resnet.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "kmALsaXdCj0m",
   "metadata": {
    "id": "kmALsaXdCj0m"
   },
   "source": [
    "# For Colab"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "DlsN4ztsCm-d",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 4908,
     "status": "ok",
     "timestamp": 1642416545049,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "DlsN4ztsCm-d",
    "outputId": "608cc4b2-e4ed-4f0b-9c26-eb93419862c6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mounted at /content/drive\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/drive', force_remount=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "tDXyGOonCqkh",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 270,
     "status": "ok",
     "timestamp": 1642416546773,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "tDXyGOonCqkh",
    "outputId": "bb83a479-587e-4aeb-c292-4a2157b88253"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "config.json\t\t\t    README\n",
      "config_recognition.json\t\t    resnet.ipynb\n",
      "config_recognition_mask-r-cnn.json  resnet_val_edge.png\n",
      "data\t\t\t\t    resnet_val_equalization.png\n",
      "detectors\t\t\t    resnet_val.png\n",
      "feature_extractors\t\t    resnet_val_sharpening.png\n",
      "metrics\t\t\t\t    run_evaluation.py\n",
      "preprocessing\t\t\t    run_recognition_evaluation.py\n",
      "Ranks.png\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "os.chdir(\"drive/My Drive/Slikovna biometrija/Assignment3\")\n",
    "!ls"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "kNZw_5XZCsM-",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "executionInfo": {
     "elapsed": 4,
     "status": "ok",
     "timestamp": 1642416547867,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "kNZw_5XZCsM-",
    "outputId": "95d8c7ed-b7dc-447d-9b2a-f8f20d9e1ae3"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.google.colaboratory.intrinsic+json": {
       "type": "string"
      },
      "text/plain": [
       "'/content/drive/My Drive/Slikovna biometrija/Assignment3'"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "os.getcwd()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d9e6ee8b-d37a-4a32-8db3-0e52f2c90de5",
   "metadata": {
    "id": "d9e6ee8b-d37a-4a32-8db3-0e52f2c90de5"
   },
   "source": [
    "# Ear based identification using ResNet50\n",
    "\n",
    "## Imports, constants and helper functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "fc11fa2c-1f19-43d1-9660-000de7802ddb",
   "metadata": {
    "executionInfo": {
     "elapsed": 1199,
     "status": "ok",
     "timestamp": 1642416551653,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "fc11fa2c-1f19-43d1-9660-000de7802ddb"
   },
   "outputs": [],
   "source": [
    "from torchvision import transforms\n",
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from data.AWEDataset import AWETestSet, AWETrainSet, AWEValSet\n",
    "from feature_extractors.your_super_extractor.CNN_detector import CNNEarDetector\n",
    "import cv2\n",
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "from preprocessing.preprocess import HistogramEqualization, EdgeEnhancement, ImageSharpening"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "005cfb05-c44e-4117-bd37-87f650923d7e",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 303,
     "status": "ok",
     "timestamp": 1642416553936,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "005cfb05-c44e-4117-bd37-87f650923d7e",
    "outputId": "381c45e3-2b2e-4a06-ce8f-f5212a18a67e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.cuda.is_available()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "8f7f7ea2-30c9-4401-b453-085b856f225b",
   "metadata": {
    "executionInfo": {
     "elapsed": 285,
     "status": "ok",
     "timestamp": 1642416556672,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "8f7f7ea2-30c9-4401-b453-085b856f225b"
   },
   "outputs": [],
   "source": [
    "INPUT_SIZE = 224\n",
    "BATCH_SIZE = 8\n",
    "# Number of classes in train set\n",
    "NUM_CLASSES = 100\n",
    "\n",
    "data_transforms = {\n",
    "    'train': transforms.Compose([\n",
    "        transforms.RandomResizedCrop(INPUT_SIZE),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "    'val': transforms.Compose([\n",
    "        transforms.Resize(INPUT_SIZE),\n",
    "        transforms.CenterCrop(INPUT_SIZE),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "}\n",
    "\n",
    "data_transforms_equalization = {\n",
    "    'train': transforms.Compose([\n",
    "        HistogramEqualization(),\n",
    "        transforms.RandomResizedCrop(INPUT_SIZE),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "    'val': transforms.Compose([\n",
    "        HistogramEqualization(),\n",
    "        transforms.Resize(INPUT_SIZE),\n",
    "        transforms.CenterCrop(INPUT_SIZE),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "}\n",
    "\n",
    "data_transforms_edge = {\n",
    "    'train': transforms.Compose([\n",
    "        EdgeEnhancement(),\n",
    "        transforms.RandomResizedCrop(INPUT_SIZE),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "    'val': transforms.Compose([\n",
    "        EdgeEnhancement(),\n",
    "        transforms.Resize(INPUT_SIZE),\n",
    "        transforms.CenterCrop(INPUT_SIZE),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "}\n",
    "\n",
    "data_transforms_sharpening = {\n",
    "    'train': transforms.Compose([\n",
    "        ImageSharpening(),\n",
    "        transforms.RandomResizedCrop(INPUT_SIZE),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "    'val': transforms.Compose([\n",
    "        ImageSharpening(),\n",
    "        transforms.Resize(INPUT_SIZE),\n",
    "        transforms.CenterCrop(INPUT_SIZE),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "029b66ae-fdf1-4087-b1c1-8291495d434d",
   "metadata": {
    "id": "029b66ae-fdf1-4087-b1c1-8291495d434d"
   },
   "source": [
    "## Perfectly detected ears\n",
    "\n",
    "### Data loading inspection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b6258993-75cf-429b-9c46-68f760d4ba86",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 4535,
     "status": "ok",
     "timestamp": 1642363935370,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "b6258993-75cf-429b-9c46-68f760d4ba86",
    "outputId": "9ef49b9f-a83d-4652-df1f-491da6c96228"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(<PIL.Image.Image image mode=RGB size=53x78 at 0x7F5E1D0759D0>, tensor(99))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data = AWETrainSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), None)\n",
    "train_data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e9aae4b3-d037-44d5-aaf3-bdda12878596",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 285
    },
    "executionInfo": {
     "elapsed": 1250,
     "status": "ok",
     "timestamp": 1642363943192,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "e9aae4b3-d037-44d5-aaf3-bdda12878596",
    "outputId": "bf6243f3-63ee-4968-8d86-4037c1dea495"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(99)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "image, label = train_data[0]\n",
    "plt.imshow(image)\n",
    "print(label)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d9bba21-57f1-4e41-ac6d-00929f5167fd",
   "metadata": {
    "id": "4d9bba21-57f1-4e41-ac6d-00929f5167fd"
   },
   "source": [
    "### Without preprocessing\n",
    "\n",
    "#### Load train and val data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "aaaae5ce-873b-49f8-b81d-3c7cb1a70716",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 291,
     "status": "ok",
     "timestamp": 1642411249589,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "aaaae5ce-873b-49f8-b81d-3c7cb1a70716",
    "outputId": "69560647-2992-4626-9b6b-95d99d828aed"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(tensor([[[ 0.3994,  0.3994,  0.3994,  ...,  1.9407,  1.9407,  1.9407],\n",
       "          [ 0.3994,  0.3994,  0.3994,  ...,  1.9407,  1.9407,  1.9407],\n",
       "          [ 0.3994,  0.3994,  0.3994,  ...,  1.9407,  1.9407,  1.9407],\n",
       "          ...,\n",
       "          [-0.0801, -0.0801, -0.0801,  ..., -0.2342, -0.2342, -0.2342],\n",
       "          [-0.0801, -0.0801, -0.0801,  ..., -0.2342, -0.2342, -0.2342],\n",
       "          [-0.0801, -0.0801, -0.0801,  ..., -0.2342, -0.2342, -0.2342]],\n",
       " \n",
       "         [[-1.0553, -1.0553, -1.0553,  ...,  0.5028,  0.5028,  0.5028],\n",
       "          [-1.0553, -1.0553, -1.0553,  ...,  0.5028,  0.5028,  0.5028],\n",
       "          [-1.0553, -1.0553, -1.0553,  ...,  0.5028,  0.5028,  0.5028],\n",
       "          ...,\n",
       "          [-1.2654, -1.2654, -1.2654,  ..., -1.3880, -1.3880, -1.3880],\n",
       "          [-1.2654, -1.2654, -1.2654,  ..., -1.3880, -1.3880, -1.3880],\n",
       "          [-1.2654, -1.2654, -1.2654,  ..., -1.3880, -1.3880, -1.3880]],\n",
       " \n",
       "         [[-1.1421, -1.1421, -1.1421,  ..., -0.3753, -0.3753, -0.3753],\n",
       "          [-1.1421, -1.1421, -1.1421,  ..., -0.3753, -0.3753, -0.3753],\n",
       "          [-1.1421, -1.1421, -1.1421,  ..., -0.3753, -0.3753, -0.3753],\n",
       "          ...,\n",
       "          [-1.5953, -1.5953, -1.5953,  ..., -1.8044, -1.8044, -1.8044],\n",
       "          [-1.5953, -1.5953, -1.5953,  ..., -1.8044, -1.8044, -1.8044],\n",
       "          [-1.5953, -1.5953, -1.5953,  ..., -1.8044, -1.8044, -1.8044]]]),\n",
       " tensor(99))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data = AWETrainSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), data_transforms[\"train\"])\n",
    "train_data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "86fbcc8a-2e03-46cc-aeea-d106586c1c4f",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 258,
     "status": "ok",
     "timestamp": 1642411253023,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "86fbcc8a-2e03-46cc-aeea-d106586c1c4f",
    "outputId": "d004f464-55d6-4545-f7f4-7d75269b933c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(tensor([[[-0.5938, -0.5938, -0.5253,  ...,  0.3309,  0.3652,  0.3652],\n",
       "          [-0.5938, -0.5938, -0.5253,  ...,  0.2967,  0.3309,  0.3309],\n",
       "          [-0.5938, -0.5938, -0.5253,  ...,  0.2624,  0.2967,  0.2967],\n",
       "          ...,\n",
       "          [-1.9809, -1.9809, -1.9638,  ...,  0.9474,  0.9646,  0.9646],\n",
       "          [-1.9809, -1.9809, -1.9638,  ...,  0.9474,  0.9646,  0.9646],\n",
       "          [-1.9809, -1.9809, -1.9638,  ...,  0.9474,  0.9646,  0.9646]],\n",
       " \n",
       "         [[-0.9678, -0.9678, -0.9678,  ..., -0.2675, -0.2150, -0.2150],\n",
       "          [-0.9678, -0.9678, -0.9678,  ..., -0.3025, -0.2500, -0.2500],\n",
       "          [-0.9678, -0.9678, -0.9678,  ..., -0.3375, -0.2850, -0.2850],\n",
       "          ...,\n",
       "          [-1.8957, -1.8957, -1.8782,  ...,  0.3803,  0.3803,  0.3803],\n",
       "          [-1.8957, -1.8957, -1.8782,  ...,  0.3803,  0.3803,  0.3803],\n",
       "          [-1.8957, -1.8957, -1.8782,  ...,  0.3803,  0.3803,  0.3803]],\n",
       " \n",
       "         [[-0.9504, -0.9504, -0.9678,  ..., -0.3753, -0.3230, -0.3230],\n",
