AutoEncoder.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "d9f9bb3c-5a47-4ef3-b6dd-4c6d68fb9c9f",
   "metadata": {},
   "source": [
    "# Auto Encoder\n",
    "- 오토인코더는 정답 없이 모델을 학습시키는 비지도 학습 모델"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1c6d717f-791e-4ff4-a73a-7f31c4dd8eb3",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torchvision\n",
    "from torchvision import transforms\n",
    "\n",
    "import torch.nn.functional as F\n",
    "import torch.nn as nn\n",
    "\n",
    "import torch.optim as optim\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "46cc56fd-8d8e-49fe-bd90-8b43e55102e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "fec6505f-2187-415b-9aaa-b34ac2548f87",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz\n",
      "Failed to download (trying next):\n",
      "HTTP Error 403: Forbidden\n",
      "\n",
      "Downloading https://ossci-datasets.s3.amazonaws.com/mnist/train-images-idx3-ubyte.gz\n",
      "Downloading https://ossci-datasets.s3.amazonaws.com/mnist/train-images-idx3-ubyte.gz to ./data/MNIST/raw/train-images-idx3-ubyte.gz\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████| 9912422/9912422 [00:02<00:00, 4504272.65it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting ./data/MNIST/raw/train-images-idx3-ubyte.gz to ./data/MNIST/raw\n",
      "\n",
      "Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz\n",
      "Failed to download (trying next):\n",
      "HTTP Error 403: Forbidden\n",
      "\n",
      "Downloading https://ossci-datasets.s3.amazonaws.com/mnist/train-labels-idx1-ubyte.gz\n",
      "Downloading https://ossci-datasets.s3.amazonaws.com/mnist/train-labels-idx1-ubyte.gz to ./data/MNIST/raw/train-labels-idx1-ubyte.gz\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████| 28881/28881 [00:00<00:00, 140090.50it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting ./data/MNIST/raw/train-labels-idx1-ubyte.gz to ./data/MNIST/raw\n",
      "\n",
      "Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz\n",
      "Failed to download (trying next):\n",
      "HTTP Error 403: Forbidden\n",
      "\n",
      "Downloading https://ossci-datasets.s3.amazonaws.com/mnist/t10k-images-idx3-ubyte.gz\n",
      "Downloading https://ossci-datasets.s3.amazonaws.com/mnist/t10k-images-idx3-ubyte.gz to ./data/MNIST/raw/t10k-images-idx3-ubyte.gz\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████| 1648877/1648877 [00:01<00:00, 984471.61it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting ./data/MNIST/raw/t10k-images-idx3-ubyte.gz to ./data/MNIST/raw\n",
      "\n",
      "Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz\n",
      "Failed to download (trying next):\n",
      "HTTP Error 403: Forbidden\n",
      "\n",
      "Downloading https://ossci-datasets.s3.amazonaws.com/mnist/t10k-labels-idx1-ubyte.gz\n",
      "Downloading https://ossci-datasets.s3.amazonaws.com/mnist/t10k-labels-idx1-ubyte.gz to ./data/MNIST/raw/t10k-labels-idx1-ubyte.gz\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████| 4542/4542 [00:00<00:00, 1306531.02it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting ./data/MNIST/raw/t10k-labels-idx1-ubyte.gz to ./data/MNIST/raw\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "dataset = torchvision.datasets.MNIST('./data/', download = True, train = True, transform = transforms.ToTensor())\n",
    "trainloader = torch.utils.data.DataLoader(dataset, batch_size = 50, shuffle = True)\n",
    "# img size in MNIST dataset is 28 * 28 = 784"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "922ce3a0-3d57-489e-a8d2-508d497335f8",
   "metadata": {},
   "source": [
    "## Construct Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "87cba025-9ab8-4d6a-b392-7b199e6e4563",
   "metadata": {},
   "outputs": [],
   "source": [
    "class AutoEncoder(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(AutoEncoder, self).__init__()\n",
    "        self.encoder = nn.Sequential(\n",
    "            nn.Linear(784, 128),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(128, 32),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(32, 10),\n",
    "            nn.ReLU())\n",
    "        self.decoder = nn.Sequential(\n",
    "            nn.Linear(10, 32),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(32, 128),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(128, 28*28),\n",
    "            nn.Sigmoid())\n",
    "\n",
    "    def forward(self, x):\n",
    "        encoded = self.encoder(x)\n",
    "        decoded = self.decoder(encoded)\n",
    "        return decoded"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "09de822e-6393-4436-8723-0b60694daed3",
   "metadata": {},
   "source": [
    "## 1. 스택 오토인코더(Stacked Autoencoder)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ea5f82cb-b432-4a4a-9cba-0e26bb3860f2",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = AutoEncoder().to(device)\n",
    "criterion = nn.MSELoss()\n",
    "optimizer = optim.Adam(model.parameters(), lr=1e-4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d967c2d2-6465-4334-bcaf-9820fd8bbea6",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 loss: 0.081\n",
      "2 loss: 0.054\n",
      "3 loss: 0.044\n",
      "4 loss: 0.039\n",
      "5 loss: 0.035\n",
      "6 loss: 0.033\n",
      "7 loss: 0.031\n",
      "8 loss: 0.030\n",
      "9 loss: 0.030\n",
      "10 loss: 0.029\n",
      "11 loss: 0.028\n",
      "12 loss: 0.028\n",
      "13 loss: 0.027\n",
      "14 loss: 0.027\n",
      "15 loss: 0.026\n",
      "16 loss: 0.026\n",
      "17 loss: 0.025\n",
      "18 loss: 0.025\n",
      "19 loss: 0.025\n",
      "20 loss: 0.024\n",
      "21 loss: 0.024\n",
