1ykDAIC2o7G/v32aMwIAANDcCCMAAMBShBEAAGApwggAALAUYQQAAFiKMAIAACxFGAEAAJYijAAAAEu1ybv2flfNumxFRUUWVwIAABqr5vf2mdZX9YswUlxcLElKTk62uBIAAOCr4uJiRUdH1/u6XywH7/F4dOjQIUVGRspmszXb+xYVFSk5OVkHDhwI2GXmA30f2T//F+j7yP75v0Dfx5bcP2OMiouLlZiYWOu+dd/lF2dG7Ha7unXr1mLvHxUVFZD/wE4V6PvI/vm/QN9H9s//Bfo+ttT+NXRGpAYTWAEAgKUIIwAAwFLtOow4nU7NnTtXTqfT6lJaTKDvI/vn/wJ9H9k//xfo+9gW9s8vJrACAIDA1a7PjAAAAOsRRgAAgKUIIwAAwFKEEQAAYCnCCAAAsFTAh5E//OEPGj16tMLDwxUTE1Nnn/3792vSpEkKDw9Xly5d9Jvf/EaVlZUNvu+xY8d03XXXKSoqSjExMbrhhht04sSJFtiDxluxYoVsNludj9WrV9e73UUXXXRa/1tuuaUVK/dNamrqafU++uijDW5TVlammTNnqlOnTurQoYN++MMfKjc3t5Uqbry9e/fqhhtuUI8ePRQWFqZzzjlHc+fOlcvlanC7tn4Mn3vuOaWmpio0NFRpaWn68ssvG+z/9ttvq1+/fgoNDdXgwYO1ePHiVqrUN5mZmRo5cqQiIyPVpUsXXXXVVdq2bVuD2yxYsOC0YxUaGtpKFfvugQceOK3efv36NbiNvxw/qe6fJzabTTNnzqyzf1s/fp988okmT56sxMRE2Ww2vffee7VeN8Zozpw56tq1q8LCwpSRkaEdO3ac8X19/Qz7KuDDiMvl0jXXXKNbb721ztfdbrcmTZokl8ulVatW6bXXXtOCBQs0Z86cBt/3uuuu06ZNm/TBBx/o3//+tz755BPdfPPNLbELjTZ69GgdPny41uPGG29Ujx49NGLEiAa3vemmm2pt9/jjj7dS1U3z0EMP1ar3tttua7D/nXfeqX/96196++239fHHH+vQoUP6r//6r1aqtvG2bt0qj8ejF154QZs2bdJTTz2l+fPn69577z3jtm31GC5cuFCzZs3S3LlztW7dOg0dOlTjx49XXl5enf1XrVqlqVOn6oYbbtD69et11VVX6aqrrtLGjRtbufIz+/jjjzVz5kx9/vnn+uCDD1RRUaHLLrtMJSUlDW4XFRVV61jt27evlSpumoEDB9aqd+XKlfX29afjJ0mrV6+utW8ffPCBJOmaa66pd5u2fPxKSko0dOhQPffcc3W+/vjjj+u///u/NX/+fH3xxReKiIjQ+PHjVVZWVu97+voZbhLTTrz66qsmOjr6tPbFixcbu91ucnJyvG3PP/+8iYqKMuXl5XW+1+bNm40ks3r1am/bf/7zH2Oz2Ux2dnaz195ULpfLdO7c2Tz00EMN9hs7dqy5/fbbW6eoZpCSkmKeeuqpRvcvKCgwwcHB5u233/a2bdmyxUgyWVlZLVBh83r88cdNjx49GuzTlo/hqFGjzMyZM73P3W63SUxMNJmZmXX2/9GPfmQmTZpUqy0tLc38/Oc/b9E6m0NeXp6RZD7++ON6+9T3s6itmjt3rhk6dGij+/vz8TPGmNtvv92cc845xuPx1Pm6Px0/Sebvf/+797nH4zEJCQnmiSee8LYVFBQYp9Np/u///q/e9/H1M9wUAX9m5EyysrI0ePBgxcfHe9vGjx+voqIibdq0qd5tYmJiap1tyMjIkN1u1xdffNHiNTfWP//5T+Xn52vGjBln7PvXv/5VcXFxGjRokGbPnq3S0tJWqLDpHn30UXXq1EnnnnuunnjiiQaH1dauXauKigplZGR42/r166fu3bsrKyurNco9K4WFhYqNjT1jv7Z4DF0ul9auXVvr795utysjI6Pev/usrKxa/aWqz6S/HCtJZzxeJ06cUEpKipKTk3XllVfW+7OmrdixY4cSExPVs2dPXXfdddq/f3+9ff35+LlcLr3++uv62c9+1uAd4v3t+NXYs2ePcnJyah2f6OhopaWl1Xt8mvIZbgq/uGtvS8rJyakVRCR5n+fk5NS7TZcuXWq1BQUFKTY2tt5trPDyyy9r/PjxZ7zj8bXXXquUlBQlJibq66+/1t13361t27bp3XffbaVKffOrX/1K5513nmJjY7Vq1SrNnj1bhw8f1pNPPlln/5ycHIWEhJw2Zyg+Pr5NHa+67Ny5U88884zmzZvXYL+2egyPHj0qt9td52ds69atdW5T32eyrR8rj8ejO+64Q+eff74GDRpUb7++ffvqlVde0ZAhQ1RYWKh58+Zp9OjR2rRpU4venbyp0tLStGDBAvXt21eHDx/Wgw8+qDFjxmjjxo2KjIw8rb+/Hj9Jeu+991RQUKDrr7++3j7+dvxOVXMMfDk+TfkMN4VfhpF77rlHjz32WIN9tmzZcsZJVv6iKft78OBBLV26VG+99dYZ3//UuS6DBw9W165ddckll2jXrl0655xzml64D3zZx1mzZnnbhgwZopCQEP385z9XZmZmm713RFOOYXZ2tr7//e/rmmuu0U033dTgtm3hGLZ3M2fO1MaNGxucTyFJ6enpSk9P9z4fPXq0+vfvrxdeeEG///3vW7pMn02YMMH79ZAhQ5SWlqaUlBS99dZbuuGGGyysrPm9/PLLmjBhghITE+vt42/Hz1/4ZRi56667GkyuktSzZ89GvVdCQsJps4JrrrJISEiod5vvTtyprKzUsWPH6t3mbDRlf1999VV16tRJV1xxhc/fLy0tTVLV/8pb6xfZ2RzTtLQ0VVZWau/everbt+9pryckJMjlcqmgoKDW2ZHc3NwWOV518XX/Dh06pHHjxmn06NF68cUXff5+VhzDusTFxcnhcJx25VJDf/cJCQk+9W8LfvnLX3onsvv6v+Pg4GCde+652rlzZwtV17xiYmLUp0+feuv1x+MnSfv27dOHH37o89lEfzp+NccgNzdXXbt29bbn5uZq2LBhdW7TlM9wkzTb7JM27kwTWHNzc71tL7zwgomKijJlZWV1vlfNBNY1a9Z425YuXdpmJrB6PB7To0cPc9dddzVp+5UrVxpJ5quvvmrmylrG66+/bux2uzl27Fidr9dMYH3nnXe8bVu3bm2zE1gPHjxoevfubX784x+bysrKJr1HWzqGo0aNMr/85S+9z91ut0lKSmpwAuvll19eqy09Pb1NToD0eDxm5syZJjEx0Wzfvr1J71FZWWn69u1r7rzzzmaurmUUFxebjh07mj/96U91vu5Px+9Uc+fONQkJCaaiosKn7dry8VM9E1jnzZvnbSssLGzUBFZfPsNNqrXZ3qmN2rdvn1m/fr158MEHTYcOHcz69evN+vXrTXFxsTGm6h/SoEGDzGWXXWY2bNhglixZYjp37mxmz57tfY8vvvjC9O3b1xw8eNDb9v3vf9+ce+655osvvjArV640vXv3NlOnTm31/avLhx9+aCSZLVu2nPbawYMHTd++fc0XX3xhjDFm586d5qGHHjJr1qwxe/bsMf/4xz9Mz549zYUXXtjaZTfKqlWrzFNPPWU2bNhgdu3aZV5//XXTuXNnM23aNG+f7+6jMcbccsstpnv37uajjz4ya9asMenp6SY9Pd2KXWjQwYMHTa9evcwll1xiDh48aA4fPux9nNrHn47hm2++aZxOp1mwYIHZvHmzufnmm01MTIz3Craf/vSn5p577vH2/+yzz0xQUJCZN2+e2bJli5k7d64JDg4233zzjVW7UK9bb73VREdHmxUrVtQ6VqWlpd4+392/Bx980CxdutTs2rXLrF271vz4xz82oaGhZtOmTVbswhndddddZsWKFWbPnj3ms88+MxkZGSYuLs7k5eUZY/z7+NVwu92me/fu5u677z7tNX87fsXFxd7fc5LMk08+adavX2/27dtnjDHm0UcfNTExMeYf//iH+frrr82VV15pevToYU6ePOl9j4svvtg888wz3udn+gw3h4API9OnTzeSTnssX77c22fv3r1mwoQJJiwszMTFxZm77rqrVjpevny5kWT27NnjbcvPzzdTp041HTp0MFFRUWbGjBnegGO1qVOnmtGjR9f52p49e2rt//79+82FF15oYmNjjdPpNL169TK/+c1vTGFhYStW3Hhr1641aWlpJjo62oSGhpr+/fubRx55pNZZrO/uozHGnDx50vziF78wHTt2NOHh4eYHP/hBrV/wbcWrr75a57/XU09i+uMxfOaZZ0z37t1NSEiIGTVqlPn888+9r40dO9ZMnz69Vv+33nrL9OnTx4SEhJiBAweaRYsWtXLFjVPfsXr11Ve9fb67f3fccYf37yI+Pt5MnDjRrFu3rvWLb6QpU6aYrl27mpCQEJOUlGSmTJlidu7c6X3dn49fjaVLlxpJZtu2bae95m/Hr+b31XcfNfvg8XjM/fffb+Lj443T6TSXXHLJafudkpJi5s6dW6utoc9wc7AZY0zzDfoAAAD4pt2vMwIAAKxFGAEAAJYijAAAAEsRRgAAgKUIIwAAwFKEEQAAYCnCCAAAsBRhBAAAWIowAgAALEUYAQAAliKMAAAAS/0/9jKUy52ew6AAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import torch\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "x = torch.linspace(-10, 10,100)\n",
    "relu = torch.nn.Sigmoid()\n",
    "y = relu(x)\n",
    "\n",
    "plt.title('Sigmoid')\n",
    "plt.plot(x.tolist(), y.tolist())\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "281038d8",
   "metadata": {},
   "source": [
    "## tanh function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "699b379c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import torch\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "x = torch.linspace(-10, 10,100)\n",
    "relu = torch.nn.Tanh()\n",
    "y = relu(x)\n",
    "\n",
    "plt.title('Tanh')\n",
    "plt.plot(x.tolist(), y.tolist())\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}