       "          [-0.9330, -0.9330, -0.9504,  ..., -0.4101, -0.3578, -0.3578],\n",
       "          [-0.9330, -0.9330, -0.9504,  ..., -0.4450, -0.4101, -0.4101],\n",
       "          ...,\n",
       "          [-1.6650, -1.6650, -1.6476,  ...,  0.1651,  0.1651,  0.1651],\n",
       "          [-1.6650, -1.6650, -1.6476,  ...,  0.1651,  0.1651,  0.1651],\n",
       "          [-1.6650, -1.6650, -1.6476,  ...,  0.1651,  0.1651,  0.1651]]]),\n",
       " tensor(10))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val_data = AWEValSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), data_transforms[\"val\"])\n",
    "val_data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "dfbc0f4a-d15a-4b03-bab7-6433094d809c",
   "metadata": {
    "executionInfo": {
     "elapsed": 273,
     "status": "ok",
     "timestamp": 1642411256702,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "dfbc0f4a-d15a-4b03-bab7-6433094d809c"
   },
   "outputs": [],
   "source": [
    "# Create training and validation dataloaders\n",
    "dataloaders_dict = {'train': torch.utils.data.DataLoader(train_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=2),\n",
    "                   'val': torch.utils.data.DataLoader(val_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=2)}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "989054b0-9a5d-417e-9c55-b3c899641585",
   "metadata": {
    "id": "989054b0-9a5d-417e-9c55-b3c899641585"
   },
   "source": [
    "#### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ae6db63e-c3a2-4772-8d77-08ea3c9cf766",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 1491725,
     "status": "ok",
     "timestamp": 1642412756219,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "ae6db63e-c3a2-4772-8d77-08ea3c9cf766",
    "outputId": "3734faf5-fa05-4a6d-8bff-24937ac84c9b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training model on GPUs\n",
      "Epoch 0/199\n",
      "----------\n",
      "train Loss: 4.6647 Acc: 0.0072\n",
      "val Loss: 4.6450 Acc: 0.0060\n",
      "\n",
      "Epoch 1/199\n",
      "----------\n",
      "train Loss: 4.6494 Acc: 0.0024\n",
      "val Loss: 4.6362 Acc: 0.0090\n",
      "\n",
      "Epoch 2/199\n",
      "----------\n",
      "train Loss: 4.6173 Acc: 0.0096\n",
      "val Loss: 4.6160 Acc: 0.0060\n",
      "\n",
      "Epoch 3/199\n",
      "----------\n",
      "train Loss: 4.5997 Acc: 0.0048\n",
      "val Loss: 4.6129 Acc: 0.0090\n",
      "\n",
      "Epoch 4/199\n",
      "----------\n",
      "train Loss: 4.5958 Acc: 0.0096\n",
      "val Loss: 4.6096 Acc: 0.0209\n",
      "\n",
      "Epoch 5/199\n",
      "----------\n",
      "train Loss: 4.5895 Acc: 0.0120\n",
      "val Loss: 4.6039 Acc: 0.0209\n",
      "\n",
      "Epoch 6/199\n",
      "----------\n",
      "train Loss: 4.5689 Acc: 0.0169\n",
      "val Loss: 4.6081 Acc: 0.0119\n",
      "\n",
      "Epoch 7/199\n",
      "----------\n",
      "train Loss: 4.5492 Acc: 0.0241\n",
      "val Loss: 4.5909 Acc: 0.0209\n",
      "\n",
      "Epoch 8/199\n",
      "----------\n",
      "train Loss: 4.5369 Acc: 0.0361\n",
      "val Loss: 4.5953 Acc: 0.0209\n",
      "\n",
      "Epoch 9/199\n",
      "----------\n",
      "train Loss: 4.5305 Acc: 0.0313\n",
      "val Loss: 4.5795 Acc: 0.0239\n",
      "\n",
      "Epoch 10/199\n",
      "----------\n",
      "train Loss: 4.5168 Acc: 0.0313\n",
      "val Loss: 4.5754 Acc: 0.0269\n",
      "\n",
      "Epoch 11/199\n",
      "----------\n",
      "train Loss: 4.5032 Acc: 0.0361\n",
      "val Loss: 4.5736 Acc: 0.0269\n",
      "\n",
      "Epoch 12/199\n",
      "----------\n",
      "train Loss: 4.4900 Acc: 0.0482\n",
      "val Loss: 4.5686 Acc: 0.0239\n",
      "\n",
      "Epoch 13/199\n",
      "----------\n",
      "train Loss: 4.4681 Acc: 0.0361\n",
      "val Loss: 4.5692 Acc: 0.0119\n",
      "\n",
      "Epoch 14/199\n",
      "----------\n",
      "train Loss: 4.4572 Acc: 0.0602\n",
      "val Loss: 4.5665 Acc: 0.0149\n",
      "\n",
      "Epoch 15/199\n",
      "----------\n",
      "train Loss: 4.4436 Acc: 0.0675\n",
      "val Loss: 4.5527 Acc: 0.0269\n",
      "\n",
      "Epoch 16/199\n",
      "----------\n",
      "train Loss: 4.4338 Acc: 0.0699\n",
      "val Loss: 4.5477 Acc: 0.0269\n",
      "\n",
      "Epoch 17/199\n",
      "----------\n",
      "train Loss: 4.4297 Acc: 0.0795\n",
      "val Loss: 4.5364 Acc: 0.0358\n",
      "\n",
      "Epoch 18/199\n",
      "----------\n",
      "train Loss: 4.4098 Acc: 0.0627\n",
      "val Loss: 4.5312 Acc: 0.0388\n",
      "\n",
      "Epoch 19/199\n",
      "----------\n",
      "train Loss: 4.3962 Acc: 0.0964\n",
      "val Loss: 4.5369 Acc: 0.0269\n",
      "\n",
      "Epoch 20/199\n",
      "----------\n",
      "train Loss: 4.3809 Acc: 0.0940\n",
      "val Loss: 4.5221 Acc: 0.0418\n",
      "\n",
      "Epoch 21/199\n",
      "----------\n",
      "train Loss: 4.3562 Acc: 0.0988\n",
      "val Loss: 4.5264 Acc: 0.0299\n",
      "\n",
      "Epoch 22/199\n",
      "----------\n",
      "train Loss: 4.3186 Acc: 0.1253\n",
      "val Loss: 4.5194 Acc: 0.0388\n",
      "\n",
      "Epoch 23/199\n",
      "----------\n",
      "train Loss: 4.3146 Acc: 0.1325\n",
      "val Loss: 4.5147 Acc: 0.0388\n",
      "\n",
      "Epoch 24/199\n",
      "----------\n",
      "train Loss: 4.3054 Acc: 0.1205\n",
      "val Loss: 4.4988 Acc: 0.0537\n",
      "\n",
      "Epoch 25/199\n",
      "----------\n",
      "train Loss: 4.2842 Acc: 0.1229\n",
      "val Loss: 4.5004 Acc: 0.0388\n",
      "\n",
      "Epoch 26/199\n",
      "----------\n",
      "train Loss: 4.2871 Acc: 0.0940\n",
      "val Loss: 4.4824 Acc: 0.0388\n",
      "\n",
      "Epoch 27/199\n",
      "----------\n",
      "train Loss: 4.2413 Acc: 0.1205\n",
      "val Loss: 4.4844 Acc: 0.0507\n",
      "\n",
      "Epoch 28/199\n",
      "----------\n",
      "train Loss: 4.2409 Acc: 0.1446\n",
      "val Loss: 4.4803 Acc: 0.0478\n",
      "\n",
      "Epoch 29/199\n",
      "----------\n",
      "train Loss: 4.2287 Acc: 0.1639\n",
      "val Loss: 4.4726 Acc: 0.0597\n",
      "\n",
      "Epoch 30/199\n",
      "----------\n",
      "train Loss: 4.1951 Acc: 0.1735\n",
      "val Loss: 4.4723 Acc: 0.0507\n",
      "\n",
      "Epoch 31/199\n",
      "----------\n",
      "train Loss: 4.1663 Acc: 0.1735\n",
      "val Loss: 4.4660 Acc: 0.0478\n",
      "\n",
      "Epoch 32/199\n",
      "----------\n",
      "train Loss: 4.1464 Acc: 0.1639\n",
      "val Loss: 4.4549 Acc: 0.0537\n",
      "\n",
      "Epoch 33/199\n",
      "----------\n",
      "train Loss: 4.1347 Acc: 0.1831\n",
      "val Loss: 4.4531 Acc: 0.0507\n",
      "\n",
      "Epoch 34/199\n",
      "----------\n",
      "train Loss: 4.1152 Acc: 0.1952\n",
      "val Loss: 4.4320 Acc: 0.0537\n",
      "\n",
      "Epoch 35/199\n",
      "----------\n",
      "train Loss: 4.1106 Acc: 0.1711\n",
      "val Loss: 4.4331 Acc: 0.0657\n",
      "\n",
      "Epoch 36/199\n",
      "----------\n",
      "train Loss: 4.0945 Acc: 0.1639\n",
      "val Loss: 4.4248 Acc: 0.0597\n",
      "\n",
      "Epoch 37/199\n",
      "----------\n",
      "train Loss: 4.0747 Acc: 0.1904\n",
      "val Loss: 4.4175 Acc: 0.0627\n",
      "\n",
      "Epoch 38/199\n",
      "----------\n",
      "train Loss: 4.0432 Acc: 0.1855\n",
      "val Loss: 4.4080 Acc: 0.0597\n",
      "\n",
      "Epoch 39/199\n",
      "----------\n",
      "train Loss: 4.0119 Acc: 0.1928\n",
      "val Loss: 4.3993 Acc: 0.0597\n",
      "\n",
      "Epoch 40/199\n",
      "----------\n",
      "train Loss: 3.9677 Acc: 0.2386\n",
      "val Loss: 4.3976 Acc: 0.0567\n",
      "\n",
      "Epoch 41/199\n",
      "----------\n",
      "train Loss: 3.9695 Acc: 0.2289\n",
      "val Loss: 4.3843 Acc: 0.0687\n",
      "\n",
      "Epoch 42/199\n",
      "----------\n",
      "train Loss: 3.9337 Acc: 0.2096\n",
      "val Loss: 4.3964 Acc: 0.0687\n",
      "\n",
      "Epoch 43/199\n",
      "----------\n",
      "train Loss: 3.9318 Acc: 0.2072\n",
      "val Loss: 4.3763 Acc: 0.0746\n",
      "\n",
      "Epoch 44/199\n",
      "----------\n",
      "train Loss: 3.9041 Acc: 0.2386\n",
      "val Loss: 4.3868 Acc: 0.0597\n",
      "\n",
      "Epoch 45/199\n",
      "----------\n",
      "train Loss: 3.8724 Acc: 0.2289\n",
      "val Loss: 4.3652 Acc: 0.0627\n",
      "\n",
      "Epoch 46/199\n",
      "----------\n",
      "train Loss: 3.8739 Acc: 0.2072\n",
      "val Loss: 4.3627 Acc: 0.0716\n",
      "\n",
      "Epoch 47/199\n",
      "----------\n",
      "train Loss: 3.8456 Acc: 0.2602\n",
      "val Loss: 4.3834 Acc: 0.0597\n",
      "\n",
      "Epoch 48/199\n",
      "----------\n",
      "train Loss: 3.8225 Acc: 0.2361\n",
      "val Loss: 4.3648 Acc: 0.0627\n",
      "\n",
      "Epoch 49/199\n",
      "----------\n",
      "train Loss: 3.8068 Acc: 0.2217\n",
      "val Loss: 4.3630 Acc: 0.0746\n",
      "\n",
      "Epoch 50/199\n",
      "----------\n",
      "train Loss: 3.7936 Acc: 0.2434\n",
      "val Loss: 4.3548 Acc: 0.0627\n",
      "\n",
      "Epoch 51/199\n",
      "----------\n",
      "train Loss: 3.7526 Acc: 0.2651\n",
      "val Loss: 4.3543 Acc: 0.0687\n",
      "\n",
      "Epoch 52/199\n",
      "----------\n",
      "train Loss: 3.7610 Acc: 0.2434\n",
      "val Loss: 4.3516 Acc: 0.0627\n",
      "\n",
      "Epoch 53/199\n",
      "----------\n",
      "train Loss: 3.7287 Acc: 0.2554\n",
      "val Loss: 4.3486 Acc: 0.0657\n",
      "\n",
      "Epoch 54/199\n",
      "----------\n",
      "train Loss: 3.6989 Acc: 0.2651\n",
      "val Loss: 4.3419 Acc: 0.0687\n",
      "\n",
      "Epoch 55/199\n",
      "----------\n",
      "train Loss: 3.6899 Acc: 0.2578\n",
      "val Loss: 4.3388 Acc: 0.0716\n",
      "\n",
      "Epoch 56/199\n",
      "----------\n",
      "train Loss: 3.6640 Acc: 0.2482\n",
      "val Loss: 4.3367 Acc: 0.0657\n",
      "\n",
      "Epoch 57/199\n",
      "----------\n",
      "train Loss: 3.6023 Acc: 0.2940\n",
      "val Loss: 4.3411 Acc: 0.0687\n",
      "\n",
      "Epoch 58/199\n",
      "----------\n",
      "train Loss: 3.6257 Acc: 0.2410\n",
      "val Loss: 4.3265 Acc: 0.0687\n",
      "\n",
      "Epoch 59/199\n",
      "----------\n",
      "train Loss: 3.6064 Acc: 0.2578\n",
      "val Loss: 4.3439 Acc: 0.0657\n",
      "\n",
      "Epoch 60/199\n",
      "----------\n",
      "train Loss: 3.6040 Acc: 0.2747\n",
      "val Loss: 4.3273 Acc: 0.0776\n",
      "\n",
      "Epoch 61/199\n",
      "----------\n",
      "train Loss: 3.5954 Acc: 0.2578\n",
      "val Loss: 4.3397 Acc: 0.0746\n",
      "\n",
      "Epoch 62/199\n",
      "----------\n",
      "train Loss: 3.5598 Acc: 0.2771\n",
      "val Loss: 4.3179 Acc: 0.0776\n",
      "\n",
      "Epoch 63/199\n",
      "----------\n",
      "train Loss: 3.5470 Acc: 0.2651\n",
      "val Loss: 4.3137 Acc: 0.0806\n",
      "\n",
      "Epoch 64/199\n",
      "----------\n",
      "train Loss: 3.4973 Acc: 0.3157\n",
      "val Loss: 4.3075 Acc: 0.0836\n",
      "\n",
      "Epoch 65/199\n",
      "----------\n",
      "train Loss: 3.4855 Acc: 0.3012\n",
      "val Loss: 4.3085 Acc: 0.0866\n",
      "\n",
      "Epoch 66/199\n",
      "----------\n",
      "train Loss: 3.4735 Acc: 0.3012\n",
      "val Loss: 4.3154 Acc: 0.0896\n",
      "\n",
      "Epoch 67/199\n",