      "22 loss: 0.024\n",
      "23 loss: 0.024\n",
      "24 loss: 0.024\n",
      "25 loss: 0.023\n",
      "26 loss: 0.023\n",
      "27 loss: 0.023\n",
      "28 loss: 0.023\n",
      "29 loss: 0.023\n",
      "30 loss: 0.023\n",
      "31 loss: 0.023\n",
      "32 loss: 0.023\n",
      "33 loss: 0.022\n",
      "34 loss: 0.022\n",
      "35 loss: 0.022\n",
      "36 loss: 0.022\n",
      "37 loss: 0.022\n",
      "38 loss: 0.022\n",
      "39 loss: 0.022\n",
      "40 loss: 0.022\n",
      "41 loss: 0.022\n",
      "42 loss: 0.022\n",
      "43 loss: 0.022\n",
      "44 loss: 0.021\n",
      "45 loss: 0.021\n",
      "46 loss: 0.021\n",
      "47 loss: 0.021\n",
      "48 loss: 0.021\n",
      "49 loss: 0.021\n",
      "50 loss: 0.021\n",
      "51 loss: 0.021\n"
     ]
    }
   ],
   "source": [
    "for epoch in range(51):\n",
    "    running_loss = 0.0\n",
    "    for data in trainloader:\n",
    "        inputs = data[0].to(device)\n",
    "        optimizer.zero_grad()\n",
    "        outputs = model(inputs.view(-1, 28*28)) # batch_size x 784 의 형태로 들어감\n",
    "        outputs = outputs.view(-1, 1, 28, 28)\n",
    "        loss = criterion(inputs, outputs)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        running_loss += loss.item()\n",
    "    cost = running_loss / len(trainloader)\n",
    "    print(f\"{epoch+1} loss: {cost:.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "97121849-5d1d-4218-b819-7e5f46c82625",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1200x500 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(2, 5, figsize=(12, 5))\n",
    "\n",
    "axes = axes.flatten()\n",
    "\n",
    "for i in range(10):\n",
    "    axes[i].imshow(outputs[i].cpu().detach().numpy().squeeze(), cmap='gray')\n",
    "    axes[i].set_title(\"Reconstructed\")\n",
    "    axes[i].axis('off') \n",
    "    \n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "94f150a9-071e-4dbf-a024-dc4ad3f0e227",
   "metadata": {},
   "source": [
    "## 2. 디노이징 오토인코더(Denoising Autoencoder)\n",
    "- 입력 이미지에 가우시안 노이즈 주입\n",
    "- 가우시안 노이즈: 평균이 0이고 표준 편차가 1인 가우시안 분포를 따르는 노이즈"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "bdc361ac-443c-42fb-9471-e68e5adc39fa",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = AutoEncoder().to(device)\n",
    "criterion = nn.MSELoss()\n",
    "optimizer = optim.Adam(model.parameters(), lr=1e-4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "ae018633-2f17-4037-b6c3-525d5c760401",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 loss: 0.083\n",
      "2 loss: 0.059\n",
      "3 loss: 0.052\n",
      "4 loss: 0.044\n",
      "5 loss: 0.042\n",
      "6 loss: 0.039\n",
      "7 loss: 0.038\n",
      "8 loss: 0.036\n",
      "9 loss: 0.035\n",
      "10 loss: 0.034\n",
      "11 loss: 0.033\n",
      "12 loss: 0.032\n",
      "13 loss: 0.032\n",
      "14 loss: 0.031\n",
      "15 loss: 0.031\n",
      "16 loss: 0.030\n",
      "17 loss: 0.030\n",
      "18 loss: 0.030\n",
      "19 loss: 0.029\n",
      "20 loss: 0.029\n",
      "21 loss: 0.029\n",
      "22 loss: 0.028\n",
      "23 loss: 0.028\n",
      "24 loss: 0.028\n",
      "25 loss: 0.028\n",
      "26 loss: 0.028\n",
      "27 loss: 0.027\n",
      "28 loss: 0.027\n",
      "29 loss: 0.027\n",
      "30 loss: 0.027\n",
      "31 loss: 0.027\n",
      "32 loss: 0.027\n",
      "33 loss: 0.027\n",
      "34 loss: 0.026\n",
      "35 loss: 0.026\n",
      "36 loss: 0.026\n",
      "37 loss: 0.026\n",
      "38 loss: 0.026\n",
      "39 loss: 0.026\n",
      "40 loss: 0.026\n",
      "41 loss: 0.026\n",
      "42 loss: 0.026\n",
      "43 loss: 0.025\n",
      "44 loss: 0.025\n",
      "45 loss: 0.025\n",
      "46 loss: 0.025\n",
      "47 loss: 0.025\n",
      "48 loss: 0.025\n",
      "49 loss: 0.025\n",
      "50 loss: 0.025\n",
      "51 loss: 0.025\n"
     ]
    }
   ],
   "source": [
    "# 가우시안 노이즈 주입\n",
    "\n",
    "for epoch in range(51):\n",
    "    running_loss = 0.0\n",
    "    for data in trainloader:\n",
    "        inputs = data[0].to(device)\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        dirty_inputs = inputs + torch.normal(0, 0.5, size = inputs.size()).to(device)\n",
    "        \n",
    "        outputs = model(dirty_inputs.view(-1, 28*28)) # batch_size x 784 의 형태로 들어감\n",
    "        outputs = outputs.view(-1, 1, 28, 28)\n",
    "        loss = criterion(inputs, outputs)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        running_loss += loss.item()\n",
    "    cost = running_loss / len(trainloader)\n",
    "    print(f\"{epoch+1} loss: {cost:.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "8585f31a-e8ec-4fde-bf3a-3023b955c3ef",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABKIAAAHxCAYAAABeRhR0AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAABaZ0lEQVR4nO3daZRV5Z3v8f+pOjWP1FzFUAWCglpM4gBRRImo4NBRY2NrL9F0zLU7K7m2Xju60pJor173qjHedZfE9E2WhlyHaIwx0ZZ0FIwTCmqYBxmKmSpqpqjxDPu+sKnuat2/B+vALg71/azlC/nynLPrnP3svc/DgR3yPM8zAAAAAAAA4ARLGeoNAAAAAAAAwPDAQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAjEsF6KefvppC4VC/f+Fw2GrrKy0hQsX2rZt24Z6846rJUuW2NNPPz2k2/Dss8/a448/fkIeu6amxhYtWnRCHhsnD+ZssJizSBRzNljMWSSKORss5iwSxZwNFnP2+BuWC1FHPfXUU7Zy5Up744037Nvf/rb97ne/swsvvNBaW1uHetOOm1N94mJ4Yc4GgzmL44U5GwzmLI4X5mwwmLM4XpizwWDOHn/hod6AoXT22WfbjBkzzMxszpw5FovFbPHixfbb3/7WbrvttiHeuuBFIpH+FXXgZMScHYg5i5Mdc3Yg5ixOdszZgZizONkxZwdiziaPYf2NqP/q6CRuaGjo/7WPPvrIrrnmGisqKrLMzEybNm2avfDCC58bu3//frvjjjts9OjRlp6eblVVVXbDDTcMeKw9e/bYLbfcYmVlZZaRkWGTJk2yH/3oRxaPx/t/z65duywUCtmjjz5qjz32mI0dO9Zyc3Nt5syZ9sEHHwx4zp07d9rChQutqqrKMjIyrLy83ObOnWtr1qwxs8++5rdx40b705/+1P+1zZqaGjMze+uttywUCtkvf/lLu/vuu23kyJGWkZFh27dvtx/84AcWCoU+9zMe/Qrorl27Bvz6s88+azNnzrTc