      "----------\n",
      "train Loss: 3.4578 Acc: 0.2940\n",
      "val Loss: 4.3129 Acc: 0.0836\n",
      "\n",
      "Epoch 68/199\n",
      "----------\n",
      "train Loss: 3.4228 Acc: 0.3036\n",
      "val Loss: 4.3216 Acc: 0.0836\n",
      "\n",
      "Epoch 69/199\n",
      "----------\n",
      "train Loss: 3.3983 Acc: 0.2892\n",
      "val Loss: 4.3290 Acc: 0.0866\n",
      "\n",
      "Epoch 70/199\n",
      "----------\n",
      "train Loss: 3.4130 Acc: 0.2819\n",
      "val Loss: 4.3062 Acc: 0.0866\n",
      "\n",
      "Epoch 71/199\n",
      "----------\n",
      "train Loss: 3.3735 Acc: 0.3108\n",
      "val Loss: 4.3258 Acc: 0.0925\n",
      "\n",
      "Epoch 72/199\n",
      "----------\n",
      "train Loss: 3.3166 Acc: 0.3349\n",
      "val Loss: 4.3242 Acc: 0.0925\n",
      "\n",
      "Epoch 73/199\n",
      "----------\n",
      "train Loss: 3.3334 Acc: 0.3012\n",
      "val Loss: 4.3376 Acc: 0.0955\n",
      "\n",
      "Epoch 74/199\n",
      "----------\n",
      "train Loss: 3.3234 Acc: 0.3446\n",
      "val Loss: 4.3302 Acc: 0.0925\n",
      "\n",
      "Epoch 75/199\n",
      "----------\n",
      "train Loss: 3.2994 Acc: 0.3422\n",
      "val Loss: 4.3060 Acc: 0.0925\n",
      "\n",
      "Epoch 76/199\n",
      "----------\n",
      "train Loss: 3.2686 Acc: 0.3349\n",
      "val Loss: 4.3187 Acc: 0.0836\n",
      "\n",
      "Epoch 77/199\n",
      "----------\n",
      "train Loss: 3.2688 Acc: 0.3205\n",
      "val Loss: 4.3310 Acc: 0.0896\n",
      "\n",
      "Epoch 78/199\n",
      "----------\n",
      "train Loss: 3.2410 Acc: 0.3181\n",
      "val Loss: 4.3262 Acc: 0.0955\n",
      "\n",
      "Epoch 79/199\n",
      "----------\n",
      "train Loss: 3.2167 Acc: 0.3398\n",
      "val Loss: 4.3123 Acc: 0.0955\n",
      "\n",
      "Epoch 80/199\n",
      "----------\n",
      "train Loss: 3.2064 Acc: 0.3398\n",
      "val Loss: 4.3276 Acc: 0.0896\n",
      "\n",
      "Epoch 81/199\n",
      "----------\n",
      "train Loss: 3.2803 Acc: 0.3398\n",
      "val Loss: 4.3119 Acc: 0.0955\n",
      "\n",
      "Epoch 82/199\n",
      "----------\n",
      "train Loss: 3.1457 Acc: 0.3229\n",
      "val Loss: 4.3254 Acc: 0.1104\n",
      "\n",
      "Epoch 83/199\n",
      "----------\n",
      "train Loss: 3.1800 Acc: 0.3663\n",
      "val Loss: 4.3250 Acc: 0.0925\n",
      "\n",
      "Epoch 84/199\n",
      "----------\n",
      "train Loss: 3.1193 Acc: 0.3590\n",
      "val Loss: 4.3278 Acc: 0.1015\n",
      "\n",
      "Epoch 85/199\n",
      "----------\n",
      "train Loss: 3.1109 Acc: 0.3590\n",
      "val Loss: 4.3203 Acc: 0.1015\n",
      "\n",
      "Epoch 86/199\n",
      "----------\n",
      "train Loss: 3.0853 Acc: 0.3976\n",
      "val Loss: 4.3219 Acc: 0.0955\n",
      "\n",
      "Epoch 87/199\n",
      "----------\n",
      "train Loss: 3.1347 Acc: 0.3566\n",
      "val Loss: 4.3130 Acc: 0.1045\n",
      "\n",
      "Epoch 88/199\n",
      "----------\n",
      "train Loss: 3.0840 Acc: 0.3711\n",
      "val Loss: 4.3280 Acc: 0.1015\n",
      "\n",
      "Epoch 89/199\n",
      "----------\n",
      "train Loss: 3.0322 Acc: 0.3759\n",
      "val Loss: 4.3337 Acc: 0.1134\n",
      "\n",
      "Epoch 90/199\n",
      "----------\n",
      "train Loss: 2.9944 Acc: 0.4000\n",
      "val Loss: 4.3180 Acc: 0.1015\n",
      "\n",
      "Epoch 91/199\n",
      "----------\n",
      "train Loss: 2.9959 Acc: 0.3880\n",
      "val Loss: 4.3363 Acc: 0.1104\n",
      "\n",
      "Epoch 92/199\n",
      "----------\n",
      "train Loss: 3.0289 Acc: 0.3639\n",
      "val Loss: 4.3293 Acc: 0.1045\n",
      "\n",
      "Epoch 93/199\n",
      "----------\n",
      "train Loss: 2.9731 Acc: 0.3976\n",
      "val Loss: 4.3456 Acc: 0.1075\n",
      "\n",
      "Epoch 94/199\n",
      "----------\n",
      "train Loss: 2.9557 Acc: 0.4120\n",
      "val Loss: 4.3386 Acc: 0.1045\n",
      "\n",
      "Epoch 95/199\n",
      "----------\n",
      "train Loss: 2.9692 Acc: 0.3928\n",
      "val Loss: 4.3304 Acc: 0.1164\n",
      "\n",
      "Epoch 96/199\n",
      "----------\n",
      "train Loss: 2.9532 Acc: 0.3807\n",
      "val Loss: 4.3385 Acc: 0.1075\n",
      "\n",
      "Epoch 97/199\n",
      "----------\n",
      "train Loss: 2.9186 Acc: 0.4241\n",
      "val Loss: 4.3147 Acc: 0.1104\n",
      "\n",
      "Epoch 98/199\n",
      "----------\n",
      "train Loss: 2.9338 Acc: 0.4048\n",
      "val Loss: 4.3197 Acc: 0.1075\n",
      "\n",
      "Epoch 99/199\n",
      "----------\n",
      "train Loss: 2.8734 Acc: 0.4193\n",
      "val Loss: 4.3524 Acc: 0.1015\n",
      "\n",
      "Epoch 100/199\n",
      "----------\n",
      "train Loss: 2.8496 Acc: 0.4217\n",
      "val Loss: 4.3383 Acc: 0.1164\n",
      "\n",
      "Epoch 101/199\n",
      "----------\n",
      "train Loss: 2.8278 Acc: 0.4530\n",
      "val Loss: 4.3446 Acc: 0.1194\n",
      "\n",
      "Epoch 102/199\n",
      "----------\n",
      "train Loss: 2.8722 Acc: 0.4627\n",
      "val Loss: 4.3367 Acc: 0.1134\n",
      "\n",
      "Epoch 103/199\n",
      "----------\n",
      "train Loss: 2.7653 Acc: 0.4627\n",
      "val Loss: 4.3378 Acc: 0.1194\n",
      "\n",
      "Epoch 104/199\n",
      "----------\n",
      "train Loss: 2.7659 Acc: 0.4530\n",
      "val Loss: 4.3229 Acc: 0.1254\n",
      "\n",
      "Epoch 105/199\n",
      "----------\n",
      "train Loss: 2.7564 Acc: 0.4410\n",
      "val Loss: 4.3168 Acc: 0.1075\n",
      "\n",
      "Epoch 106/199\n",
      "----------\n",
      "train Loss: 2.7064 Acc: 0.4482\n",
      "val Loss: 4.3575 Acc: 0.1194\n",
      "\n",
      "Epoch 107/199\n",
      "----------\n",
      "train Loss: 2.6866 Acc: 0.4988\n",
      "val Loss: 4.3277 Acc: 0.1134\n",
      "\n",
      "Epoch 108/199\n",
      "----------\n",
      "train Loss: 2.7261 Acc: 0.4771\n",
      "val Loss: 4.3267 Acc: 0.1164\n",
      "\n",
      "Epoch 109/199\n",
      "----------\n",
      "train Loss: 2.6889 Acc: 0.4747\n",
      "val Loss: 4.3525 Acc: 0.1045\n",
      "\n",
      "Epoch 110/199\n",
      "----------\n",
      "train Loss: 2.6417 Acc: 0.4627\n",
      "val Loss: 4.3355 Acc: 0.0955\n",
      "\n",
      "Epoch 111/199\n",
      "----------\n",
      "train Loss: 2.6499 Acc: 0.4699\n",
      "val Loss: 4.3264 Acc: 0.1164\n",
      "\n",
      "Epoch 112/199\n",
      "----------\n",
      "train Loss: 2.6060 Acc: 0.4843\n",
      "val Loss: 4.3648 Acc: 0.1224\n",
      "\n",
      "Epoch 113/199\n",
      "----------\n",
      "train Loss: 2.6103 Acc: 0.4940\n",
      "val Loss: 4.3495 Acc: 0.1194\n",
      "\n",
      "Epoch 114/199\n",
      "----------\n",
      "train Loss: 2.5566 Acc: 0.5084\n",
      "val Loss: 4.3295 Acc: 0.1164\n",
      "\n",
      "Epoch 115/199\n",
      "----------\n",
      "train Loss: 2.5466 Acc: 0.4940\n",
      "val Loss: 4.3695 Acc: 0.1194\n",
      "\n",
      "Epoch 116/199\n",
      "----------\n",
      "train Loss: 2.5893 Acc: 0.5157\n",
      "val Loss: 4.3739 Acc: 0.1104\n",
      "\n",
      "Epoch 117/199\n",
      "----------\n",
      "train Loss: 2.6111 Acc: 0.4867\n",
      "val Loss: 4.3556 Acc: 0.1194\n",
      "\n",
      "Epoch 118/199\n",
      "----------\n",
      "train Loss: 2.5359 Acc: 0.5084\n",
      "val Loss: 4.3606 Acc: 0.1164\n",
      "\n",
      "Epoch 119/199\n",
      "----------\n",
      "train Loss: 2.4711 Acc: 0.5349\n",
      "val Loss: 4.3394 Acc: 0.1104\n",
      "\n",
      "Epoch 120/199\n",
      "----------\n",
      "train Loss: 2.4740 Acc: 0.4988\n",
      "val Loss: 4.3667 Acc: 0.1104\n",
      "\n",
      "Epoch 121/199\n",
      "----------\n",
      "train Loss: 2.5262 Acc: 0.4892\n",
      "val Loss: 4.3698 Acc: 0.1104\n",
      "\n",
      "Epoch 122/199\n",
      "----------\n",
      "train Loss: 2.4702 Acc: 0.5157\n",
      "val Loss: 4.3598 Acc: 0.1194\n",
      "\n",
      "Epoch 123/199\n",
      "----------\n",
      "train Loss: 2.3945 Acc: 0.5494\n",
      "val Loss: 4.3768 Acc: 0.1194\n",
      "\n",
      "Epoch 124/199\n",
      "----------\n",
      "train Loss: 2.5297 Acc: 0.5084\n",
      "val Loss: 4.3797 Acc: 0.1224\n",
      "\n",
      "Epoch 125/199\n",
      "----------\n",
      "train Loss: 2.3727 Acc: 0.5614\n",
      "val Loss: 4.3637 Acc: 0.1164\n",
      "\n",
      "Epoch 126/199\n",
      "----------\n",
      "train Loss: 2.3941 Acc: 0.5301\n",
      "val Loss: 4.3395 Acc: 0.1015\n",
      "\n",
      "Epoch 127/199\n",
      "----------\n",
      "train Loss: 2.4421 Acc: 0.5349\n",
      "val Loss: 4.3750 Acc: 0.1284\n",
      "\n",
      "Epoch 128/199\n",
      "----------\n",
      "train Loss: 2.2846 Acc: 0.5711\n",
      "val Loss: 4.3465 Acc: 0.1134\n",
      "\n",
      "Epoch 129/199\n",
      "----------\n",
      "train Loss: 2.3133 Acc: 0.5735\n",
      "val Loss: 4.3667 Acc: 0.1164\n",
      "\n",
      "Epoch 130/199\n",
      "----------\n",
      "train Loss: 2.4390 Acc: 0.4867\n",
      "val Loss: 4.3656 Acc: 0.1134\n",
      "\n",
      "Epoch 131/199\n",
      "----------\n",
      "train Loss: 2.2938 Acc: 0.5880\n",
      "val Loss: 4.3649 Acc: 0.1164\n",
      "\n",
      "Epoch 132/199\n",
      "----------\n",
      "train Loss: 2.3027 Acc: 0.5735\n",
      "val Loss: 4.4009 Acc: 0.1045\n",
      "\n",
      "Epoch 133/199\n",
      "----------\n",
      "train Loss: 2.2586 Acc: 0.5422\n",
      "val Loss: 4.3439 Acc: 0.1224\n",
      "\n",
      "Epoch 134/199\n",
      "----------\n",
      "train Loss: 2.2886 Acc: 0.5566\n",
      "val Loss: 4.3789 Acc: 0.1224\n",
      "\n",
      "Epoch 135/199\n",
      "----------\n",
      "train Loss: 2.2059 Acc: 0.5880\n",
      "val Loss: 4.3721 Acc: 0.1164\n",
      "\n",
      "Epoch 136/199\n",
      "----------\n",
      "train Loss: 2.1892 Acc: 0.5952\n",
      "val Loss: 4.4205 Acc: 0.1075\n",
      "\n",
      "Epoch 137/199\n",
      "----------\n",
      "train Loss: 2.1461 Acc: 0.5976\n",
      "val Loss: 4.3690 Acc: 0.1104\n",
      "\n",
      "Epoch 138/199\n",
      "----------\n",
      "train Loss: 2.2239 Acc: 0.5566\n",
      "val Loss: 4.4069 Acc: 0.1254\n",
      "\n",
      "Epoch 139/199\n",
      "----------\n",
      "train Loss: 2.1698 Acc: 0.5711\n",
      "val Loss: 4.3452 Acc: 0.1254\n",
      "\n",
      "Epoch 140/199\n",
      "----------\n",
      "train Loss: 2.1915 Acc: 0.5735\n",
      "val Loss: 4.3799 Acc: 0.1164\n",
      "\n",
      "Epoch 141/199\n",
      "----------\n",
      "train Loss: 2.1328 Acc: 0.5952\n",
      "val Loss: 4.3837 Acc: 0.1313\n",
      "\n",
      "Epoch 142/199\n",
      "----------\n",
      "train Loss: 2.0762 Acc: 0.6145\n",
      "val Loss: 4.3629 Acc: 0.1224\n",
      "\n",
      "Epoch 143/199\n",
      "----------\n",
      "train Loss: 2.1897 Acc: 0.5566\n",
      "val Loss: 4.3874 Acc: 0.1224\n",
      "\n",
      "Epoch 144/199\n",
      "----------\n",
      "train Loss: 2.0839 Acc: 0.6337\n",
      "val Loss: 4.3981 Acc: 0.1194\n",
      "\n",
      "Epoch 145/199\n",
      "----------\n",
      "train Loss: 2.0377 Acc: 0.5928\n",
      "val Loss: 4.3537 Acc: 0.1254\n",
      "\n",
      "Epoch 146/199\n",
      "----------\n",
      "train Loss: 2.0306 Acc: 0.6265\n",
      "val Loss: 4.3705 Acc: 0.1313\n",
      "\n",
      "Epoch 147/199\n",
      "----------\n",
      "train Loss: 2.0423 Acc: 0.6120\n",