3FzLzc21qVOn2s9//nMz++yA+Nprr9nu3bsHfH30qL6+Pvunf/onmzhxomVkZFhpaanddttt1tjYOOA5IpGI3XvvvVZRUWHZ2dl24YUX2qpVq/zeSgwTzFnmLJILc5Y5i+TCnGXOIrkwZ5mzyYKlwv+krq7OzMxOP/10MzNbsWKFXXHFFXb++efbk08+aQUFBfb888/bX/7lX1pXV1f/3+Xcv3+/nXvuuRaJROz++++3yZMnW3Nzs/3hD3+w1tZWKy8vt8bGRps1a5b19fXZQw89ZDU1Nfbqq6/aPffcYzt27LAlS5YM2JYnnnjCJk6c2P8VwH/8x3+0+fPnW11dnRUUFJiZ2fz58y0Wi9nDDz9sY8aMsaamJnv//fetra3NzMxefvllu+GGG6ygoKD/8TMyMgY8z3333WczZ860J5980lJSUqysrOxLvWYPPPCAPfTQQ3bdddfZ3XffbQUFBbZhwwbbvXu3mX32Vco77rjDduzYYS+//PKAsfF43K699lp755137N5777VZs2bZ7t27bfHixTZnzhz76KOPLCsry8zMvvnNb9rSpUvtnnvuscsuu8w2bNhg1113nXV0dHyp7cWphTnLnEVyYc4yZ5FcmLPMWSQX5ixzNml4w9BTTz3lmZn3wQcfeJFIxOvo6PCWLVvmVVRUeLNnz/YikYjneZ43ceJEb9q0af3/f9RVV13lVVZWerFYzPM8z7v99tu9tLQ0b9OmTb7P+b3vfc8zM+/DDz8c8Ot33nmnFwqFvK1bt3qe53l1dXWemXm1tbVeNBrt/32rVq3yzMx77rnnPM/zvKamJs/MvMcff1z+rGeddZZ38cUXf+7XV6xY4ZmZN3v27M+1xYsXe1+0axx93erq6jzP87ydO3d6qamp3s033yy3YcGCBV51dfXnfv25557zzMx76aWXBvz66tWrPTPzlixZ4nme523evNkzM++uu+4a8PueeeYZz8y8W2+9VT4/kh9zljmL5MKcZc4iuTBnmbNILsxZ5myyG9Z/Ne+CCy6wtLQ0y8vLsyuuuMJGjBhhr7zyioXDYdu+fbtt2bLFbr75ZjMzi0aj/f/Nnz/fDh48aFu3bjUzs9dff90uueQSmzRpku9zLV++3M4880w777zzBvz6okWLzPM8W758+YBfX7BggaWmpvb//+TJk83M+ldmi4qK7LTTTrNHHnnEHnvsMfvzn/884CuRx+r666//0mOO+uMf/2ixWMz+7u/+blDjX331VSssLLSrr756wOs7depUq6iosLfeesvMPlvJN7P+9+KoG2+8kb//O8wwZ5mzSC7MWeYskgtzljmL5MKcZc4mq2G9ELV06VJbvXq1LV++3L71rW/Z5s2b7aabbjKz//h7tffcc4+lpaUN+O9v//ZvzcysqanJzMwaGxtt1KhR8rmam5utsrLyc79eVVXV3/+z4uLiAf9/9CuI3d3dZmYWCoXszTfftMsvv9wefvhhmz59upWWltp3vvOdL/X1vi/apmN19O+9un52Pw0NDdbW1mbp6emfe43r6+v7X9+jr01FRcWA8eFw+HOvE05tzFnmLJILc5Y5i+TCnGXOIrkwZ5mzyWp4Lr/9u0mTJvX/g26XXHKJxWIx+9nPfma//vWvrba21sw++zun11133ReOP+OMM8zMrLS01Pbt2yefq7i42A4ePPi5Xz9w4ICZmZWUlHzp7a+uru7/R9Q+/fRTe+GFF+wHP/iB9fX12ZNPPnlMj/FF/4hbZmammZn19vYO+Du4RyfSUaWlpWZmtm/fPhs9evSX3v6SkhIrLi62ZcuWfWHPy8szs/84iNXX19vIkSP7ezQa/dwBD6c25ixzFsmFOcucRXJhzjJnkVyYs8zZZDWsvxH1Xz388MM2YsQIe+CBB2zChAk2YcIEW7t2rc2YMeML/zu6Y1155ZW2YsWK/q82fpG5c+fapk2b7JNPPhnw60uXLrVQKGSXXHJJQtt++umn2/e//32rra0d8BwZGRn9q87H6uidCNatWzfg13//+98P+P958+ZZamqq/eQnP5GP57cNV111lTU3N1ssFvvC1/fogXHOnDlmZvbMM88MGP/CCy9YNBr9Mj8aTjHM2c8wZ5EsmLOfYc4iWTBnP8OcRbJgzn6GOXvyG9bfiPqvRowYYffdd5/de++99uyzz9pPf/pTu/LKK+3yyy+3RYsW2ciRI62lpcU2b95sn3zyib344otmZvbggw/a66+/brNnz7b777/famtrra2tzZYtW2Z///d/bxMnTrS77rrLli5dagsWLLAHH3zQqqur7bXXXrMlS5bYnXfe2X9ng2O1bt06+/a3v21f//rXbcKECZaenm7Lly+3devW2fe+973+31dbW2vPP/+8/epXv7Jx48ZZZmZm/+q4n/nz51tRUZF94xvfsAcffNDC4bA9/fTTtnfv3gG/r6amxu6//3576KGHrLu722666SYrKCiwTZs2WVNTk/3whz/s34bf/OY39pOf/MTOOeccS0lJsRkzZtjChQvtmWeesfnz59t3v/tdO++88ywtLc327dtnK1assGuvvda+9rWv2aRJk+yWW26xxx9/3NLS0uyrX/2qbdiwwR599FHLz8//Uq8bTi3M2c8wZ5EsmLOfYc4iWTBnP8OcRbJgzn6GOZsEhvSfSh8iR/+1/NWrV3+udXd3e2PGjPEmTJjgRaNRb+3atd6NN97olZWVeWlpaV5FRYV36aWXek8++eSAcXv37vVuv/12r6KiwktLS/Oqqqq8G2+80WtoaOj/Pbt37/b+6q/+yisuLvbS0tK8M844w3vkkUf671bgef9xl4FHHnnkc9tmZt7ixYs9z/O8hoYGb9GiRd7EiRO9nJwcLzc315s8ebL34x//eMDdCXbt2uXNmzfPy8vL88ys/1/7P3qXgRdffPELX6NVq1Z5s2bN8nJycryRI0d6ixcv9n72s58NuMvAUUuXLvXOPfdcLzMz08vNzfWmTZvmPfXUU/29paXFu+GGG7zCwkIvFAoNuINBJBLxHn30UW/KlCn94ydOnOh961vf8rZt29b/+3p7e727777bKysr8zIzM70LLrjAW7lypVddXT0s7zIw3DBnmbNILsxZ5iySC3OWOYvkwpxlzia7kOd53gle6wIAAAAAAAD4N6IAAAAAAAAQDBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQiPCx/sZQKHQitwPAv/M877g8DnMWCAZzFkguzFkguTBngeRyLHOWb0QBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBDhod4ADJ1QKJTQeM/zjtOWAAAAAACA4YBvRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBDhod6AU10oFBrqTRg0z/Nkd/1srvEATh0cDzBUXPteVlaW7GlpabKPHj3at6Wnp8uxLi0tLbI3NTXJ3tPTI3s0Gv3S2wScyjhXAaeWEz2nw2G9XKKuA1zXF+3t7YPaplMF34gCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgWIgCAAAAAABAIMJDvQHDXWpqakLjPc/zbaFQKKHHPtHjY7HYoMeqn9v12MCpyjVvXMebjIyMQfecnBw5NiVF/7lHa2ur7JFIRPbu7m7ZcXJT+25aWpocW1JSIvuECRNkv/DCC2U///zzfduZZ54px2ZnZ8vuOpe1tLTI/v7778v+/PPP+7Z169bJsW1tbbJznsVQcZ1PVA+H9Ucf13nS1V3nYde5rK+vz7clOudcxxsgGbnmnGvOZ2Zmyl5YWCi7uv7Nzc2VY12OHDkiezQalX379u0JPf+JxjeiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEIjwUG/AySAUCsnueZ5vS01NlWPDYf0SZ2Vlye56/JQU/7XESCSS0GPn5eXJnpmZKXtZWZnsbW1tg2pmZs3NzbL39vbKHo/HE+rAYLmON4qa72buOTl+/HjZ58+fL3t5eblvcx3L2tvbZV+5cqXs69evl33//v2yd3d3y44Ty3W+Ub2qqkqOHTNmjOwTJkyQ/fzzz5d95MiRvk1dH5i554XrGiEnJ0f2efPmyb5v3z7f9umnn8qxwGC5znOuc1l2drbs6enpso8YMcK3qfOYmVlhYaHssVhM9vr6etl7enpk7+zs9G2tra0JPbZr213HM+BEUccE1/HCdZ51fZ6tqamR3XUNUlpa6ttcc66jo0N215x2fdZ39d27d8t+ovGNKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAAQiPNQbcDLwPE/2UCjk29LS0uTY/Px82SsrK2UvKSmRvaenx7e1t7cPeqyZWXZ2tuyuny0zM1P2wsJC3+baNtfr3tfXJ7uLes/N3PsMhi/XvpOS4r/+n5qaKsemp6fLPmXKFNmvuuoq2b/61a/Kro5Hrvnuel3OP/982ZctWyb7Cy+8IPuOHTtkx4mVyHm2u7tbju3q6pLddS585ZVXZN+3b59vc51rampqZL/22mtlP+ecc2TPy8uTfdasWb7tF7/4hRzLeQ6KmrOuc1VBQYHsZWVlsk+dOlX2Sy+91Lepa08zs0mTJsleX18v++HDh2XfuXOn7IcOHfJt+/fvl2Pffvtt2dva2mR3HUsTvbbG8OW6BlSfOV2fhSdOnCj7tGnTZC8vL5fddczYu3evb3MdD1zrAK4557r2jsVisu/evVv2E41vRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQ4aHegGSXkqLX8kpKSmQfN26c7LFYTPbGxkbf1tDQIMd2d3fLnpmZKbvrZ8/Ly5M9IyNj0I8dj8dl9zxP9kSFQqEhe24MLde+mZaWJvuIESN825gxY+TYmTNnyn7FFVfIXlxcLHtubq7siut4kZ2dLXttba3srmPprl27ZN+7d69v6+vrk2ORONcxOxKJ+LbW1lY5Vh2PzcwOHToku+tc2NHR4dvUdpuZvffee7LX1dXJ/r/+1/+S3TWn29vbfVtvb68cy7kMipp36jxnZnbRRRfJftNNN8k+YcIE2SsrK32b6xztOsePGjVK9s7OTtlnzZol+759+3zb4cOH5diFCxfK/s4778j+2muvyb5lyxbZOZcOX4l+Jrz00kt921//9V/Lseeee67sqampsu/fv1/2bdu2ya4+q7teF9f1S0VFhez5+fmyt7W1yb5161bftnPnTjn2eOAbUQAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAhEeKg3IBmEQiHflpOTI8eedtppsldUVMje1NQke3Nzs29ra2uTY2OxmOyRSER21+NPmTJF9nDYf/fr7OyUY12vS1dXl+ye5yXUcepS893MLCsrS/azzz5b9iuuuMK3XXjhhXJsVVWV7Pn5+bJ3dHTI7prTPT09vs31upx++umyu8Tjcdmzs7Nlj0ajCT0/Tiz1/rrOVe3t7bKnpOg/c+vu7pbdte8lMtY1J0eMGCF7Xl6e7EeOHPFtiZ4ncWpznQuLiop823e/+1059i/+4i9kLykpkd01p13brrjOFX19fbK75o269jUzGzt2rG9zXZerc7RZYscLM7P6+nrZDx06JDuSl2tOufathQsXyn733Xf7tpqaGjnWxXWN0NraKvvWrVsH3V1zynXtXFxcLLtrHcLFdX11ovGNKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABELfQxRmZpaamurbysvL5dhx48bJ7rqN67Zt22Rvbm72bYnerjzRW9AWFBTInp6e7ttct+ZN9HaT3JZ6+FL7nZn7VqkLFiyQ/dZbb5V99OjRvs21ba6ujgfH0t966y3Z16xZ49tct/a95pprZJ88ebLsjY2NsrtukRuPx2XHycv13rluae46V53I80FmZqbsd955p+yVlZWyd3V1yf7ee+/5tu7ubjkWpzbXMbu0tFT2n//8575t5syZcmxubq7srjnd29sruzrX7d69W451XV8eOnRIdtd5urq6Wnb12lRUVMixhYWFstfU1Mh+5plnyu5631yvDU5ers9deXl5sl977bWy/8M//IPsVVVVsiuu67833nhD9tdff132DRs2yK72e9fxJC0tTfaysjLZXe9LR0eH7EM9Z/lGFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAAIRHuoNOBmEQiHZs7OzfVtVVZUcm5eXJ/vGjRtlP3TokOx9fX2+zfM8OdbF9bpkZGTInpWVNejHD4f1rhmJRGTHqc21b6p9b/r06XLslVdeKfvf/M3fyO6a8x0dHb6tt7dXjj148KDsO3bskH3JkiWyf/rpp7J3dXX5tpycHDnWNWfVsexYtLe3JzQeJy/XuSwej8sei8Vkdx1PlNTUVNldx5sbb7xRdte50HWNsGzZMt8WjUblWCS3lBT9Z82uc9U//MM/yH7ZZZf5Ntd+65qT6jxpZvb222/L/oc//MG37dq1S45V5zkz95wfMWKE7KWlpbKPGjXKt91www1ybE1NjewFBQWyjx07VnbXZ56dO3fKjqGljgmufWPWrFmyf+c735G9pKRE9u7ubt925MgROfY3v/mN7M8++6zsrv3WdX2qrt1dx+G0tLSEntt1PHIda4f68zTfiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgwkO9ASeDcFi/DKWlpb6tsLBQjt23b5/sdXV1sre1tcnueZ7sSigUkj0lRa9TlpSUyJ6dnS17X1+fb2tpaZFjY7GY7PF4XHYMLde+5+qufWvy5Mm+7b777pNjZ8+eLXtmZqbsHR0dsh86dMi3rVmzRo7ds2eP7B999JHsW7Zskd11vIlGo4Nqx/LYOTk5sruOdUeOHJEdpy7XvuHaN1NTU2VPS0vzbXl5eXLsAw88ILvreNLT0yP7E088IXtDQ4PsOHUVFBTI/jd/8zey33HHHbKnp6f7Ntec3LFjh+y//OUvZX/ttddkb2xs9G2uOXUijxdmZpFIRPaamhrfdtppp8mxVVVVsufm5speXV0te21trezvvvuu7BhaGRkZvk1dN5uZ/bf/9t9kHzlypOydnZ2yHzx40Le98847cqzreLFr1y7Zu7q6ZHd9plSfSRP9vOPqrtdVfdY2G/rPy3wjCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgWAhCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgWAhCgAAAAAAAIEID/UGBCEUCsmemZkpe1FRkW+rrKyUY7du3Sp7fX297JFIRHbP82RXXK9LXl6e7GeeeabsEydOlL2urs639fT0yLGxWEx2nNxc+15aWprsI0eOlP3222/3bTNnzpRjMzIyZI9Go7J3dnbK/tFHH/m2NWvWyLH79++XfceOHbK75o3rfUlNTfVtrtetrKxMdtexrK+vT/ampibZcepK5DxoZpaeni772Wef7dvmzp0rx15wwQWyu7Z9/fr1sv+///f/ZI/H47IjeaWk6D9LnjNnjux33nmn7NnZ2bKr80VjY6Mc+9Of/lT2Z599VnbXeVadp13ncNd5MFGu9019Lvjggw/k2PPOO0/2cFh/7CsoKJC9sLBQdvXaJXqchptr3yopKfFtCxYskGPHjx8vu2vedHR0yK6uX13Xxq7zXH5+vuyuedHV1SW7urZW181m7jUI17xxXde71hGG+hqBb0QBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgEOGh3oDjIRQKyZ6WliZ7UVGR7KWlpb4tFovJsQcOHJC9o6ND9ng8LrvrZ1cyMzNlHzdunOwXXXSR7K7XdefOnb4tJUWvkXqel1DHieV6/8JhfeiprKyU/ZJLLpH9zDPPHPRz9/b2yt7d3S3722+/Lfu//du/+baVK1fKsdFoVHbX6+46XrlkZWX5tlGjRsmx06ZNk931vnzyySeyHzx4UHYMX67zZEFBgeznnHOOb/v6178ux7rO4eo8aGb2/e9/X/YjR47IjlNXXl6e7N/85jdld51nXeeTSCTi215//XU59tlnn5W9ublZ9hN5DZjIdfWxcG2bmtP19fVyrOszR2Fhoeyu99yFa+8Ty7Vvuj7XjR8/3re5PvPl5OTI7vqsrY4XZnrfcV1fup67vb1ddtf14759+2R3fZZXUlNTZXd95nB9Lkj0uv9E4xtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIhL5fdpJw3fqwqKhIdtctK8877zzf1tnZKccmeht7V1e3u1S3Wjczq6mpkf3yyy+X/bTTTpO9p6dn0N11y+sTfXtdnFj5+fmyu/atM844Q/bc3FzftmPHDjn28OHDsj/99NOy//GPf5Rd3Zbadetj1y1qXfr6+mR33fq3vLzct82ePVuOdd1+13X73A8++ED2rq4u2TF8uc4XxcXFss+YMcO3VVZWyrFtbW2yP/jgg7K///77smP4UrdiN3OfJzMyMmR3nY9aWlp8269+9atBjzVzXwOeSK6fO1Gu41Ei17eu99TFdf2zf//+hB4fiXHtG3l5ebLn5OT4tsbGRjnWNWdd57q6ujrZt27d6tuys7Pl2NGjR8teUFAgezQald11fdnb2+vbXNft3d3dg35sM7NIJCL7UB5LjwXfiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgwkO9AcdDRkaG7GVlZbLPmDFD9ksvvdS3FRUVybGjRo2S/b333pP94MGDspeXlw/6uadOnSr7uHHjBv3cZmY7d+6U/dChQ74tHo/LsZ7nyY6hlZaWJntubq7sc+bMkT07O1v2P/3pT75t9+7dcuwf//hH2bdt2yZ7b2+v7EooFJI9HNaH7L6+Ptld78vo0aNlr62t9W0lJSVyrGvbN2/eLPuWLVtkj8VismP4Ki4ulv22226TXR2PXPN9zZo1sq9bt072RI4nOLWNHDlS9szMTNld11HRaFT2uro637Zp0yY51nWNdyK5zrOu7tp213jXefi0007zbXPnzpVjq6qqZHe9p+vXr5fd9b7ixEpNTZXdte81Njb6to0bN8qxhw8flv3AgQOyt7a2yq5+NtdnBtfr4rr+dF37pqTo7+2ox3f93K7rdtfxJtHj0VB/nuYbUQAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAhEeKg3wMwsFArJHg7rzSwqKpK9tLRU9vLyctlHjRrl20aOHCnHnnnmmbJfe+21snd1dcmenp7u22KxmBzb2dkpe35+vuyHDx+Wfffu3bJv2bLFt7W2tsqxnufJjhPLNWczMjJknzRpkuypqamyNzc3y75ixQrftmrVKjn2yJEjsrvmlYvrtVMikYjsaWlpspeUlMg+a9Ys2S+88ELf5nrPt2/fLrvrfXEdb+LxuOw4dZWVlcm+ePFi2W+55RbZ1b795ptvyrGvvvqq7K7zJODHtd+7zqOuY2Y0GpV9w4YNg37sRKWkDP7P0V2vi6v39vbK7vrMUllZKfvVV1/t2+bMmSPHuq4BXOfR9957T/Zdu3bJjsS4rg9dn31c+6763Pfuu+/KsS5tbW2yu+asuj4tKCiQY13rANnZ2bJXVFTIPmbMGNlzc3N927Zt2+RY1+vmes+T/fMw34gCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgWIgCAAAAAABAIMJDvQFmZp7nyR6NRhPqoVBI9v3798v+pz/9ybddcsklcmxhYaHs6enpsnd2dsq+a9cu33b48GE5tru7W/a8vDzZI5GI7B9//LHs27dv921HjhyRY+PxuOw4sVxztqKiQvbZs2fLPnPmTNl/8YtfyL5+/Xrf5tq3YrGY7IlSr11qaqocm5GRIXtJSYns119/vexXX3217MrGjRtl/9d//VfZd+/eLbvrOI/kps7TVVVVcuyPf/xj2b/2ta/JHg7rS6G9e/f6tueee06Ora+vl90151NS9J8Xci4cvvbt2yd7R0eH7MXFxQmNV9eYrmsE15xz7deu6/qcnBzf5rrudj22i+t4ddddd8k+Y8YM31ZWVibHNjY2yu66Ln/llVdkb21tlR2aa99yzYvs7GzZXfNOXf+65lxPT4/siV6jqXOda8729fXJ7nrdxo4dK3taWprsH374oW9zfRZ3fSZxce1Trn1iqPGNKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAAQiPNQbcCw8z5O9vr5e9s7OTtm3bdsm+7p163zbypUr5dhp06bJ3tzcLPvWrVtl37Fjh29LTU2VY8877zzZJ02aJHt7e7vsdXV1sh8+fNi3RSIROda1T7g6EhMKhWSvqqqS/YILLpC9pqZG9tLSUtmj0ahvS0nR6+/xeFx2F9dro54/JydHjp0xY4bs3/jGN2S/8MILZXc9/4YNG3zbCy+8MOixZmZHjhyRPdH3BSe3/Px83/bDH/5Qjr3mmmtkD4f1pU5XV5fsS5Ys8W2uc/i4ceNkd23b5s2bZT9w4IDsrnMpkldTU5Ps+/fvl72goEB2176jrhFnzZolx7777ruyu67hiouLZS8sLPRtrjnpeuyLLrpI9ilTpsheUVEhu7qGcM33999/X/Zf/vKXsrsePxaLyQ7Ndf2ZlZUl+9ixY2XPzs6WXV1HueZ7W1ub7B0dHbJnZmbKnpub69tc1/yua+PJkyfLXl5eLntra6vs6jOHa06pscfSk/3amG9EAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBDhod6A48HzPNkPHz6cUN+zZ49ve+edd+TY3/72t7LH43HZu7u7ZVc/e35+vhxbVFQk+6RJk2RvbW2Vvb6+Xvaenh7ZFdd7jqFVWVkpe3V1teyufXfChAmyl5SU+Lbe3l45NhqNyh4KhWTPyMiQvba21rdddtllcuyiRYtkLy4ult21bfv27ZP9Rz/6kW9bu3atHNvZ2Sm761iI5Jaamir7tGnTfNu8efPk2HBYX8pEIhHZX3rpJdnXrFnj22bMmCHHzpo1S/by8nLZV61aJfsTTzwh+5YtW3xbLBaTYznPntx27dolu+v6s7CwUPa8vDzZx44d69vuuusuOfbqq6+WvbGxUXbXvCktLfVtrmvbzMxM2XNycmR3nWdd86qhocG3vfHGG3Lsiy++KPv69etld10fITGu60fX5zJ1njQzq6iokL2vr8+3tbS0yLGuOen6LO2aF5deeqlvO//88+XYmpoa2V3HMtd1v2vevP32276tq6sroec+1a+N+UYUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACoe95DDPTt1pVt8I0Mzt06JDsrltau6jxrlvEum5/67q1s+uWlN3d3bKnpLAOmqzS09Nlz8/Pl921b2ZnZ8t+ySWXyL527VrftnHjRjnWtd+OHz9e9ssvv1z2iy66yLdVVlbKsa7X3XWb+gMHDsh+7733yr58+XLf1tPTI8dyq/jhzbXvzps3z7eVlJQk9NyuOe2aN7Nnz/Ztl112mRw7YcIE2V23ile3oTcz27Vrl+xqXu7du1eOHerbSnNM0Do6OmR/4403ZC8oKJD9q1/9quzqXOjab8866yzZXXMyFArJnpubO+ixifbe3l7ZGxoaZH/rrbd829KlS+XYrVu3yt7W1iY7cy4xiX6ucR1zXdfGEydOlD0nJ8e3ZWRkyLGu431hYaHso0ePll0dM1znSZfOzk7Zt2zZIvvzzz8v+759+3yb63jgel1P9TnJSgAAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACER7qDTjVeZ4neywWkz0UCsmemprq2zIzM+XYESNGyO4an5eXJ3t6errsKSn+66CqmblfV1dHYiKRiOytra2yd3R0yN7X1yd7VVWV7D/84Q9925YtW+RY175XXV0te2FhoexpaWm+zTXfXa/rmjVrZL/33ntlX79+vezqfWdODm+ueeOaszNmzPBt4XBilyqubfvKV74i++HDh31bRUWFHOva713H0q6uLtkrKytlV6+76zjc3t4uu+s4HY/HZUdiXK//zp07ZX/mmWdkz87Olj0rK8u3uc6T6trVzH396TomqHnnmpPd3d2yNzQ0yL5hwwbZP/jgA9mXLVvm2w4cOCDHqmOVGefhE811zHOdi1zng6amJtldnynHjx/v29R8NjMrKCiQ3XXtm5GRIbviel1d+/1HH30k+2OPPSb7+++/L3tvb69vc237cJ+TfCMKAAAAAAAAgWAhCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgWAhCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgQgP9QYgMSkp/muJubm5cmw0GpW9vr4+od7X1yd7amqqb1M/l5lZPB6X3fM82ZEY1+u/YsUK2c8991zZ586dK3tNTY3soVBo0GObm5tl7+zslD0Wi8ne2Njo21atWiXHPv/887KvXr1a9t7eXtld264w54Y3NefMzNLS0mRX+09bW5sc6zpfuM5FrvHhsP+lkmvbmpqaEurvvvuu7L/+9a9lV8ebnp4eOTaR4wFOPNcxt7u7W/a9e/fK/sQTT8j+8ccf+7bLLrtMjq2trZU9Pz9f9oKCAtnV9eXOnTvlWNd5+MMPP5S9rq5Odte1c2trq29L9PqD8/TQcl07d3R0yP7RRx/Jrs5VZmaHDx/2baeffrocm5eXJ3t1dbXsrjmt9u1PP/1Ujv3d734n++9//3vZDx48KLvrXKneV+acxjeiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEIiQ53neMf3GUOhEb8uw5HpdU1L0WmF6erpvq66ulmNPO+002UtLS2VvaGiQfd26dbK3t7f7tp6eHjk2FovJfoy79UnpeG37UM5Z13OPGzdO9rlz58peW1sre2FhoW9ra2uTY/fv3y+7a79fvXq17Hv27PFtR44ckWNd+0Yy7/fJ7FSYs4lybXtBQYHsM2fO9G3nnHNOQs/tOp90dXXJvmvXrkE1M7NwOCy7Og+amTU2Nsre29srezwel/1EjT3ZMWfd2+7ad7Ozs32b6/px1KhRsrvGV1RUyL5p0ybf9umnn8qxnZ2dsruOJ4kej9T4aDQqx57KhsOcdX3my8jIkN11nlXXxmVlZXJsVlaW7JmZmbK7rr337dvn2w4dOiTHuuZFX1+f7K5zHdfWg3MsrxvfiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgQp7necf0G0OhE70tw5LrdXX1cDjs27Kzs+XY3NzchLpLfX297J2dnb4tGo3Ksce42yal4/Wzncxz1rVt6enpsmdlZQ16fCwWk2P7+vpk7+npkX0477vD1XCYs0NpqF8X5uyphznrlpKi/6xa/eyu18V1Dnedp13vXyQSGfRju7Y90X3H9fjxeDyhxz9VMWcT/8yYyJxNdF649mv2+1PPscxZvhEFAAAAAACAQLAQBQAAAAAAgECwEAUAAAAAAIBAsBAFAAAAAACAQLAQBQAAAAAAgECwEAUAAAAAAIBAsBAFAAAAAACAQIQ8z/OO6TeGQid6WzAI6n1JSdHrjOFwOKEej8dl7+vrG/T4Y9wtT0nH62dnzg6O63UbzvsmvhhzFkguzFkguTBngeRyLHOWb0QBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQIe8Y74fJ7S6BYHCLWiC5MGeB5MKcBZILcxZILscyZ/lGFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAhHyPM8b6o0AAAAAAADAqW9YfiPq6aeftlAo1P9fOBy2yspKW7hwoW3btm2oN++4WrJkiT399NNDug3PPvusPf744yfksWtqamzRokUn5LFx8mDOBos5i0QxZ4PFnEWimLPBYs4iUczZYDFnj79huRB11FNPPWUrV660N954w7797W/b7373O7vwwguttbV1qDftuDnVJy6GF+ZsMJizOF6Ys8FgzuJ4Yc4GgzmL44U5Gwzm7PEXHuoNGEpnn322zZgxw8zM5syZY7FYzBYvXmy//e1v7bbbbhvirQteJBLpX1EHTkbM2YGYszjZMWcHYs7iZMecHYg5i5Mdc3Yg5mzyGNbfiPqvjk7ihoaG/l/76KOP7JprrrGioiLLzMy0adOm2QsvvPC5sfv377c77rjDRo8ebenp6VZVVWU33HDDgMfas2eP3XLLLVZWVmYZGRk2adIk+9GPfmTxeLz/9+zatctCoZA9+uij9thjj9nYsWMtNzfXZs6caR988MGA59y5c6ctXLjQqqqqLCMjw8rLy23u3Lm2Zs0aM/vsa34bN260P/3pT/1f26ypqTEzs7feestCoZD98pe/tLvvvttGjhxpGRkZtn37dvvBD35goVDocz/j0a+A7tq1a8CvP/vsszZz5kzLzc213Nxcmzp1qv385z83s88OiK+99prt3r17wNdHj+rr67N/+qd/sokTJ1pGRoaVlpbabbfdZo2NjQOeIxKJ2L333msVFRWWnZ1tF154oa1atcrvrcQwwZxlziK5MGeZs0guzFnmLJILc5Y5myxYKvxP6urqzMzs9NNPNzOzFStW2BVXXGHnn3++Pfnkk1ZQUGDPP/+8/eVf/qV1dXX1/13O/fv327nnnmuRSMTuv/9+mzx5sjU3N9sf/vAHa21ttfLycmtsbLRZs2ZZX1+fPfTQQ1ZTU2Ovvvqq3XPPPbZjxw5bsmTJgG154oknbOLEif1fAfzHf/xHmz9/vtXV1VlBQYGZmc2fP99isZg9/PDDNmbMGGtqarL333/f2trazMzs5ZdfthtuuMEKCgr6Hz8jI2PA89x33302c+ZMe/LJJy0lJcXKysq+1Gv2wAMP2EMPPWTXXXed3X333VZQUGAbNmyw3bt3m9lnX6W84447bMeOHfbyyy8PGBuPx+3aa6+1d955x+69916bNWuW7d692xYvXmxz5syxjz76yLKysszM7Jvf/KYtXbrU7rnnHrvssstsw4YNdt1111lHR8eX2l6cWpizzFkkF+YscxbJhTnLnEVyYc4yZ5OGNww99dRTnpl5H3zwgReJRLyOjg5v2bJlXkVFhTd79mwvEol4nud5EydO9KZNm9b//0ddddVVXmVlpReLxTzP87zbb7/dS0tL8zZt2uT7nN/73vc8M/M+/PDDAb9+5513eqFQyNu6davneZ5XV1fnmZlXW1vrRaPR/t+3atUqz8y85557zvM8z2tqavLMzHv88cflz3rWWWd5F1988ed+fcWKFZ6ZebNnz/5cW7x4sfdFu8bR162urs7zPM/buXOnl5qa6t18881yGxYsWOBVV1d/7tefe+45z8y8l156acCvr1692jMzb8mSJZ7ned7mzZs9M/PuuuuuAb/vmWee8czMu/XWW+XzI/kxZ5mzSC7MWeYskgtzljmL5MKcZc4mu2H9V/MuuOACS0tLs7y8PLviiitsxIgR9sorr1g4HLbt27fbli1b7OabbzYzs2g02v/f/Pnz7eDBg7Z161YzM3v99dftkksusUmTJvk+1/Lly+3MM8+08847b8CvL1q0yDzPs+XLlw/49QULFlhqamr//0+ePNnMrH9ltqioyE477TR75JFH7LHHHrM///nPA74Seayuv/76Lz3mqD/+8Y8Wi8Xs7/7u7wY1/tVXX7XCwkK7+uqrB7y+U6dOtYqKCnvrrbfM7LOVfDPrfy+OuvHGG/n7v8MMc5Y5i+TCnGXOIrkwZ5mzSC7MWeZsshrWC1FLly611atX2/Lly+1b3/qWbd682W666SYz+4+/V3vPPfdYWlragP/+9m//1szMmpqazMyssbHRRo0aJZ+rubnZKisrP/frVVVV/f0/Ky4uHvD/R7+C2N3dbWZmoVDI3nzzTbv88svt4YcftunTp1tpaal95zvf+VJf7/uibTpWR//eq+tn99PQ0GBtbW2Wnp7+ude4vr6+//U9+tpUVFQMGB8Ohz/3OuHUxpxlziK5MGeZs0guzFnmLJILc5Y5m6yG5/Lbv5s0aVL/P+h2ySWXWCwWs5/97Gf261//2mpra83ss79zet11133h+DPOOMPMzEpLS23fvn3yuYqLi+3gwYOf+/UDBw6YmVlJScmX3v7q6ur+f0Tt008/tRdeeMF+8IMfWF9fnz355JPH9Bhf9I+4ZWZmmplZb2/vgL+De3QiHVVaWmpmZvv27bPRo0d/6e0vKSmx4uJiW7Zs2Rf2vLw8M/uPg1h9fb2NHDmyv0ej0c8d8HBqY84yZ5FcmLPMWSQX5ixzFsmFOcucTVbD+htR/9XDDz9sI0aMsAceeMAmTJhgEyZMsLVr19qMGTO+8L+jO9aVV15pK1as6P9q4xeZO3eubdq0yT755JMBv7506VILhUJ2ySWXJLTtp59+un3/+9+32traAc+RkZHRv+p8rI7eiWDdunUDfv33v//9gP+fN2+epaam2k9+8hP5eH7bcNVVV1lzc7PFYrEvfH2PHhjnzJljZmbPPPPMgPEvvPCCRaPRL/Oj4RTDnP0McxbJgjn7GeYskgVz9jPMWSQL5uxnmLMnv2H9jaj/asSIEXbffffZvffea88++6z99Kc/tSuvvNIuv/xyW7RokY0cOdJaWlps8+bN9sknn9iLL75oZmYPPvigvf766zZ79my7//77rba21tra2mzZsmX293//9zZx4kS76667bOnSpbZgwQJ78MEHrbq62l577TVbsmSJ3Xnnnf13NjhW69ats29/+9v29a9/3SZMmGDp6em2fPlyW7dunX3ve9/r/321tbX2/PPP269+9SsbN26cZWZm9q+O+5k/f74VFRXZN77xDXvwwQctHA7b008/bXv37h3w+2pqauz++++3hx56yLq7u+2mm26ygoIC27RpkzU1NdkPf/jD/m34zW9+Yz/5yU/snHPOsZSUFJsxY4YtXLjQnnnmGZs/f75997vftfPOO8/S0tJs3759tmLFCrv22mvta1/7mk2aNMluueUWe/zxxy0tLc2++tWv2oYNG+zRRx+1/Pz8L/W64dTCnP0McxbJgjn7GeYskgVz9jPMWSQL5uxnmLNJYEj/qfQhcvRfy1+9evXnWnd3tzdmzBhvwoQJXjQa9dauXevdeOONXllZmZeWluZVVFR4l156qffkk08OGLd3717v9ttv9yoqKry0tDSvqqrKu/HGG72Ghob+37N7927vr/7qr7zi4mIvLS3NO+OMM7xHHnmk/24Fnvcfdxl45JFHPrdtZuYtXrzY8zzPa2ho8BYtWuRNnDjRy8nJ8XJzc73Jkyd7P/7xjwfcnWDXrl3evHnzvLy8PM/M+v+1/6N3GXjxxRe/8DVatWqVN2vWLC8nJ8cbOXKkt3jxYu9nP/vZgLsMHLV06VLv3HPP9TIzM73c3Fxv2rRp3lNPPdXfW1pavBtuuMErLCz0QqHQgDsYRCIR79FHH/WmTJnSP37ixInet771LW/btm39v6+3t9e7++67vbKyMi8zM9O74IILvJUrV3rV1dXD8i4Dww1zljmL5MKcZc4iuTBnmbNILsxZ5myyC3me553gtS4AAAAAAACAfyMKAAAAAAAAwWAhCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgWAhCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgQgf628MhUIncjsA/DvP847L4zBngWAwZ4HkwpwFkgtzFkguxzJn+UYUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAAIRHuoNgBYKhWTPzs72bSNHjpRjL774YtknT54s+86dO2X/8Y9/LDsAAMksJSWxP8+Lx+PHaUsAHAvXnHV113W5a07HYjHZAWC44BtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIRHioNwCa6zayeXl5vm327Nly7Ne+9jXZS0tLZS8vL5e9p6dH9ldeecW3HTx4UI71PE92AACOhet27Oo8nJubK8dWVlbKXlJSklBvamqSfe3atb7tyJEjciznWSgFBQW+TV2bmpmNGzdO9tGjR8s+YcIE2bu7uwfVzMzq6+tl7+vrk911/RqNRmXv6Ojwbc3NzXKsqwPAyYRvRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQ4aHegOEuFArJnpeXJ3ttba1vW7BggRw7atQo2VtaWmTfsGGD7G+//bbszc3NsgNIHq5jmYvneQk9vms8hi/XvhMO60uhESNG+LZJkybJsVddddWgH9vMrL29XfY33nhD9vT0dN+WlpYmx/b19cmOk1tKiv6zZte8GDlypOwTJkzwbRdccIEce/7558vumleZmZmyu+a0ouaMmXtebNu2TfaVK1fKrua06+fOysqS3XXd3dvbK3s8HpcdQ8s153n/cLLhG1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIRHioN2C4S0nRa4FVVVWyL1y40LeVlJTIsW1tbbIvW7ZM9t/85jey7927V/ZIJOLbPM+TYwEELxQKDaodC9exMNHe29v7pbcJycG176Wmpso+YsQI2efOnevbzjvvPDn24osvlr2xsVH2PXv2yL5t2zbZlXg8LrtrTrnG48RyvT+u/T4rK0v2srKyQXfXnOro6JB9x44dsru2XR0TiouL5dicnBzZCwoKZJ8yZYrsrvdFzfnW1lY51vW6qutuDL1Er6MSeXw+d2Eo8I0oAAAAAAAABIKFKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABCI81BtwqguFQrJXVVXJ/j/+x/+Qfc6cOb6ttbVVjl26dKnsv/jFL2Rvb2+XPR6Pyw4gWK7jUWpq6qDHJ/rY6enpsufn58vusnfv3oTGI3m59r3c3FzZzzzzTN82Y8YMObawsFD25uZm2devX5/QeHWe5hyd3DzPk911TO7r65O9qalJ9iNHjvi2TZs2ybGRSER217alpOg/R+/s7PRtJSUlcmxRUZHsl156qezjx4+XvaysTPbMzEzf1tPTI8e6XlfXPpPoPuUaj8S4Xv9Exrse29Vdc9JF7TuxWGzQY48H9vsTh29EAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBDhod6AU104rF/ia6+9VvbZs2fLnpOT49v+7d/+TY79v//3/8re1dUlu+d5sgN+QqGQ7MN133K9Likp+s8OUlNTZU9LS5NdHU/MzDIzM31bPB6XYwsKCmQfM2aM7FVVVbL39vbKvmfPHt/29ttvy7E4ubmOF+np6bJPmjRJ9rPPPtu3ufb7bdu2yb5mzRrZ//znP8ve0dEhu2v7kLxc15eu84XrfFNfXz/o8dFoVI51cc1p17ar3tTUJMe6XlfXtrn61q1bZc/OzvZtPT09cqzrPU/02mq4XpsdL67350RT14hZWVlyrOv6MSMjQ3a1X5vpfdt1novFYrK79lvXeTLRriS67S4n+5zlG1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIRHioN+BUV1paKvv1118v+9ixY2XfuXOnb/vnf/5nObazs1N2YLBCodBQb8KQcf3saWlpvq2oqEiOraiokD0lRf/ZQklJiezl5eWyp6am+rb09HQ51nUsy87Oln3cuHGye54n+7vvvuvb3n77bTkWJze1X5qZFRcXyz558mTZ1b7V3t4ux65du1b25cuXy97U1CR7NBqVHcNXLBZLqLuOqa6uJHqN4Bqvts11nnQdT44cOSJ7PB6XvaamRnZ1XZ+RkZHQc7sk8p7CbOLEibLv379f9kgkIrvr/XHtH6NHj/ZtEyZMkGMLCgpkz83Nlb2rq0t2dTzq6OiQY1taWmR3zflDhw7J7jrPHj58eFDtWB67r69P9mSfs3wjCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgWAhCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgQgP9QYkO9dtXu+8807Z58yZI7vrto7/83/+T9+2d+9eORY4WSVya+YT/dyu28Cq2+Oamc2cOdO33XDDDXKs6zb0rtvEum7X7bpFrnptXLcddnEdS9va2mTv7OyUfc+ePV92k3CScM3J9PR02b/yla/Ifu2118peVlbm21xzbuPGjbK7bjvtugbA8OU65rrmjcuJPM+6HjvRbQ+H/T/euI4XrnP4ueeeK/uUKVNkd93Gvrm52be5znOuc3yy3+r9ZOe6RmttbZW9r69P9oKCAtlPP/102a+++mrfNn36dDnWpb29Xfa8vDzZi4qKfJvr2tR1XV5XVyf75s2bZXcda3ft2uXbPvzwQzn24MGDsruuAVxz/mTHN6IAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQiPBQb0CyKy0tlf2v//qvZY/FYrK/+eabsr/00ku+LR6Py7HAiRIKhRIa73leQo+vumtsRkaG7NOnT5f95ptvHnR3zdn9+/fL3t7eLvvHH38se3p6uuwHDx70bd3d3XKs6z3t6emRvaWlRfauri7Zm5ubZUfyKigokL2qqiqh8eXl5b7Ntd+Hw/oyy3U8cl0jAH5cx9yTWUpKYn9OruZ0TU2NHHvTTTfJXltbK7vrXLZhwwbZ165d69uOHDkix0ajUdmTeZ9IBu+9957s1dXVsmdmZso+fvx42W+99VbZZ82a5dtc5zJ1/Wfm3jf37t076PE5OTlybFZWluyua+fKykrZ8/PzZc/NzfVtu3fvlmNbW1tld70vyY5vRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQ4aHegGQQCoV827x58+TYgoIC2bdu3Sr7Qw89JHt7e7vsJ1JKil7HdPV4PD6ohpOf6733PC+hx09NTR302NzcXNmnTp0q+3//7/9d9ssuu0z2rq4u3/baa6/JsW+99Zbsa9eulf3IkSOyZ2Zmyh6NRn1bdnb2oMeaubdNvW7H0nt6emRH8nK9tzk5ObKXlJTInpWV5dtKS0vlWNc1QEdHh+yJHiuBk5G6rj4Wrnl1zjnn+LZbbrll0GPNzBobG2XfsmWL7Js3b5Z9+/btvs11nuTa+eTmen9c+3VlZaXsrvOJ+sy5fv16OdZ1fXnw4EHZe3t7ZVfXiJFIRI7Nz8+X3fWZwXXdP2bMGNnVNcLo0aPlWDXfzRI/Vp7s+EYUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAhEe6g1IBpmZmb7t1ltvlWMbGxtlf+mll2Rfs2aN7J7nya6kpOh1yNzcXNkLCgpkj8fjsre1tfm2SCQix0aj0YSeGyeW6/UPhUIJddd+n5OT49vOOussOfbee++V/Stf+Yrsrm175ZVXfNuLL74ox65bt05217xxva7Z2dmyq2PhoUOH5FjXtrmOR+p4YWbW29srO8eE5OWaU6733nUezsvLk901b5QDBw7InpGRMejHBpKV63jvOhdNnz5d9kWLFvm2efPmybGu443reLJy5UrZXdf13d3dvi0Wi8mxiXwmwInX09Mj++jRo2VvaGiQvb6+XvYdO3b4trVr18qx27dvl/3w4cOyu67B1L7r2u9dSkpKZHd9nq2urpbddbxSXD9bItcfyYBvRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQ4aHegGQwbdo031ZQUCDHNjQ0yP7OO+/I3tfXJ3soFPJtqampcmxVVZXsc+bMkX3kyJGy9/T0yL5582bftm/fPjl27969snd2dsoejUZlR2Li8bjsKSl6DTwc1oemtLQ02QsLC33bRRddJMfOmjVL9ry8PNnfeust2dWcb2trk2NLSkpkP3LkiOyu1931vinqWGTmPpa55mwsFvvS2/SfeZ6X0HicvFzH846ODtkPHz4suzrPt7a2yrH19fWyu45lrjkLnIxc+21OTo7skydPlv3OO++U/eKLL/ZtRUVFcuyaNWtkf++992RvamqS3XUu6+3t9W1cu57cXPt1JBKR3fW5qby8XPYdO3bIrvafnTt3yrGu/bqrq0t2FzUvXNeXrs+7rnO8a7zr2ljNWdf1h+t44PrZkx1XOAAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACAQLUQAAAAAAAAgEC1EAAAAAAAAIBAtRAAAAAAAACIS+R/owkZ6eLvuECRN8m7plo5n71s179+6V3XXbxuzsbN929tlny7ELFiyQvba2VnbX67Zr1y7Z1S0rZ86cKcdu27ZN9tdee032trY22bnV+4nl2q9dt3527Xvqluiu27Ru2rRJ9oyMDNm3bNkiu7oN7JQpU+TY7du3y56fny+7i+v2vEeOHPFtfX19cqzrtsWu7pqTrtvr4tTleu/37Nkju+t8kpub69taWlrkWNecdN3uWx3LzNzzDjhR1HncdZ50XV/edtttsp933nmyq3nz6aefyrFvv/227B9//LHsnZ2dsrvmLNefyct1HZPo8fzw4cOyd3V1yd7R0eHbenp65NhE99sTeY3m+kzheu7GxkbZXT+7+kyprpvBN6IAAAAAAAAQEBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQiPBQb8DJIDU1VfZx48b5tsLCQjl227Ztsufm5so+evRo2adPn+7brr76ajm2vLxc9oMHD8r+wQcfyL5jxw7ZR40a5dsmTpwox06dOlX2LVu2yL5mzRrZo9Go7EiMa86lpOg18szMTNnj8bhv27hxoxzb1taW0HM3NDTIrvatESNGyLE1NTWyZ2dny75hwwbZw2F9SlDvm+d5cqxrTrnGq/cUUHp6emTv7OyUvaSkxLfl5eXJsa5zeEFBgexpaWmyh0Ih2V3zCvDjOg9nZGT4tgsuuECOXbhwoewXX3yx7K7zyfvvv+/bli5dKseuXr1adtc1gOt445qT6nV3XTvFYrGEnhuJcR2PXe9PS0uL7K73r6ysbNDPr+azmVlWVpbsrms01/FEcV2buri2zfVZ3DXv1LV7JBKRY13v6ak+Z/lGFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAAIRHuoNCEIoFJI9MzNT9tNPP923ZWdny7Fnn3227DfccIPsrm2bOHGibysqKpJjly9fLvu//Mu/yN7a2ip7RkaG7NFo1Lfdcsstcmx+fr7spaWlsru49hnP8xJ6/FOd6/VLSdFr4Onp6Qk9f29vr29bv369HFtXVyd7X1+f7FlZWbKPHTvWt9XX18uxruNJJBKR3fW6u/Zr9fidnZ1yrOt1Y05hsMJhfSlz1llnyV5eXi67Opft379fjo3FYrKfdtppsq9Zs0b2jo6OhJ4fw1ei18YzZ870bfPmzRv0WDOztLQ02V966SXZn3zySd+2Z88eOdZ1PElNTZXdte2u8YWFhb6tq6tLju3u7pZdXXebuc/DnKc11zWYy+HDh2Xv6emRvaWlRXa177muu137tWvexONx2dX1qes85jqWud4X17xwXSOox090n3C9bsmOb0QBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgEOGh3oCTQSwWk/2tt97ybdOnT5djKysrZb/mmmtk7+npkb2zs9O3/epXv5Jjn3vuOdmbmppkd71uubm5so8fP963VVdXy7EHDhyQPRKJyO7ieV5C44e7lBS9xp2amip7Xl5eQuPj8bhvi0ajcmxbW5vs4bA+bKanp8ve19fn28466yw5tqqqSvZDhw7J7vrZXXNevTbq5zJzzynXPsOchJ/8/HzZ/+Iv/kL2UaNGya7Oszt27JBjs7OzZXcdT0pLS2U/fPiw7GrbcXILhUIJddcxNZFrNDOzqVOn+rZzzz1XjnV5+eWXZX/00Udlb2ho8G2uc4nr2jYtLU12df1hZlZcXCy7uoZwXRu1tLTI7vpM4equn83VT3Wun7+3t1d21zWa67NNd3e37IrrXOT62RKdV6q7Htu1ba7rctex8JxzzpF97dq1vs11HE702jnZ8Y0oAAAAAAAABIKFKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABCI81BsQBM/zZO/u7pb9008/9W1bt26VY8855xzZy8rKZF+3bp3sq1at8m0rV66UY10/d2pqquzhsN59pkyZIvvVV18tu9LQ0CB7XV2d7LFYbNDPDTfXvpGdnS37mDFjZHfN6c7OTtmVUCgke0ZGhuyuOX3uuef6tvHjx8uxrjn50Ucfya6OZWZmzc3NskejUdkT4XpPMXy55mR5ebnsZ5xxRkKPr843XV1dcqzreFFVVSW763hy4MAB2dX2MedObikp+s+K8/PzZXedL8aNGyf7vHnzZJ88ebJv6+3tlWPfeOMN2f/lX/5F9paWFtnj8bhvc813V3dd3+Tk5MheUlIi+9lnn+3b2tvb5VjXte/BgwdlV6+bmfsaQHXXY8P92cTVXfuu4jofuI5HiTy3md4/Ej1XuY6Frmvv6upq2V3HIyUSich+qp+n+UYUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAsFCFAAAAAAAAALBQhQAAAAAAAACwUIUAAAAAAAAAhEe6g04GUQiEdnfeecd31ZTUyPHjho1SvbRo0fLXlpaKvv06dN9Wzis3953331X9lAoJHtJSYnsc+fOlX38+PG+raOjQ47913/9V9kPHDggO04sz/Nkz8zMlL2qqkr2sWPHyq7mdCwWG/RYM7O8vDzZq6urZZ80aZJvS0nRfzawZs0a2d9++23ZGxoaZI/H47KfSK59BsOXa17MmDFDdtfxxjUvWlpafJvrXOM6lhUWFspeVFQke2pqquxIXq5ruPT0dNlHjhwp+5QpU2SfOnWq7GeccYZvO3TokBy7bds22V3nItf1qXptXHOmoKBAdtd1e21treyu92XMmDG+rb29XY51Xd90dXXJ3tfXJ3sixxvXYyNxJ/I6aiivDxPlOlZedNFFsruuIdra2nxbfX29HBuNRmU/1fGNKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABELfGxZmpm+t+Nxzz8mxxcXFst94442yu25Tf/rpp/u2OXPmyLG33nqr7K7b7zY3N8uekZEx6Md33Yb+jTfekN11O0xuFZ8Y1+3UXbcQ7ujokL2pqUl2td+bmU2bNs23jRgxQo5NS0uT3fWzu7q6ffInn3wixz7zzDOyb926VfZkvv0uhi/XrZNdt0N3He8jkYjs6lxVXl4ux7q6a05OmjRJ9uXLl8uujkeu5+Y8ObTCYX2J7jrPum5ZXlhYKLvrXKaus1zPfeaZZ8q+adMm2fv6+mQfNWrUoJ97/PjxstfW1sqe6DWGmnerV6+WY0tLS2Vft26d7K59ynVt7TqWAkPBNSeLiopk7+7uln3Lli2+ra2tTY4d7udZvhEFAAAAAACAQLAQBQAAAAAAgECwEAUAAAAAAIBAsBAFAAAAAACAQLAQBQAAAAAAgECwEAUAAAAAAIBAsBAFAAAAAACAQISHegOSXU9Pj+z/+3//b9l//etfy3799dfL/o1vfMO3FRQUyLHxeFz2ffv2yV5XVyf7hg0bZD906JBv+/jjj+XYpqYm2SORiOxIjGvfSU1NlT0ajcq+Z88e2TMyMmRX7391dbUcW1ZWJns4rA+b69evl33jxo2+7c0335RjW1paZHe9L8DJKiXF/8/FXPN9+/btsrvOVSNGjJC9pqbGt7mOB0VFRbL39vbKvm7dOtldx1L1unqeJ8e6JDoeWl9fn+zZ2dmyHzx4UHbXvBg/frzsSkVFhexqTpmZ3XjjjbK3t7fLXltb69vOPvtsOdY1p9WcMkv82vm9994b9GN/8sknsqvrbjP3tbPrGoNjAoaCa06OHj1a9vLyctldnznfeecd2eGPb0QBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgEOGh3oBTXSQSkX3Xrl2yP/bYY7L/n//zf3xbbm6uHBsKhWQPh/Xu4Xme7NFodNDje3p65FhXx9By7Rt9fX2yt7S0yL5mzRrZDx8+7Nt2794tx7o0NzfLvm/fPtnV87v2a9frCiSreDzu2zo7O+XYDRs2yP6HP/xB9lmzZsk+cuRI3/aVr3xFjnWd4999913ZXcebRI4JrrEcb4aW6/pRnefM3PPm/fffl33v3r2yV1dX+7bS0lI5dsqUKbLn5+fLXlhYKHtGRoZva2pqkmNd52HX9YdrTq9fv152tX2ua6fu7m7Z1XHWLLHrdmCoZGdnyz59+nTZjxw5ktDzZ2Zm+raUFP2dH1ePxWKD2qZkwTeiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEIjwUG8ANM/zZO/r6/NtLS0tx3tzvpRQKCR7amqqb4vFYnKs63XB0IrH47K73j/X+x+NRmXfvXu3bzt06JAc293dLXt7e3tC49Vrw34NfJ7reLB3717ZV65cKXtmZqbshw8f9m0NDQ1y7LZt22T/85//LPv27dtl7+zslN11rETyikQisrvee9e+29bWJvuGDRt8W0ZGhhz74osvyp6WlpZQLy8v922un1vNdzP3697b2yu7i3r8lBT9/YFEr72AoaI+M6rPi2ZmeXl5speWliY03nXMKC4u9m3Z2dlyrOt44/osnexzmm9EAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBAsRAEAAAAAACAQLEQBAAAAAAAgECxEAQAAAAAAIBDhod4AnLo8z5M9Go0GtCUIWigUkj0lRa+Bu/YN1+Or8YcPH07ouXt7e2V37feuDmAg15zs6OiQ/b333pN948aNsofD/pdKnZ2dcqxrvrvGx+Nx2TF8neh9o6ura9BjXfv1ibZr164hff4TJRaLDfUmAIOSyOeCtLQ0OTYnJ0f2rKws2V3XEHV1dYN+/vz8fDm2ublZ9lN9zvONKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAASChSgAAAAAAAAEgoUoAAAAAAAABIKFKAAAAAAAAAQiPNQbAODU43me7LFYTPbe3l7ZU1NTE3p8JSMjQ3bXz+bqAI4v15yLRCKyNzY2Hs/NAQAA/4nrPB2Px31bNBqVY12fGT799FPZR40aJXtPT4/s3d3dvi07O1uOzcrKSui5kx3fiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBYiAIAAAAAAEAgWIgCAAAAAABAIFiIAgAAAAAAQCBC3jHeazwUCp3obQFg7lucHivmLBAM5iyQXJizQHJhzmI4cu2vx2tenAjHsm18IwoAAAAAAACBYCEKAAAAAAAAgWAhCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgWAhCgAAAAAAAIFgIQoAAAAAAACBCA/1BgAAAAAAAOAznucN9SacUHwjCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgWAhCgAAAAAAAIFgIQoAAAAAAACBYCEKAAAAAAAAgWAhCgAAAAAAAIEIeZ7nDfVGAAAAAAAA4NTHN6IAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABAIFqIAAAAAAAAQCBaiAAAAAAAAEAgWogAAAAAAABCI/w9wbT2JKSaedwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1200x500 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(2, 5, figsize=(12, 5))\n",