      "val Loss: 4.3914 Acc: 0.1194\n",
      "\n",
      "Epoch 148/199\n",
      "----------\n",
      "train Loss: 2.0807 Acc: 0.6337\n",
      "val Loss: 4.3940 Acc: 0.1284\n",
      "\n",
      "Epoch 149/199\n",
      "----------\n",
      "train Loss: 2.0191 Acc: 0.6289\n",
      "val Loss: 4.4016 Acc: 0.1164\n",
      "\n",
      "Epoch 150/199\n",
      "----------\n",
      "train Loss: 1.9341 Acc: 0.6506\n",
      "val Loss: 4.3375 Acc: 0.1284\n",
      "\n",
      "Epoch 151/199\n",
      "----------\n",
      "train Loss: 1.9715 Acc: 0.6313\n",
      "val Loss: 4.3834 Acc: 0.1224\n",
      "\n",
      "Epoch 152/199\n",
      "----------\n",
      "train Loss: 1.9574 Acc: 0.6482\n",
      "val Loss: 4.3944 Acc: 0.1224\n",
      "\n",
      "Epoch 153/199\n",
      "----------\n",
      "train Loss: 1.9667 Acc: 0.6410\n",
      "val Loss: 4.3696 Acc: 0.1313\n",
      "\n",
      "Epoch 154/199\n",
      "----------\n",
      "train Loss: 1.9611 Acc: 0.6482\n",
      "val Loss: 4.3969 Acc: 0.1284\n",
      "\n",
      "Epoch 155/199\n",
      "----------\n",
      "train Loss: 1.9558 Acc: 0.6169\n",
      "val Loss: 4.3884 Acc: 0.1224\n",
      "\n",
      "Epoch 156/199\n",
      "----------\n",
      "train Loss: 1.9847 Acc: 0.6145\n",
      "val Loss: 4.3937 Acc: 0.1104\n",
      "\n",
      "Epoch 157/199\n",
      "----------\n",
      "train Loss: 1.8590 Acc: 0.6506\n",
      "val Loss: 4.4246 Acc: 0.1194\n",
      "\n",
      "Epoch 158/199\n",
      "----------\n",
      "train Loss: 1.9114 Acc: 0.6530\n",
      "val Loss: 4.3890 Acc: 0.1284\n",
      "\n",
      "Epoch 159/199\n",
      "----------\n",
      "train Loss: 1.8045 Acc: 0.6916\n",
      "val Loss: 4.3648 Acc: 0.1224\n",
      "\n",
      "Epoch 160/199\n",
      "----------\n",
      "train Loss: 1.9128 Acc: 0.6482\n",
      "val Loss: 4.3916 Acc: 0.1284\n",
      "\n",
      "Epoch 161/199\n",
      "----------\n",
      "train Loss: 1.8749 Acc: 0.6530\n",
      "val Loss: 4.4089 Acc: 0.1254\n",
      "\n",
      "Epoch 162/199\n",
      "----------\n",
      "train Loss: 1.7823 Acc: 0.6940\n",
      "val Loss: 4.4286 Acc: 0.1194\n",
      "\n",
      "Epoch 163/199\n",
      "----------\n",
      "train Loss: 1.8109 Acc: 0.6458\n",
      "val Loss: 4.4958 Acc: 0.1254\n",
      "\n",
      "Epoch 164/199\n",
      "----------\n",
      "train Loss: 1.7965 Acc: 0.6506\n",
      "val Loss: 4.4841 Acc: 0.1373\n",
      "\n",
      "Epoch 165/199\n",
      "----------\n",
      "train Loss: 1.8166 Acc: 0.6578\n",
      "val Loss: 4.4357 Acc: 0.1284\n",
      "\n",
      "Epoch 166/199\n",
      "----------\n",
      "train Loss: 1.7758 Acc: 0.6964\n",
      "val Loss: 4.4231 Acc: 0.1313\n",
      "\n",
      "Epoch 167/199\n",
      "----------\n",
      "train Loss: 1.7312 Acc: 0.7084\n",
      "val Loss: 4.4317 Acc: 0.1313\n",
      "\n",
      "Epoch 168/199\n",
      "----------\n",
      "train Loss: 1.7594 Acc: 0.6651\n",
      "val Loss: 4.4491 Acc: 0.1343\n",
      "\n",
      "Epoch 169/199\n",
      "----------\n",
      "train Loss: 1.7627 Acc: 0.6940\n",
      "val Loss: 4.3988 Acc: 0.1284\n",
      "\n",
      "Epoch 170/199\n",
      "----------\n",
      "train Loss: 1.7304 Acc: 0.6675\n",
      "val Loss: 4.3833 Acc: 0.1343\n",
      "\n",
      "Epoch 171/199\n",
      "----------\n",
      "train Loss: 1.6344 Acc: 0.7277\n",
      "val Loss: 4.3801 Acc: 0.1343\n",
      "\n",
      "Epoch 172/199\n",
      "----------\n",
      "train Loss: 1.6908 Acc: 0.6867\n",
      "val Loss: 4.4592 Acc: 0.1194\n",
      "\n",
      "Epoch 173/199\n",
      "----------\n",
      "train Loss: 1.7036 Acc: 0.7036\n",
      "val Loss: 4.4562 Acc: 0.1373\n",
      "\n",
      "Epoch 174/199\n",
      "----------\n",
      "train Loss: 1.6816 Acc: 0.7157\n",
      "val Loss: 4.4048 Acc: 0.1433\n",
      "\n",
      "Epoch 175/199\n",
      "----------\n",
      "train Loss: 1.6218 Acc: 0.7157\n",
      "val Loss: 4.4293 Acc: 0.1313\n",
      "\n",
      "Epoch 176/199\n",
      "----------\n",
      "train Loss: 1.6280 Acc: 0.7084\n",
      "val Loss: 4.4448 Acc: 0.1403\n",
      "\n",
      "Epoch 177/199\n",
      "----------\n",
      "train Loss: 1.5315 Acc: 0.7614\n",
      "val Loss: 4.4128 Acc: 0.1254\n",
      "\n",
      "Epoch 178/199\n",
      "----------\n",
      "train Loss: 1.6343 Acc: 0.7253\n",
      "val Loss: 4.4444 Acc: 0.1284\n",
      "\n",
      "Epoch 179/199\n",
      "----------\n",
      "train Loss: 1.5802 Acc: 0.7494\n",
      "val Loss: 4.4921 Acc: 0.1343\n",
      "\n",
      "Epoch 180/199\n",
      "----------\n",
      "train Loss: 1.5290 Acc: 0.7807\n",
      "val Loss: 4.4465 Acc: 0.1433\n",
      "\n",
      "Epoch 181/199\n",
      "----------\n",
      "train Loss: 1.5717 Acc: 0.7229\n",
      "val Loss: 4.4123 Acc: 0.1403\n",
      "\n",
      "Epoch 182/199\n",
      "----------\n",
      "train Loss: 1.4668 Acc: 0.7807\n",
      "val Loss: 4.4227 Acc: 0.1463\n",
      "\n",
      "Epoch 183/199\n",
      "----------\n",
      "train Loss: 1.5363 Acc: 0.7663\n",
      "val Loss: 4.4633 Acc: 0.1403\n",
      "\n",
      "Epoch 184/199\n",
      "----------\n",
      "train Loss: 1.4877 Acc: 0.7422\n",
      "val Loss: 4.4515 Acc: 0.1373\n",
      "\n",
      "Epoch 185/199\n",
      "----------\n",
      "train Loss: 1.4527 Acc: 0.7687\n",
      "val Loss: 4.4441 Acc: 0.1254\n",
      "\n",
      "Epoch 186/199\n",
      "----------\n",
      "train Loss: 1.4726 Acc: 0.7590\n",
      "val Loss: 4.4591 Acc: 0.1284\n",
      "\n",
      "Epoch 187/199\n",
      "----------\n",
      "train Loss: 1.5408 Acc: 0.7253\n",
      "val Loss: 4.4623 Acc: 0.1254\n",
      "\n",
      "Epoch 188/199\n",
      "----------\n",
      "train Loss: 1.3885 Acc: 0.7711\n",
      "val Loss: 4.4385 Acc: 0.1284\n",
      "\n",
      "Epoch 189/199\n",
      "----------\n",
      "train Loss: 1.5075 Acc: 0.7470\n",
      "val Loss: 4.4618 Acc: 0.1343\n",
      "\n",
      "Epoch 190/199\n",
      "----------\n",
      "train Loss: 1.4671 Acc: 0.7373\n",
      "val Loss: 4.4932 Acc: 0.1403\n",
      "\n",
      "Epoch 191/199\n",
      "----------\n",
      "train Loss: 1.4373 Acc: 0.7470\n",
      "val Loss: 4.4411 Acc: 0.1194\n",
      "\n",
      "Epoch 192/199\n",
      "----------\n",
      "train Loss: 1.4014 Acc: 0.7807\n",
      "val Loss: 4.4552 Acc: 0.1284\n",
      "\n",
      "Epoch 193/199\n",
      "----------\n",
      "train Loss: 1.3859 Acc: 0.7566\n",
      "val Loss: 4.4769 Acc: 0.1373\n",
      "\n",
      "Epoch 194/199\n",
      "----------\n",
      "train Loss: 1.3953 Acc: 0.7735\n",
      "val Loss: 4.4622 Acc: 0.1433\n",
      "\n",
      "Epoch 195/199\n",
      "----------\n",
      "train Loss: 1.3622 Acc: 0.7976\n",
      "val Loss: 4.5071 Acc: 0.1164\n",
      "\n",
      "Epoch 196/199\n",
      "----------\n",
      "train Loss: 1.4206 Acc: 0.7663\n",
      "val Loss: 4.5113 Acc: 0.1254\n",
      "\n",
      "Epoch 197/199\n",
      "----------\n",
      "train Loss: 1.3352 Acc: 0.8024\n",
      "val Loss: 4.4522 Acc: 0.1313\n",
      "\n",
      "Epoch 198/199\n",
      "----------\n",
      "train Loss: 1.3870 Acc: 0.7711\n",
      "val Loss: 4.4713 Acc: 0.1403\n",
      "\n",
      "Epoch 199/199\n",
      "----------\n",
      "train Loss: 1.4945 Acc: 0.7277\n",
      "val Loss: 4.4713 Acc: 0.1463\n",
      "\n",
      "Training complete in 25m 25s\n",
      "Best val Acc: 0.146269\n"
     ]
    }
   ],
   "source": [
    "model = CNNEarDetector(model=\"resnet\", num_classes=train_data.get_num_classes(), extract_features=False)\n",
    "\n",
    "# Setup the loss fxn\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "\n",
    "# Train and evaluate\n",
    "num_epochs = 200\n",
    "hist = model.train(dataloaders_dict, criterion, num_epochs=num_epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "8a8d2d6c-dcbe-4d5a-876d-3f5d7e192c53",
   "metadata": {
    "executionInfo": {
     "elapsed": 4675,
     "status": "ok",
     "timestamp": 1642412770347,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "8a8d2d6c-dcbe-4d5a-876d-3f5d7e192c53"
   },
   "outputs": [],
   "source": [
    "model.save(os.getcwd(), \"trained_resnet\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "664a74bc-a4e1-45d5-8230-b141ea879717",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "executionInfo": {
     "elapsed": 868,
     "status": "ok",
     "timestamp": 1642412776427,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "664a74bc-a4e1-45d5-8230-b141ea879717",
    "outputId": "4a3131b9-4703-4710-f2d2-8bd44f5069da"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "val_hist = [h.cpu().numpy() for h in hist]\n",
    "\n",
    "plt.title(\"Validation Accuracy vs. Number of Training Epochs\")\n",
    "plt.xlabel(\"Training Epochs\")\n",
    "plt.ylabel(\"Validation Accuracy\")\n",
    "plt.plot(range(1,num_epochs+1), val_hist)\n",
    "plt.savefig(os.getcwd()+\"/resnet_val.png\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "q55xtc0Ghj0i",
   "metadata": {
    "id": "q55xtc0Ghj0i"
   },
   "source": [
    "### Histogram equalization\n",
    "\n",
    "#### Inspect data loading"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "nBiALfpEhrj2",
   "metadata": {
    "id": "nBiALfpEhrj2"
   },
   "outputs": [],
   "source": [
    "train_data = AWETrainSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), transforms.Compose([HistogramEqualization()]))\n",
    "train_data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "JdFPWWytkUfa",
   "metadata": {
    "id": "JdFPWWytkUfa"
   },
   "outputs": [],
   "source": [
    "image, label = train_data[0]\n",
    "plt.imshow(image)\n",
    "print(label)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "68OIqdQ4ifsW",
   "metadata": {
    "id": "68OIqdQ4ifsW"
   },
   "source": [
    "#### Load train and val data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "B8NNdXEBiprW",
   "metadata": {
    "executionInfo": {
     "elapsed": 275,
     "status": "ok",
     "timestamp": 1642412911051,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "B8NNdXEBiprW"
   },
   "outputs": [],
   "source": [
    "train_data = AWETrainSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), data_transforms_equalization[\"train\"])\n",
    "val_data = AWEValSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), data_transforms_equalization[\"val\"])\n",
    "\n",
    "# Create training and validation dataloaders\n",
    "dataloaders_dict = {'train': torch.utils.data.DataLoader(train_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=2),\n",
    "                   'val': torch.utils.data.DataLoader(val_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=2)}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ektXabjsjNUr",
   "metadata": {
    "id": "ektXabjsjNUr"
   },
   "source": [
    "#### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "rTxsWmxxjSoJ",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 1528681,
     "status": "ok",
     "timestamp": 1642414452291,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "rTxsWmxxjSoJ",
    "outputId": "e76631d6-31d5-40f3-c3d3-83e7cc9e8147"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training model on GPUs\n",
      "Epoch 0/199\n",
      "----------\n",