    "\n",
    "axes = axes.flatten()\n",
    "\n",
    "for i in range(10):\n",
    "    axes[i].imshow(outputs[i].cpu().detach().numpy().squeeze(), cmap='gray')\n",
    "    axes[i].set_title(\"Reconstructed\")\n",
    "    axes[i].axis('off') \n",
    "    \n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "526d11d9-fd80-499b-9d01-e4fa897a6b0f",
   "metadata": {},
   "source": [
    "## 3. 합성곱 오토인코더(Convolutional Autoencoder)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "3fe55f18-19a8-4458-a776-f1f061e1eecc",
   "metadata": {},
   "outputs": [],
   "source": [
    "class Flatten(nn.Module):\n",
    "    def forward(self, x):\n",
    "        batch_size = x.shape[0]\n",
    "        return x.view(batch_size, -1)\n",
    "\n",
    "class Deflatten(nn.Module):\n",
    "    def __init__(self, k, height, width):\n",
    "        super(Deflatten, self).__init__()\n",
    "        self.k = k\n",
    "        self.height = height\n",
    "        self.width = width\n",
    "\n",
    "    def forward(self, x):\n",
    "        s = x.size()\n",
    "        # Reshape to (batch_size, channels, height, width)\n",
    "        return x.view(s[0], self.k, self.height, self.width)\n",
    "\n",
    "class AutoEncoder(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(AutoEncoder, self).__init__()\n",
    "        k = 16\n",
    "        self.encoder = nn.Sequential(\n",
    "            nn.Conv2d(1, k, 3, stride=2),  # Output: (batch_size, 16, 13, 13)\n",
    "            nn.ReLU(),\n",
    "            nn.Conv2d(k, 2*k, 3, stride=2),  # Output: (batch_size, 32, 6, 6)\n",
    "            nn.ReLU(),\n",
    "            nn.Conv2d(2*k, 4*k, 3, stride=1),  # Output: (batch_size, 64, 4, 4)\n",
    "            nn.ReLU(),\n",
    "            Flatten(),\n",
    "            nn.Linear(4*k*4*4, 10),  # Output: (batch_size, 10)\n",
    "            nn.ReLU()\n",
    "        )\n",
    "        self.decoder = nn.Sequential(\n",
    "            nn.Linear(10, 4*k*4*4),  # Output: (batch_size, 1024)\n",
    "            nn.ReLU(),\n",
    "            Deflatten(4*k, 4, 4),  # Reshape to (batch_size, 64, 4, 4)\n",
    "            nn.ConvTranspose2d(4*k, 2*k, 3, stride=1),  # Output: (batch_size, 32, 6, 6)\n",
    "            nn.ReLU(),\n",
    "            nn.ConvTranspose2d(2*k, k, 3, stride=2, output_padding=1),  # Output: (batch_size, 16, 13, 13)\n",
    "            nn.ReLU(),\n",
    "            nn.ConvTranspose2d(k, 1, 3, stride=2, padding=1, output_padding=1),  # Adjust this layer\n",
    "            nn.Sigmoid()  # Output: (batch_size, 1, 28, 28)\n",
    "        )\n",
    "\n",
    "    def forward(self, x):\n",
    "        encoded = self.encoder(x)\n",
    "        decoded = self.decoder(encoded)\n",
    "        return decoded\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6a0b0a93-6171-456e-9dc2-b1abf531a39a",
   "metadata": {},
   "source": [
    "- 위 인코더를 거친 feature map의 size는 (batch_size, channels, height, width)\n",
    "- 배치 사이즈가 현재 이미지의 개수이므로 벡터가 배치 사이즈 만큼 존재해야 함\n",
    "- 즉, 각 feature data를 일렬로 변환해야 함"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "b632fbaf-4825-4d28-9392-59146e6eb7d0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input shape: torch.Size([1, 1, 28, 28])\n",
      "Output shape: torch.Size([1, 1, 28, 28])\n"
     ]
    }
   ],
   "source": [
    "# Instantiate the model\n",
    "model = AutoEncoder().to(device)\n",
    "\n",
    "# Pass a dummy input through the model\n",
    "dummy_input = torch.randn(1, 1, 28, 28).to(device)\n",
    "output = model(dummy_input)\n",
    "\n",
    "print(f\"Input shape: {dummy_input.shape}\")\n",
    "print(f\"Output shape: {output.shape}\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "fead8c8e-8ff5-41f2-8a5c-8c06dbb63102",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = AutoEncoder().to(device)\n",
    "criterion = nn.MSELoss()\n",
    "optimizer = optim.Adam(model.parameters(), lr = 1e-3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "2f87bea6-3cf6-4a95-9c34-689ef34b6611",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 loss: 0.063\n",
      "11 loss: 0.025\n",
      "21 loss: 0.024\n",
      "31 loss: 0.023\n",
      "41 loss: 0.023\n",
      "51 loss: 0.022\n"
     ]
    }
   ],
   "source": [
    "for epoch in range(51):\n",
    "    running_loss = 0.0\n",
    "    for data in trainloader:\n",
    "        inputs = data[0].to(device)\n",
    "        optimizer.zero_grad()\n",
    "        outputs = model(inputs)\n",
    "        loss = criterion(inputs, outputs)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        running_loss += loss.item()\n",
    "\n",
    "    cost = running_loss / len(trainloader)\n",
    "    if epoch % 10 == 0:\n",
    "        print(f\"{epoch + 1} loss: {cost:.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "3bd60eaa-0e0e-466b-87cc-c465cbde03ca",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1200x500 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(2, 5, figsize=(12, 5))\n",
    "\n",
    "axes = axes.flatten()\n",
    "\n",
    "for i in range(10):\n",
    "    axes[i].imshow(outputs[i].cpu().detach().numpy().squeeze(), cmap='gray')\n",
    "    axes[i].set_title(\"Reconstructed\")\n",
    "    axes[i].axis('off') \n",
    "    \n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d2c7cc61-f768-441d-879b-d00cda4f7419",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.10.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}