      "train Loss: 4.6803 Acc: 0.0072\n",
      "val Loss: 4.6367 Acc: 0.0179\n",
      "\n",
      "Epoch 1/199\n",
      "----------\n",
      "train Loss: 4.6443 Acc: 0.0072\n",
      "val Loss: 4.6273 Acc: 0.0090\n",
      "\n",
      "Epoch 2/199\n",
      "----------\n",
      "train Loss: 4.6249 Acc: 0.0096\n",
      "val Loss: 4.6186 Acc: 0.0119\n",
      "\n",
      "Epoch 3/199\n",
      "----------\n",
      "train Loss: 4.6177 Acc: 0.0096\n",
      "val Loss: 4.6008 Acc: 0.0119\n",
      "\n",
      "Epoch 4/199\n",
      "----------\n",
      "train Loss: 4.5880 Acc: 0.0120\n",
      "val Loss: 4.5905 Acc: 0.0179\n",
      "\n",
      "Epoch 5/199\n",
      "----------\n",
      "train Loss: 4.5755 Acc: 0.0193\n",
      "val Loss: 4.5816 Acc: 0.0119\n",
      "\n",
      "Epoch 6/199\n",
      "----------\n",
      "train Loss: 4.5705 Acc: 0.0096\n",
      "val Loss: 4.5764 Acc: 0.0239\n",
      "\n",
      "Epoch 7/199\n",
      "----------\n",
      "train Loss: 4.5632 Acc: 0.0241\n",
      "val Loss: 4.5710 Acc: 0.0299\n",
      "\n",
      "Epoch 8/199\n",
      "----------\n",
      "train Loss: 4.5461 Acc: 0.0145\n",
      "val Loss: 4.5692 Acc: 0.0358\n",
      "\n",
      "Epoch 9/199\n",
      "----------\n",
      "train Loss: 4.5332 Acc: 0.0265\n",
      "val Loss: 4.5577 Acc: 0.0537\n",
      "\n",
      "Epoch 10/199\n",
      "----------\n",
      "train Loss: 4.5221 Acc: 0.0434\n",
      "val Loss: 4.5612 Acc: 0.0269\n",
      "\n",
      "Epoch 11/199\n",
      "----------\n",
      "train Loss: 4.5112 Acc: 0.0410\n",
      "val Loss: 4.5461 Acc: 0.0358\n",
      "\n",
      "Epoch 12/199\n",
      "----------\n",
      "train Loss: 4.5075 Acc: 0.0289\n",
      "val Loss: 4.5471 Acc: 0.0209\n",
      "\n",
      "Epoch 13/199\n",
      "----------\n",
      "train Loss: 4.4899 Acc: 0.0361\n",
      "val Loss: 4.5382 Acc: 0.0358\n",
      "\n",
      "Epoch 14/199\n",
      "----------\n",
      "train Loss: 4.4686 Acc: 0.0410\n",
      "val Loss: 4.5373 Acc: 0.0358\n",
      "\n",
      "Epoch 15/199\n",
      "----------\n",
      "train Loss: 4.4560 Acc: 0.0482\n",
      "val Loss: 4.5257 Acc: 0.0358\n",
      "\n",
      "Epoch 16/199\n",
      "----------\n",
      "train Loss: 4.4389 Acc: 0.0627\n",
      "val Loss: 4.5290 Acc: 0.0388\n",
      "\n",
      "Epoch 17/199\n",
      "----------\n",
      "train Loss: 4.4305 Acc: 0.0627\n",
      "val Loss: 4.5263 Acc: 0.0388\n",
      "\n",
      "Epoch 18/199\n",
      "----------\n",
      "train Loss: 4.4148 Acc: 0.0651\n",
      "val Loss: 4.5237 Acc: 0.0358\n",
      "\n",
      "Epoch 19/199\n",
      "----------\n",
      "train Loss: 4.4173 Acc: 0.0723\n",
      "val Loss: 4.5133 Acc: 0.0358\n",
      "\n",
      "Epoch 20/199\n",
      "----------\n",
      "train Loss: 4.3769 Acc: 0.0819\n",
      "val Loss: 4.5052 Acc: 0.0269\n",
      "\n",
      "Epoch 21/199\n",
      "----------\n",
      "train Loss: 4.3726 Acc: 0.1012\n",
      "val Loss: 4.4989 Acc: 0.0358\n",
      "\n",
      "Epoch 22/199\n",
      "----------\n",
      "train Loss: 4.3483 Acc: 0.0940\n",
      "val Loss: 4.4927 Acc: 0.0567\n",
      "\n",
      "Epoch 23/199\n",
      "----------\n",
      "train Loss: 4.3523 Acc: 0.0988\n",
      "val Loss: 4.4871 Acc: 0.0478\n",
      "\n",
      "Epoch 24/199\n",
      "----------\n",
      "train Loss: 4.3332 Acc: 0.0892\n",
      "val Loss: 4.4766 Acc: 0.0418\n",
      "\n",
      "Epoch 25/199\n",
      "----------\n",
      "train Loss: 4.3098 Acc: 0.1325\n",
      "val Loss: 4.4770 Acc: 0.0388\n",
      "\n",
      "Epoch 26/199\n",
      "----------\n",
      "train Loss: 4.2788 Acc: 0.1108\n",
      "val Loss: 4.4734 Acc: 0.0537\n",
      "\n",
      "Epoch 27/199\n",
      "----------\n",
      "train Loss: 4.2824 Acc: 0.1277\n",
      "val Loss: 4.4604 Acc: 0.0507\n",
      "\n",
      "Epoch 28/199\n",
      "----------\n",
      "train Loss: 4.2756 Acc: 0.1398\n",
      "val Loss: 4.4426 Acc: 0.0507\n",
      "\n",
      "Epoch 29/199\n",
      "----------\n",
      "train Loss: 4.2357 Acc: 0.1518\n",
      "val Loss: 4.4494 Acc: 0.0507\n",
      "\n",
      "Epoch 30/199\n",
      "----------\n",
      "train Loss: 4.2354 Acc: 0.1422\n",
      "val Loss: 4.4307 Acc: 0.0567\n",
      "\n",
      "Epoch 31/199\n",
      "----------\n",
      "train Loss: 4.2088 Acc: 0.1470\n",
      "val Loss: 4.4359 Acc: 0.0478\n",
      "\n",
      "Epoch 32/199\n",
      "----------\n",
      "train Loss: 4.1777 Acc: 0.1373\n",
      "val Loss: 4.4342 Acc: 0.0478\n",
      "\n",
      "Epoch 33/199\n",
      "----------\n",
      "train Loss: 4.1747 Acc: 0.1470\n",
      "val Loss: 4.4324 Acc: 0.0597\n",
      "\n",
      "Epoch 34/199\n",
      "----------\n",
      "train Loss: 4.1668 Acc: 0.1855\n",
      "val Loss: 4.4204 Acc: 0.0627\n",
      "\n",
      "Epoch 35/199\n",
      "----------\n",
      "train Loss: 4.1360 Acc: 0.1904\n",
      "val Loss: 4.4197 Acc: 0.0627\n",
      "\n",
      "Epoch 36/199\n",
      "----------\n",
      "train Loss: 4.1227 Acc: 0.1446\n",
      "val Loss: 4.4073 Acc: 0.0627\n",
      "\n",
      "Epoch 37/199\n",
      "----------\n",
      "train Loss: 4.1097 Acc: 0.1735\n",
      "val Loss: 4.4054 Acc: 0.0657\n",
      "\n",
      "Epoch 38/199\n",
      "----------\n",
      "train Loss: 4.1216 Acc: 0.1446\n",
      "val Loss: 4.3953 Acc: 0.0716\n",
      "\n",
      "Epoch 39/199\n",
      "----------\n",
      "train Loss: 4.0635 Acc: 0.2000\n",
      "val Loss: 4.3966 Acc: 0.0716\n",
      "\n",
      "Epoch 40/199\n",
      "----------\n",
      "train Loss: 4.0543 Acc: 0.1952\n",
      "val Loss: 4.3902 Acc: 0.0567\n",
      "\n",
      "Epoch 41/199\n",
      "----------\n",
      "train Loss: 4.0190 Acc: 0.1952\n",
      "val Loss: 4.3845 Acc: 0.0657\n",
      "\n",
      "Epoch 42/199\n",
      "----------\n",
      "train Loss: 4.0206 Acc: 0.1880\n",
      "val Loss: 4.3610 Acc: 0.0746\n",
      "\n",
      "Epoch 43/199\n",
      "----------\n",
      "train Loss: 3.9718 Acc: 0.2145\n",
      "val Loss: 4.3711 Acc: 0.0627\n",
      "\n",
      "Epoch 44/199\n",
      "----------\n",
      "train Loss: 3.9567 Acc: 0.2024\n",
      "val Loss: 4.3579 Acc: 0.0657\n",
      "\n",
      "Epoch 45/199\n",
      "----------\n",
      "train Loss: 3.9432 Acc: 0.2145\n",
      "val Loss: 4.3424 Acc: 0.0806\n",
      "\n",
      "Epoch 46/199\n",
      "----------\n",
      "train Loss: 3.9439 Acc: 0.1976\n",
      "val Loss: 4.3333 Acc: 0.0657\n",
      "\n",
      "Epoch 47/199\n",
      "----------\n",
      "train Loss: 3.9101 Acc: 0.2361\n",
      "val Loss: 4.3528 Acc: 0.0746\n",
      "\n",
      "Epoch 48/199\n",
      "----------\n",
      "train Loss: 3.8947 Acc: 0.2289\n",
      "val Loss: 4.3370 Acc: 0.0746\n",
      "\n",
      "Epoch 49/199\n",
      "----------\n",
      "train Loss: 3.8710 Acc: 0.2217\n",
      "val Loss: 4.3343 Acc: 0.0657\n",
      "\n",
      "Epoch 50/199\n",
      "----------\n",
      "train Loss: 3.8484 Acc: 0.2530\n",
      "val Loss: 4.3275 Acc: 0.0687\n",
      "\n",
      "Epoch 51/199\n",
      "----------\n",
      "train Loss: 3.8309 Acc: 0.2120\n",
      "val Loss: 4.3206 Acc: 0.0657\n",
      "\n",
      "Epoch 52/199\n",
      "----------\n",
      "train Loss: 3.8428 Acc: 0.2217\n",
      "val Loss: 4.3190 Acc: 0.0597\n",
      "\n",
      "Epoch 53/199\n",
      "----------\n",
      "train Loss: 3.7535 Acc: 0.2602\n",
      "val Loss: 4.3118 Acc: 0.0597\n",
      "\n",
      "Epoch 54/199\n",
      "----------\n",
      "train Loss: 3.7599 Acc: 0.2386\n",
      "val Loss: 4.3343 Acc: 0.0657\n",
      "\n",
      "Epoch 55/199\n",
      "----------\n",
      "train Loss: 3.7635 Acc: 0.2530\n",
      "val Loss: 4.2964 Acc: 0.0657\n",
      "\n",
      "Epoch 56/199\n",
      "----------\n",
      "train Loss: 3.7557 Acc: 0.2193\n",
      "val Loss: 4.3199 Acc: 0.0687\n",
      "\n",
      "Epoch 57/199\n",
      "----------\n",
      "train Loss: 3.7297 Acc: 0.2530\n",
      "val Loss: 4.2968 Acc: 0.0746\n",
      "\n",
      "Epoch 58/199\n",
      "----------\n",
      "train Loss: 3.7133 Acc: 0.2723\n",
      "val Loss: 4.2961 Acc: 0.0716\n",
      "\n",
      "Epoch 59/199\n",
      "----------\n",
      "train Loss: 3.7040 Acc: 0.2578\n",
      "val Loss: 4.2940 Acc: 0.0687\n",
      "\n",
      "Epoch 60/199\n",
      "----------\n",
      "train Loss: 3.6205 Acc: 0.2892\n",
      "val Loss: 4.2979 Acc: 0.0746\n",
      "\n",
      "Epoch 61/199\n",
      "----------\n",
      "train Loss: 3.6249 Acc: 0.2602\n",
      "val Loss: 4.2666 Acc: 0.0716\n",
      "\n",
      "Epoch 62/199\n",
      "----------\n",
      "train Loss: 3.6333 Acc: 0.2916\n",
      "val Loss: 4.2723 Acc: 0.0836\n",
      "\n",
      "Epoch 63/199\n",
      "----------\n",
      "train Loss: 3.5907 Acc: 0.3036\n",
      "val Loss: 4.2650 Acc: 0.0746\n",
      "\n",
      "Epoch 64/199\n",
      "----------\n",
      "train Loss: 3.5751 Acc: 0.2482\n",
      "val Loss: 4.2684 Acc: 0.0896\n",
      "\n",
      "Epoch 65/199\n",
      "----------\n",
      "train Loss: 3.5526 Acc: 0.2916\n",
      "val Loss: 4.2821 Acc: 0.0776\n",
      "\n",
      "Epoch 66/199\n",
      "----------\n",
      "train Loss: 3.5702 Acc: 0.2940\n",
      "val Loss: 4.2532 Acc: 0.0836\n",
      "\n",
      "Epoch 67/199\n",
      "----------\n",
      "train Loss: 3.4979 Acc: 0.3084\n",
      "val Loss: 4.2478 Acc: 0.0806\n",
      "\n",
      "Epoch 68/199\n",
      "----------\n",
      "train Loss: 3.4816 Acc: 0.3157\n",
      "val Loss: 4.2699 Acc: 0.0746\n",
      "\n",
      "Epoch 69/199\n",
      "----------\n",
      "train Loss: 3.4954 Acc: 0.2940\n",
      "val Loss: 4.2746 Acc: 0.0687\n",
      "\n",
      "Epoch 70/199\n",
      "----------\n",
      "train Loss: 3.4628 Acc: 0.2940\n",
      "val Loss: 4.2545 Acc: 0.0776\n",
      "\n",
      "Epoch 71/199\n",
      "----------\n",
      "train Loss: 3.4108 Acc: 0.3301\n",
      "val Loss: 4.2650 Acc: 0.0866\n",
      "\n",
      "Epoch 72/199\n",
      "----------\n",
      "train Loss: 3.3994 Acc: 0.3325\n",
      "val Loss: 4.2575 Acc: 0.0776\n",
      "\n",
      "Epoch 73/199\n",
      "----------\n",
      "train Loss: 3.4111 Acc: 0.3253\n",
      "val Loss: 4.2365 Acc: 0.0776\n",
      "\n",
      "Epoch 74/199\n",
      "----------\n",
      "train Loss: 3.3605 Acc: 0.3518\n",
      "val Loss: 4.2424 Acc: 0.0806\n",
      "\n",
      "Epoch 75/199\n",
      "----------\n",
      "train Loss: 3.3727 Acc: 0.3084\n",
      "val Loss: 4.2372 Acc: 0.0806\n",
      "\n",
      "Epoch 76/199\n",
      "----------\n",
      "train Loss: 3.3090 Acc: 0.3398\n",
      "val Loss: 4.2355 Acc: 0.0806\n",
      "\n",
      "Epoch 77/199\n",
      "----------\n",
      "train Loss: 3.2813 Acc: 0.3735\n",
      "val Loss: 4.2214 Acc: 0.0836\n",
      "\n",
      "Epoch 78/199\n",
      "----------\n",
      "train Loss: 3.3306 Acc: 0.3301\n",
      "val Loss: 4.2283 Acc: 0.0806\n",
      "\n",
      "Epoch 79/199\n",
      "----------\n",
      "train Loss: 3.2611 Acc: 0.3614\n",
      "val Loss: 4.2605 Acc: 0.0896\n",
      "\n",
      "Epoch 80/199\n",
      "----------\n",
      "train Loss: 3.2507 Acc: 0.3446\n",
      "val Loss: 4.2323 Acc: 0.0716\n",
      "\n",
      "Epoch 81/199\n",
      "----------\n",
      "train Loss: 3.2577 Acc: 0.3566\n",
      "val Loss: 4.2503 Acc: 0.0776\n",
      "\n",
      "Epoch 82/199\n",
      "----------\n",
      "train Loss: 3.1885 Acc: 0.3687\n",
      "val Loss: 4.2561 Acc: 0.0955\n",
      "\n",
      "Epoch 83/199\n",
      "----------\n",
      "train Loss: 3.2042 Acc: 0.4000\n",
      "val Loss: 4.2309 Acc: 0.0836\n",
      "\n",
      "Epoch 84/199\n",
      "----------\n",
      "train Loss: 3.1986 Acc: 0.3687\n",
      "val Loss: 4.2320 Acc: 0.0985\n",
      "\n",
      "Epoch 85/199\n",
      "----------\n",
      "train Loss: 3.1490 Acc: 0.3807\n",
      "val Loss: 4.2256 Acc: 0.0866\n",
      "\n",
      "Epoch 86/199\n",
      "----------\n",
      "train Loss: 3.0949 Acc: 0.4096\n",
      "val Loss: 4.2420 Acc: 0.0866\n",
      "\n",
      "Epoch 87/199\n",
      "----------\n",
      "train Loss: 3.0916 Acc: 0.3952\n",
      "val Loss: 4.2106 Acc: 0.1075\n",
      "\n",
      "Epoch 88/199\n",
      "----------\n",
      "train Loss: 3.0920 Acc: 0.3855\n",
      "val Loss: 4.2335 Acc: 0.0955\n",
      "\n",
      "Epoch 89/199\n",
      "----------\n",
      "train Loss: 3.1311 Acc: 0.3614\n",
      "val Loss: 4.2293 Acc: 0.0896\n",
      "\n",
      "Epoch 90/199\n",
      "----------\n",
      "train Loss: 3.0745 Acc: 0.4120\n",
      "val Loss: 4.2371 Acc: 0.0896\n",
      "\n",
      "Epoch 91/199\n",
      "----------\n",
      "train Loss: 3.0469 Acc: 0.4120\n",
      "val Loss: 4.2627 Acc: 0.0896\n",
      "\n",
      "Epoch 92/199\n",
      "----------\n",
      "train Loss: 3.0388 Acc: 0.3855\n",
      "val Loss: 4.2599 Acc: 0.0896\n",
      "\n",
      "Epoch 93/199\n",
      "----------\n",
      "train Loss: 3.0137 Acc: 0.4096\n",
      "val Loss: 4.2525 Acc: 0.0985\n",
      "\n",
      "Epoch 94/199\n",
      "----------\n",
      "train Loss: 3.0256 Acc: 0.3952\n",
      "val Loss: 4.2246 Acc: 0.1015\n",
      "\n",
      "Epoch 95/199\n",
      "----------\n",
      "train Loss: 2.9745 Acc: 0.3928\n",
      "val Loss: 4.2103 Acc: 0.1045\n",
      "\n",
      "Epoch 96/199\n",
      "----------\n",
      "train Loss: 2.9581 Acc: 0.4048\n",
      "val Loss: 4.2243 Acc: 0.0896\n",
      "\n",
      "Epoch 97/199\n",
      "----------\n",
      "train Loss: 2.9470 Acc: 0.4602\n",
      "val Loss: 4.2160 Acc: 0.0955\n",
      "\n",
      "Epoch 98/199\n",
      "----------\n",
      "train Loss: 2.8783 Acc: 0.4482\n",
      "val Loss: 4.2145 Acc: 0.0925\n",
      "\n",
      "Epoch 99/199\n",
      "----------\n",
      "train Loss: 2.8795 Acc: 0.4241\n",
      "val Loss: 4.2208 Acc: 0.0896\n",
      "\n",
      "Epoch 100/199\n",
      "----------\n",
      "train Loss: 2.8594 Acc: 0.4217\n",
      "val Loss: 4.2027 Acc: 0.1045\n",
      "\n",
      "Epoch 101/199\n",
      "----------\n",
      "train Loss: 2.8583 Acc: 0.4458\n",
      "val Loss: 4.2311 Acc: 0.1045\n",
      "\n",
      "Epoch 102/199\n",
      "----------\n",
      "train Loss: 2.8228 Acc: 0.4337\n",
      "val Loss: 4.2417 Acc: 0.0955\n",
      "\n",
      "Epoch 103/199\n",
      "----------\n",
      "train Loss: 2.7896 Acc: 0.4578\n",
      "val Loss: 4.2349 Acc: 0.0955\n",
      "\n",
      "Epoch 104/199\n",
      "----------\n",
      "train Loss: 2.8290 Acc: 0.4554\n",
      "val Loss: 4.2278 Acc: 0.1075\n",
      "\n",
      "Epoch 105/199\n",
      "----------\n",
      "train Loss: 2.8055 Acc: 0.4699\n",
      "val Loss: 4.2363 Acc: 0.1075\n",
      "\n",
      "Epoch 106/199\n",
      "----------\n",
      "train Loss: 2.7611 Acc: 0.4265\n",
      "val Loss: 4.2442 Acc: 0.1104\n",
      "\n",
      "Epoch 107/199\n",
      "----------\n",
      "train Loss: 2.8055 Acc: 0.4554\n",
      "val Loss: 4.2213 Acc: 0.0955\n",
      "\n",
      "Epoch 108/199\n",
      "----------\n",
      "train Loss: 2.7580 Acc: 0.4554\n",
      "val Loss: 4.2166 Acc: 0.1075\n",
      "\n",
      "Epoch 109/199\n",
      "----------\n",
      "train Loss: 2.7662 Acc: 0.4843\n",
      "val Loss: 4.2154 Acc: 0.1164\n",
      "\n",
      "Epoch 110/199\n",
      "----------\n",
      "train Loss: 2.7930 Acc: 0.4554\n",
      "val Loss: 4.2386 Acc: 0.1045\n",
      "\n",
      "Epoch 111/199\n",
      "----------\n",
      "train Loss: 2.7251 Acc: 0.4482\n",
      "val Loss: 4.2345 Acc: 0.1045\n",
      "\n",
      "Epoch 112/199\n",
      "----------\n",
      "train Loss: 2.6919 Acc: 0.4675\n",
      "val Loss: 4.2217 Acc: 0.1104\n",
      "\n",
      "Epoch 113/199\n",
      "----------\n",
      "train Loss: 2.6273 Acc: 0.4988\n",
      "val Loss: 4.2238 Acc: 0.1104\n",
      "\n",
      "Epoch 114/199\n",
      "----------\n",
      "train Loss: 2.6509 Acc: 0.5012\n",
      "val Loss: 4.2027 Acc: 0.0985\n",
      "\n",
      "Epoch 115/199\n",
      "----------\n",
      "train Loss: 2.6423 Acc: 0.5133\n",
      "val Loss: 4.1967 Acc: 0.1075\n",
      "\n",
      "Epoch 116/199\n",
      "----------\n",
      "train Loss: 2.5994 Acc: 0.5205\n",
      "val Loss: 4.2088 Acc: 0.0985\n",
      "\n",
      "Epoch 117/199\n",
      "----------\n",
      "train Loss: 2.5488 Acc: 0.5277\n",
      "val Loss: 4.1992 Acc: 0.1104\n",
      "\n",
      "Epoch 118/199\n",
      "----------\n",
      "train Loss: 2.5100 Acc: 0.5398\n",
      "val Loss: 4.2005 Acc: 0.1075\n",
      "\n",
      "Epoch 119/199\n",
      "----------\n",
      "train Loss: 2.5219 Acc: 0.5205\n",
      "val Loss: 4.2007 Acc: 0.1104\n",
      "\n",
      "Epoch 120/199\n",
      "----------\n",
      "train Loss: 2.5262 Acc: 0.5157\n",
      "val Loss: 4.2075 Acc: 0.1254\n",
      "\n",
      "Epoch 121/199\n",
      "----------\n",
      "train Loss: 2.4428 Acc: 0.5711\n",
      "val Loss: 4.1612 Acc: 0.1194\n",
      "\n",
      "Epoch 122/199\n",
      "----------\n",
      "train Loss: 2.5259 Acc: 0.4988\n",
      "val Loss: 4.2003 Acc: 0.1134\n",
      "\n",
      "Epoch 123/199\n",
      "----------\n",
      "train Loss: 2.4342 Acc: 0.5494\n",
      "val Loss: 4.1997 Acc: 0.1164\n",
      "\n",
      "Epoch 124/199\n",
      "----------\n",
      "train Loss: 2.3780 Acc: 0.5687\n",
      "val Loss: 4.1958 Acc: 0.1134\n",
      "\n",
      "Epoch 125/199\n",
      "----------\n",
      "train Loss: 2.4303 Acc: 0.5663\n",
      "val Loss: 4.2259 Acc: 0.1104\n",
      "\n",
      "Epoch 126/199\n",
      "----------\n",
      "train Loss: 2.3460 Acc: 0.5807\n",
      "val Loss: 4.1886 Acc: 0.1164\n",
      "\n",
      "Epoch 127/199\n",
      "----------\n",
      "train Loss: 2.4525 Acc: 0.5157\n",
      "val Loss: 4.2298 Acc: 0.1104\n",
      "\n",
      "Epoch 128/199\n",
      "----------\n",
      "train Loss: 2.2959 Acc: 0.5446\n",
      "val Loss: 4.2096 Acc: 0.1194\n",
      "\n",
      "Epoch 129/199\n",
      "----------\n",
      "train Loss: 2.4136 Acc: 0.5157\n",
      "val Loss: 4.2148 Acc: 0.1104\n",
      "\n",
      "Epoch 130/199\n",
      "----------\n",
      "train Loss: 2.3756 Acc: 0.5422\n",
      "val Loss: 4.2381 Acc: 0.1134\n",
      "\n",
      "Epoch 131/199\n",
      "----------\n",
      "train Loss: 2.3680 Acc: 0.5663\n",
      "val Loss: 4.2201 Acc: 0.1134\n",
      "\n",
      "Epoch 132/199\n",
      "----------\n",
      "train Loss: 2.3801 Acc: 0.5446\n",
      "val Loss: 4.2570 Acc: 0.1224\n",
      "\n",
      "Epoch 133/199\n",
      "----------\n",
      "train Loss: 2.3147 Acc: 0.5735\n",
      "val Loss: 4.2230 Acc: 0.1284\n",
      "\n",
      "Epoch 134/199\n",
      "----------\n",
      "train Loss: 2.3128 Acc: 0.5639\n",
      "val Loss: 4.2003 Acc: 0.1284\n",
      "\n",
      "Epoch 135/199\n",
      "----------\n",
      "train Loss: 2.2688 Acc: 0.5807\n",
      "val Loss: 4.2491 Acc: 0.1224\n",
      "\n",
      "Epoch 136/199\n",
      "----------\n",
      "train Loss: 2.2793 Acc: 0.5614\n",
      "val Loss: 4.2169 Acc: 0.1164\n",
      "\n",
      "Epoch 137/199\n",
      "----------\n",
      "train Loss: 2.2058 Acc: 0.6410\n",
      "val Loss: 4.2107 Acc: 0.1104\n",
      "\n",
      "Epoch 138/199\n",
      "----------\n",
      "train Loss: 2.2246 Acc: 0.5759\n",
      "val Loss: 4.2410 Acc: 0.1254\n",
      "\n",
      "Epoch 139/199\n",
      "----------\n",
      "train Loss: 2.1653 Acc: 0.6048\n",
      "val Loss: 4.2647 Acc: 0.1254\n",
      "\n",
      "Epoch 140/199\n",
      "----------\n",
      "train Loss: 2.1168 Acc: 0.6145\n",
      "val Loss: 4.2784 Acc: 0.1343\n",
      "\n",
      "Epoch 141/199\n",
      "----------\n",
      "train Loss: 2.1873 Acc: 0.6096\n",
      "val Loss: 4.2475 Acc: 0.1194\n",
      "\n",
      "Epoch 142/199\n",
      "----------\n",
      "train Loss: 2.0873 Acc: 0.6361\n",
      "val Loss: 4.2920 Acc: 0.1224\n",
      "\n",
      "Epoch 143/199\n",
      "----------\n",
      "train Loss: 2.0803 Acc: 0.6337\n",
      "val Loss: 4.2632 Acc: 0.1045\n",
      "\n",
      "Epoch 144/199\n",
      "----------\n",
      "train Loss: 2.1020 Acc: 0.6145\n",
      "val Loss: 4.2375 Acc: 0.1284\n",
      "\n",
      "Epoch 145/199\n",
      "----------\n",
      "train Loss: 2.1004 Acc: 0.6193\n",
      "val Loss: 4.2881 Acc: 0.1313\n",
      "\n",
      "Epoch 146/199\n",
      "----------\n",
      "train Loss: 2.1024 Acc: 0.6096\n",
      "val Loss: 4.2534 Acc: 0.1343\n",
      "\n",
      "Epoch 147/199\n",
      "----------\n",
      "train Loss: 2.0845 Acc: 0.6337\n",
      "val Loss: 4.2861 Acc: 0.1224\n",
      "\n",
      "Epoch 148/199\n",
      "----------\n",
      "train Loss: 2.0723 Acc: 0.6024\n",
      "val Loss: 4.2476 Acc: 0.1284\n",
      "\n",
      "Epoch 149/199\n",
      "----------\n",
      "train Loss: 2.0701 Acc: 0.6217\n",
      "val Loss: 4.2564 Acc: 0.1313\n",
      "\n",
      "Epoch 150/199\n",
      "----------\n",
      "train Loss: 1.9665 Acc: 0.6506\n",
      "val Loss: 4.2329 Acc: 0.1403\n",
      "\n",
      "Epoch 151/199\n",
      "----------\n",
      "train Loss: 2.0530 Acc: 0.6193\n",
      "val Loss: 4.2442 Acc: 0.1313\n",
      "\n",
      "Epoch 152/199\n",
      "----------\n",
      "train Loss: 1.9663 Acc: 0.6554\n",
      "val Loss: 4.2521 Acc: 0.1164\n",
      "\n",
      "Epoch 153/199\n",
      "----------\n",
      "train Loss: 1.9353 Acc: 0.6458\n",
      "val Loss: 4.2864 Acc: 0.1284\n",
      "\n",
      "Epoch 154/199\n",
      "----------\n",
      "train Loss: 1.9283 Acc: 0.6627\n",
      "val Loss: 4.2270 Acc: 0.1284\n",
      "\n",
      "Epoch 155/199\n",
      "----------\n",
      "train Loss: 1.9053 Acc: 0.6964\n",
      "val Loss: 4.2633 Acc: 0.1254\n",
      "\n",
      "Epoch 156/199\n",
      "----------\n",
      "train Loss: 2.0047 Acc: 0.6289\n",
      "val Loss: 4.2918 Acc: 0.1313\n",
      "\n",
      "Epoch 157/199\n",
      "----------\n",
      "train Loss: 1.8953 Acc: 0.6964\n",
      "val Loss: 4.2492 Acc: 0.1284\n",
      "\n",
      "Epoch 158/199\n",
      "----------\n",
      "train Loss: 1.9039 Acc: 0.6771\n",
      "val Loss: 4.3067 Acc: 0.1284\n",
      "\n",
      "Epoch 159/199\n",
      "----------\n",
      "train Loss: 1.8944 Acc: 0.6651\n",
      "val Loss: 4.2476 Acc: 0.1493\n",
      "\n",
      "Epoch 160/199\n",
      "----------\n",
      "train Loss: 1.8703 Acc: 0.6675\n",
      "val Loss: 4.2296 Acc: 0.1313\n",
      "\n",
      "Epoch 161/199\n",
      "----------\n",
      "train Loss: 1.8391 Acc: 0.6988\n",
      "val Loss: 4.2858 Acc: 0.1254\n",
      "\n",
      "Epoch 162/199\n",
      "----------\n",
      "train Loss: 1.8189 Acc: 0.6964\n",
      "val Loss: 4.2859 Acc: 0.1343\n",
      "\n",
      "Epoch 163/199\n",
      "----------\n",
      "train Loss: 1.7805 Acc: 0.7060\n",
      "val Loss: 4.2916 Acc: 0.1313\n",
      "\n",
      "Epoch 164/199\n",
      "----------\n",
      "train Loss: 1.7690 Acc: 0.7084\n",
      "val Loss: 4.3081 Acc: 0.1343\n",
      "\n",
      "Epoch 165/199\n",
      "----------\n",
      "train Loss: 1.6583 Acc: 0.7542\n",
      "val Loss: 4.2688 Acc: 0.1373\n",
      "\n",
      "Epoch 166/199\n",
      "----------\n",
      "train Loss: 1.7369 Acc: 0.7133\n",
      "val Loss: 4.2532 Acc: 0.1313\n",
      "\n",
      "Epoch 167/199\n",
      "----------\n",
      "train Loss: 1.7466 Acc: 0.7205\n",
      "val Loss: 4.2893 Acc: 0.1373\n",
      "\n",
      "Epoch 168/199\n",
      "----------\n",
      "train Loss: 1.7236 Acc: 0.7036\n",
      "val Loss: 4.2845 Acc: 0.1373\n",
      "\n",
      "Epoch 169/199\n",
      "----------\n",
      "train Loss: 1.7323 Acc: 0.7181\n",
      "val Loss: 4.2437 Acc: 0.1403\n",
      "\n",
      "Epoch 170/199\n",
      "----------\n",
      "train Loss: 1.7874 Acc: 0.6699\n",
      "val Loss: 4.2467 Acc: 0.1343\n",
      "\n",
      "Epoch 171/199\n",
      "----------\n",
      "train Loss: 1.6686 Acc: 0.7157\n",
      "val Loss: 4.2597 Acc: 0.1373\n",
      "\n",
      "Epoch 172/199\n",
      "----------\n",
      "train Loss: 1.6562 Acc: 0.7084\n",
      "val Loss: 4.2427 Acc: 0.1493\n",
      "\n",
      "Epoch 173/199\n",
      "----------\n",
      "train Loss: 1.6622 Acc: 0.7325\n",
      "val Loss: 4.2396 Acc: 0.1224\n",
      "\n",
      "Epoch 174/199\n",
      "----------\n",
      "train Loss: 1.6056 Acc: 0.7301\n",
      "val Loss: 4.2472 Acc: 0.1284\n",
      "\n",
      "Epoch 175/199\n",
      "----------\n",
      "train Loss: 1.6287 Acc: 0.7349\n",
      "val Loss: 4.2817 Acc: 0.1313\n",
      "\n",
      "Epoch 176/199\n",
      "----------\n",
      "train Loss: 1.5764 Acc: 0.7446\n",
      "val Loss: 4.2809 Acc: 0.1254\n",
      "\n",
      "Epoch 177/199\n",
      "----------\n",
      "train Loss: 1.6507 Acc: 0.7349\n",
      "val Loss: 4.3073 Acc: 0.1373\n",
      "\n",
      "Epoch 178/199\n",
      "----------\n",
      "train Loss: 1.5746 Acc: 0.7446\n",
      "val Loss: 4.2558 Acc: 0.1373\n",
      "\n",
      "Epoch 179/199\n",
      "----------\n",
      "train Loss: 1.5797 Acc: 0.7205\n",
      "val Loss: 4.3187 Acc: 0.1373\n",
      "\n",
      "Epoch 180/199\n",
      "----------\n",
      "train Loss: 1.5108 Acc: 0.7566\n",
      "val Loss: 4.2599 Acc: 0.1224\n",
      "\n",
      "Epoch 181/199\n",
      "----------\n",
      "train Loss: 1.6796 Acc: 0.6964\n",
      "val Loss: 4.3174 Acc: 0.1254\n",
      "\n",
      "Epoch 182/199\n",
      "----------\n",
      "train Loss: 1.5770 Acc: 0.7373\n",
      "val Loss: 4.2759 Acc: 0.1284\n",
      "\n",
      "Epoch 183/199\n",
      "----------\n",
      "train Loss: 1.4924 Acc: 0.7470\n",
      "val Loss: 4.2756 Acc: 0.1373\n",
      "\n",
      "Epoch 184/199\n",
      "----------\n",
      "train Loss: 1.5022 Acc: 0.7277\n",
      "val Loss: 4.3168 Acc: 0.1254\n",
      "\n",
      "Epoch 185/199\n",
      "----------\n",
      "train Loss: 1.4758 Acc: 0.7614\n",
      "val Loss: 4.2944 Acc: 0.1313\n",
      "\n",
      "Epoch 186/199\n",
      "----------\n",
      "train Loss: 1.5268 Acc: 0.7590\n",
      "val Loss: 4.2756 Acc: 0.1373\n",
      "\n",
      "Epoch 187/199\n",
      "----------\n",
      "train Loss: 1.4626 Acc: 0.7759\n",
      "val Loss: 4.3516 Acc: 0.1403\n",
      "\n",
      "Epoch 188/199\n",
      "----------\n",
      "train Loss: 1.3995 Acc: 0.7976\n",
      "val Loss: 4.2978 Acc: 0.1254\n",
      "\n",
      "Epoch 189/199\n",
      "----------\n",
      "train Loss: 1.3807 Acc: 0.7952\n",
      "val Loss: 4.3438 Acc: 0.1313\n",
      "\n",
      "Epoch 190/199\n",
      "----------\n",
      "train Loss: 1.4095 Acc: 0.7735\n",
      "val Loss: 4.2996 Acc: 0.1313\n",
      "\n",
      "Epoch 191/199\n",
      "----------\n",
      "train Loss: 1.3903 Acc: 0.7831\n",
      "val Loss: 4.2712 Acc: 0.1343\n",
      "\n",
      "Epoch 192/199\n",
      "----------\n",
      "train Loss: 1.3630 Acc: 0.8072\n",
      "val Loss: 4.3582 Acc: 0.1313\n",
      "\n",
      "Epoch 193/199\n",
      "----------\n",
      "train Loss: 1.3456 Acc: 0.7904\n",
      "val Loss: 4.2914 Acc: 0.1343\n",
      "\n",
      "Epoch 194/199\n",
      "----------\n",
      "train Loss: 1.3689 Acc: 0.7952\n",
      "val Loss: 4.3290 Acc: 0.1224\n",
      "\n",
      "Epoch 195/199\n",
      "----------\n",
      "train Loss: 1.3974 Acc: 0.7831\n",
      "val Loss: 4.3483 Acc: 0.1433\n",
      "\n",
      "Epoch 196/199\n",
      "----------\n",
      "train Loss: 1.3046 Acc: 0.8145\n",
      "val Loss: 4.2997 Acc: 0.1343\n",
      "\n",
      "Epoch 197/199\n",
      "----------\n",
      "train Loss: 1.3544 Acc: 0.7735\n",
      "val Loss: 4.3091 Acc: 0.1224\n",
      "\n",
      "Epoch 198/199\n",
      "----------\n",
      "train Loss: 1.3202 Acc: 0.7855\n",
      "val Loss: 4.3197 Acc: 0.1313\n",
      "\n",
      "Epoch 199/199\n",
      "----------\n",
      "train Loss: 1.2581 Acc: 0.8193\n",
      "val Loss: 4.2830 Acc: 0.1373\n",
      "\n",
      "Training complete in 26m 10s\n",
      "Best val Acc: 0.149254\n"
     ]
    }
   ],
   "source": [
    "model = CNNEarDetector(model=\"resnet\", num_classes=train_data.get_num_classes(), extract_features=False)\n",
    "\n",
    "# Setup the loss fxn\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "\n",
    "# Train and evaluate\n",
    "num_epochs = 200\n",
    "hist = model.train(dataloaders_dict, criterion, num_epochs=num_epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "m-AZP6e_jiNL",
   "metadata": {
    "executionInfo": {
     "elapsed": 3284,
     "status": "ok",
     "timestamp": 1642414630712,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "m-AZP6e_jiNL"
   },
   "outputs": [],
   "source": [
    "model.save(os.getcwd(), \"trained_resnet_equalization\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "bX5Iw2RSjqDT",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "executionInfo": {
     "elapsed": 903,
     "status": "ok",
     "timestamp": 1642414634649,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "bX5Iw2RSjqDT",
    "outputId": "69d4fe50-d625-4f40-948f-13467724dae1"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "val_hist = [h.cpu().numpy() for h in hist]\n",
    "\n",
    "plt.title(\"Validation Accuracy vs. Number of Training Epochs\")\n",
    "plt.xlabel(\"Training Epochs\")\n",
    "plt.ylabel(\"Validation Accuracy\")\n",
    "plt.plot(range(1,num_epochs+1), val_hist)\n",
    "plt.savefig(os.getcwd()+\"/resnet_val_equalization.png\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cRdyFBQij1xo",
   "metadata": {
    "id": "cRdyFBQij1xo"
   },
   "source": [
    "### Edge enhancement\n",
    "\n",
    "#### Inspect the data loading"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "vTd-KYDHkHGN",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 480,
     "status": "ok",
     "timestamp": 1642368945662,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "vTd-KYDHkHGN",
    "outputId": "a689d486-7e18-4d03-8ffa-45d2d616c1aa"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(<PIL.Image.Image image mode=RGB size=53x78 at 0x7F979498F1D0>, tensor(99))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data = AWETrainSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), transforms.Compose([EdgeEnhancement()]))\n",
    "train_data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "z4wgz7DrkR-o",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 285
    },
    "executionInfo": {
     "elapsed": 715,
     "status": "ok",
     "timestamp": 1642368948733,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "z4wgz7DrkR-o",
    "outputId": "84b40356-3e23-4758-d132-4f19abe9429b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(99)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "image, label = train_data[0]\n",
    "plt.imshow(image)\n",
    "print(label)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3WCTbejykgwI",
   "metadata": {
    "id": "3WCTbejykgwI"
   },
   "source": [
    "#### Load train and val data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "vjwpiAUfkd7s",
   "metadata": {
    "executionInfo": {
     "elapsed": 281,
     "status": "ok",
     "timestamp": 1642414696572,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "vjwpiAUfkd7s"
   },
   "outputs": [],
   "source": [
    "train_data = AWETrainSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), data_transforms_edge[\"train\"])\n",
    "val_data = AWEValSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), data_transforms_edge[\"val\"])\n",
    "\n",
    "# Create training and validation dataloaders\n",
    "dataloaders_dict = {'train': torch.utils.data.DataLoader(train_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=2),\n",
    "                   'val': torch.utils.data.DataLoader(val_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=2)}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "sWggSaGslH1Q",
   "metadata": {
    "id": "sWggSaGslH1Q"
   },
   "source": [
    "#### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "fS99HHBalLED",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 1526005,
     "status": "ok",
     "timestamp": 1642416231657,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "fS99HHBalLED",
    "outputId": "5b32d029-1fba-4433-abea-9fc6d4ee0990"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training model on GPUs\n",
      "Epoch 0/199\n",
      "----------\n",
      "train Loss: 4.6479 Acc: 0.0145\n",
      "val Loss: 4.6570 Acc: 0.0090\n",
      "\n",
      "Epoch 1/199\n",
      "----------\n",
      "train Loss: 4.6349 Acc: 0.0048\n",
      "val Loss: 4.6343 Acc: 0.0030\n",
      "\n",
      "Epoch 2/199\n",
      "----------\n",
      "train Loss: 4.6166 Acc: 0.0072\n",
      "val Loss: 4.6396 Acc: 0.0000\n",
      "\n",
      "Epoch 3/199\n",
      "----------\n",
      "train Loss: 4.5920 Acc: 0.0145\n",
      "val Loss: 4.6220 Acc: 0.0119\n",
      "\n",
      "Epoch 4/199\n",
      "----------\n",
      "train Loss: 4.5792 Acc: 0.0096\n",
      "val Loss: 4.6120 Acc: 0.0179\n",
      "\n",
      "Epoch 5/199\n",
      "----------\n",
      "train Loss: 4.5647 Acc: 0.0145\n",
      "val Loss: 4.6152 Acc: 0.0090\n",
      "\n",
      "Epoch 6/199\n",
      "----------\n",
      "train Loss: 4.5314 Acc: 0.0217\n",
      "val Loss: 4.6068 Acc: 0.0149\n",
      "\n",
      "Epoch 7/199\n",
      "----------\n",
      "train Loss: 4.5285 Acc: 0.0169\n",
      "val Loss: 4.5992 Acc: 0.0119\n",
      "\n",
      "Epoch 8/199\n",
      "----------\n",
      "train Loss: 4.5228 Acc: 0.0289\n",
      "val Loss: 4.5921 Acc: 0.0090\n",
      "\n",
      "Epoch 9/199\n",
      "----------\n",
      "train Loss: 4.5142 Acc: 0.0265\n",
      "val Loss: 4.5959 Acc: 0.0119\n",
      "\n",
      "Epoch 10/199\n",
      "----------\n",
      "train Loss: 4.4829 Acc: 0.0361\n",
      "val Loss: 4.5827 Acc: 0.0090\n",
      "\n",
      "Epoch 11/199\n",
      "----------\n",
      "train Loss: 4.4845 Acc: 0.0434\n",
      "val Loss: 4.5840 Acc: 0.0209\n",
      "\n",
      "Epoch 12/199\n",
      "----------\n",
      "train Loss: 4.4618 Acc: 0.0361\n",
      "val Loss: 4.5716 Acc: 0.0209\n",
      "\n",
      "Epoch 13/199\n",
      "----------\n",
      "train Loss: 4.4417 Acc: 0.0506\n",
      "val Loss: 4.5749 Acc: 0.0209\n",
      "\n",
      "Epoch 14/199\n",
      "----------\n",
      "train Loss: 4.4437 Acc: 0.0506\n",
      "val Loss: 4.5644 Acc: 0.0209\n",
      "\n",
      "Epoch 15/199\n",
      "----------\n",
      "train Loss: 4.4269 Acc: 0.0482\n",
      "val Loss: 4.5612 Acc: 0.0179\n",
      "\n",
      "Epoch 16/199\n",
      "----------\n",
      "train Loss: 4.3991 Acc: 0.0699\n",
      "val Loss: 4.5616 Acc: 0.0119\n",
      "\n",
      "Epoch 17/199\n",
      "----------\n",
      "train Loss: 4.4048 Acc: 0.0627\n",
      "val Loss: 4.5587 Acc: 0.0149\n",
      "\n",
      "Epoch 18/199\n",
      "----------\n",
      "train Loss: 4.3785 Acc: 0.0675\n",
      "val Loss: 4.5551 Acc: 0.0119\n",
      "\n",
      "Epoch 19/199\n",
      "----------\n",
      "train Loss: 4.3481 Acc: 0.0651\n",
      "val Loss: 4.5521 Acc: 0.0119\n",
      "\n",
      "Epoch 20/199\n",
      "----------\n",
      "train Loss: 4.3624 Acc: 0.0723\n",
      "val Loss: 4.5457 Acc: 0.0209\n",
      "\n",
      "Epoch 21/199\n",
      "----------\n",
      "train Loss: 4.3151 Acc: 0.0916\n",
      "val Loss: 4.5433 Acc: 0.0328\n",
      "\n",
      "Epoch 22/199\n",
      "----------\n",
      "train Loss: 4.3067 Acc: 0.1108\n",
      "val Loss: 4.5435 Acc: 0.0209\n",
      "\n",
      "Epoch 23/199\n",
      "----------\n",
      "train Loss: 4.2785 Acc: 0.1060\n",
      "val Loss: 4.5369 Acc: 0.0328\n",
      "\n",
      "Epoch 24/199\n",
      "----------\n",
      "train Loss: 4.2833 Acc: 0.1181\n",
      "val Loss: 4.5299 Acc: 0.0388\n",
      "\n",
      "Epoch 25/199\n",
      "----------\n",
      "train Loss: 4.2417 Acc: 0.1108\n",
      "val Loss: 4.5337 Acc: 0.0328\n",
      "\n",
      "Epoch 26/199\n",
      "----------\n",
      "train Loss: 4.2072 Acc: 0.1253\n",
      "val Loss: 4.5272 Acc: 0.0418\n",
      "\n",
      "Epoch 27/199\n",
      "----------\n",
      "train Loss: 4.2024 Acc: 0.1205\n",
      "val Loss: 4.5192 Acc: 0.0478\n",
      "\n",
      "Epoch 28/199\n",
      "----------\n",
      "train Loss: 4.2013 Acc: 0.1253\n",
      "val Loss: 4.5313 Acc: 0.0507\n",
      "\n",
      "Epoch 29/199\n",
      "----------\n",
      "train Loss: 4.1751 Acc: 0.1229\n",
      "val Loss: 4.5241 Acc: 0.0418\n",
      "\n",
      "Epoch 30/199\n",
      "----------\n",
      "train Loss: 4.1917 Acc: 0.1373\n",
      "val Loss: 4.5148 Acc: 0.0478\n",
      "\n",
      "Epoch 31/199\n",
      "----------\n",
      "train Loss: 4.1563 Acc: 0.1253\n",
      "val Loss: 4.5175 Acc: 0.0537\n",
      "\n",
      "Epoch 32/199\n",
      "----------\n",
      "train Loss: 4.1251 Acc: 0.1470\n",
      "val Loss: 4.5073 Acc: 0.0507\n",
      "\n",
      "Epoch 33/199\n",
      "----------\n",
      "train Loss: 4.1019 Acc: 0.1518\n",
      "val Loss: 4.5140 Acc: 0.0418\n",
      "\n",
      "Epoch 34/199\n",
      "----------\n",
      "train Loss: 4.1265 Acc: 0.1301\n",
      "val Loss: 4.5128 Acc: 0.0418\n",
      "\n",
      "Epoch 35/199\n",
      "----------\n",
      "train Loss: 4.0755 Acc: 0.1398\n",
      "val Loss: 4.5155 Acc: 0.0418\n",
      "\n",
      "Epoch 36/199\n",
      "----------\n",
      "train Loss: 4.0688 Acc: 0.1470\n",
      "val Loss: 4.5106 Acc: 0.0418\n",
      "\n",
      "Epoch 37/199\n",
      "----------\n",
      "train Loss: 4.0653 Acc: 0.1542\n",
      "val Loss: 4.5129 Acc: 0.0597\n",
      "\n",
      "Epoch 38/199\n",
      "----------\n",
      "train Loss: 4.0311 Acc: 0.1783\n",
      "val Loss: 4.4966 Acc: 0.0567\n",
      "\n",
      "Epoch 39/199\n",
      "----------\n",
      "train Loss: 4.0070 Acc: 0.1639\n",
      "val Loss: 4.4916 Acc: 0.0627\n",
      "\n",
      "Epoch 40/199\n",
      "----------\n",
      "train Loss: 4.0118 Acc: 0.1398\n",
      "val Loss: 4.4980 Acc: 0.0567\n",
      "\n",
      "Epoch 41/199\n",
      "----------\n",
      "train Loss: 3.9827 Acc: 0.1663\n",
      "val Loss: 4.4861 Acc: 0.0657\n",
      "\n",
      "Epoch 42/199\n",
      "----------\n",
      "train Loss: 3.9363 Acc: 0.2096\n",
      "val Loss: 4.4978 Acc: 0.0537\n",
      "\n",
      "Epoch 43/199\n",
      "----------\n",
      "train Loss: 3.9459 Acc: 0.1639\n",
      "val Loss: 4.4993 Acc: 0.0597\n",
      "\n",
      "Epoch 44/199\n",
      "----------\n",
      "train Loss: 3.9496 Acc: 0.1831\n",
      "val Loss: 4.4875 Acc: 0.0537\n",
      "\n",
      "Epoch 45/199\n",
      "----------\n",
      "train Loss: 3.8794 Acc: 0.1735\n",
      "val Loss: 4.4912 Acc: 0.0657\n",
      "\n",
      "Epoch 46/199\n",
      "----------\n",
      "train Loss: 3.8859 Acc: 0.1904\n",
      "val Loss: 4.4956 Acc: 0.0716\n",
      "\n",
      "Epoch 47/199\n",
      "----------\n",
      "train Loss: 3.8544 Acc: 0.1928\n",
      "val Loss: 4.4909 Acc: 0.0687\n",
      "\n",
      "Epoch 48/199\n",
      "----------\n",
      "train Loss: 3.8290 Acc: 0.2096\n",
      "val Loss: 4.4923 Acc: 0.0657\n",
      "\n",
      "Epoch 49/199\n",
      "----------\n",
      "train Loss: 3.7953 Acc: 0.2048\n",
      "val Loss: 4.4930 Acc: 0.0746\n",
      "\n",
      "Epoch 50/199\n",
      "----------\n",
      "train Loss: 3.7923 Acc: 0.2000\n",
      "val Loss: 4.4835 Acc: 0.0776\n",
      "\n",
      "Epoch 51/199\n",
      "----------\n",
      "train Loss: 3.7744 Acc: 0.2169\n",
      "val Loss: 4.4782 Acc: 0.0746\n",
      "\n",
      "Epoch 52/199\n",
      "----------\n",
      "train Loss: 3.7668 Acc: 0.2048\n",
      "val Loss: 4.4764 Acc: 0.0776\n",
      "\n",
      "Epoch 53/199\n",
      "----------\n",
      "train Loss: 3.7778 Acc: 0.2145\n",
      "val Loss: 4.4975 Acc: 0.0627\n",
      "\n",
      "Epoch 54/199\n",
      "----------\n",
      "train Loss: 3.7384 Acc: 0.2241\n",
      "val Loss: 4.4972 Acc: 0.0627\n",
      "\n",
      "Epoch 55/199\n",
      "----------\n",
      "train Loss: 3.7297 Acc: 0.2120\n",
      "val Loss: 4.4917 Acc: 0.0806\n",
      "\n",
      "Epoch 56/199\n",
      "----------\n",
      "train Loss: 3.6778 Acc: 0.2096\n",
      "val Loss: 4.4770 Acc: 0.0776\n",
      "\n",
      "Epoch 57/199\n",
      "----------\n",
      "train Loss: 3.6730 Acc: 0.2361\n",
      "val Loss: 4.4767 Acc: 0.0776\n",
      "\n",
      "Epoch 58/199\n",
      "----------\n",
      "train Loss: 3.6714 Acc: 0.2193\n",
      "val Loss: 4.4891 Acc: 0.0657\n",
      "\n",
      "Epoch 59/199\n",
      "----------\n",
      "train Loss: 3.6177 Acc: 0.2361\n",
      "val Loss: 4.4656 Acc: 0.0746\n",
      "\n",
      "Epoch 60/199\n",
      "----------\n",
      "train Loss: 3.5670 Acc: 0.2410\n",
      "val Loss: 4.4928 Acc: 0.0716\n",
      "\n",
      "Epoch 61/199\n",
      "----------\n",
      "train Loss: 3.6175 Acc: 0.2120\n",
      "val Loss: 4.4974 Acc: 0.0597\n",
      "\n",
      "Epoch 62/199\n",
      "----------\n",
      "train Loss: 3.6168 Acc: 0.2265\n",
      "val Loss: 4.4879 Acc: 0.0776\n",
      "\n",
      "Epoch 63/199\n",
      "----------\n",
      "train Loss: 3.5596 Acc: 0.2651\n",
      "val Loss: 4.4818 Acc: 0.0687\n",
      "\n",
      "Epoch 64/199\n",
      "----------\n",
      "train Loss: 3.5438 Acc: 0.2482\n",
      "val Loss: 4.4906 Acc: 0.0866\n",
      "\n",
      "Epoch 65/199\n",
      "----------\n",
      "train Loss: 3.5244 Acc: 0.2651\n",
      "val Loss: 4.4740 Acc: 0.0716\n",
      "\n",
      "Epoch 66/199\n",
      "----------\n",
      "train Loss: 3.5044 Acc: 0.2627\n",
      "val Loss: 4.4912 Acc: 0.0716\n",
      "\n",
      "Epoch 67/199\n",
      "----------\n",
      "train Loss: 3.4910 Acc: 0.2819\n",
      "val Loss: 4.4836 Acc: 0.0627\n",
      "\n",
      "Epoch 68/199\n",
      "----------\n",
      "train Loss: 3.4262 Acc: 0.2747\n",
      "val Loss: 4.4867 Acc: 0.0716\n",
      "\n",
      "Epoch 69/199\n",
      "----------\n",
      "train Loss: 3.4851 Acc: 0.2313\n",
      "val Loss: 4.4804 Acc: 0.0746\n",
      "\n",
      "Epoch 70/199\n",
      "----------\n",
      "train Loss: 3.3838 Acc: 0.2723\n",
      "val Loss: 4.4908 Acc: 0.0716\n",
      "\n",
      "Epoch 71/199\n",
      "----------\n",
      "train Loss: 3.3794 Acc: 0.2916\n",
      "val Loss: 4.4859 Acc: 0.0687\n",
      "\n",
      "Epoch 72/199\n",
      "----------\n",
      "train Loss: 3.4285 Acc: 0.2482\n",
      "val Loss: 4.5149 Acc: 0.0806\n",
      "\n",
      "Epoch 73/199\n",
      "----------\n",
      "train Loss: 3.3588 Acc: 0.2651\n",
      "val Loss: 4.4853 Acc: 0.0716\n",
      "\n",
      "Epoch 74/199\n",
      "----------\n",
      "train Loss: 3.3540 Acc: 0.2578\n",
      "val Loss: 4.4929 Acc: 0.0716\n",
      "\n",
      "Epoch 75/199\n",
      "----------\n",
      "train Loss: 3.3383 Acc: 0.2819\n",
      "val Loss: 4.4911 Acc: 0.0687\n",
      "\n",
      "Epoch 76/199\n",
      "----------\n",
      "train Loss: 3.3404 Acc: 0.2627\n",
      "val Loss: 4.4847 Acc: 0.0746\n",
      "\n",
      "Epoch 77/199\n",
      "----------\n",
      "train Loss: 3.2934 Acc: 0.2916\n",
      "val Loss: 4.4984 Acc: 0.0746\n",
      "\n",
      "Epoch 78/199\n",
      "----------\n",
      "train Loss: 3.2348 Acc: 0.3229\n",
      "val Loss: 4.5092 Acc: 0.0746\n",
      "\n",
      "Epoch 79/199\n",
      "----------\n",
      "train Loss: 3.3188 Acc: 0.2819\n",
      "val Loss: 4.5080 Acc: 0.0776\n",
      "\n",
      "Epoch 80/199\n",
      "----------\n",
      "train Loss: 3.1840 Acc: 0.3157\n",
      "val Loss: 4.4943 Acc: 0.0716\n",
      "\n",
      "Epoch 81/199\n",
      "----------\n",
      "train Loss: 3.2058 Acc: 0.2964\n",
      "val Loss: 4.5080 Acc: 0.0776\n",
      "\n",
      "Epoch 82/199\n",
      "----------\n",
      "train Loss: 3.1981 Acc: 0.3446\n",
      "val Loss: 4.4973 Acc: 0.0657\n",
      "\n",
      "Epoch 83/199\n",
      "----------\n",
      "train Loss: 3.1697 Acc: 0.3398\n",
      "val Loss: 4.4947 Acc: 0.0716\n",
      "\n",
      "Epoch 84/199\n",
      "----------\n",
      "train Loss: 3.1635 Acc: 0.3108\n",
      "val Loss: 4.4937 Acc: 0.0746\n",
      "\n",
      "Epoch 85/199\n",
      "----------\n",
      "train Loss: 3.1214 Acc: 0.3373\n",
      "val Loss: 4.5115 Acc: 0.0776\n",
      "\n",
      "Epoch 86/199\n",
      "----------\n",
      "train Loss: 3.1020 Acc: 0.3253\n",
      "val Loss: 4.5124 Acc: 0.0746\n",
      "\n",
      "Epoch 87/199\n",
      "----------\n",
      "train Loss: 3.1087 Acc: 0.3301\n",
      "val Loss: 4.5144 Acc: 0.0716\n",
      "\n",
      "Epoch 88/199\n",
      "----------\n",
      "train Loss: 3.0821 Acc: 0.3494\n",
      "val Loss: 4.5228 Acc: 0.0746\n",
      "\n",
      "Epoch 89/199\n",
      "----------\n",
      "train Loss: 3.1072 Acc: 0.3566\n",
      "val Loss: 4.5143 Acc: 0.0746\n",
      "\n",
      "Epoch 90/199\n",
      "----------\n",
      "train Loss: 3.0978 Acc: 0.3253\n",
      "val Loss: 4.5242 Acc: 0.0806\n",
      "\n",
      "Epoch 91/199\n",
      "----------\n",
      "train Loss: 3.0315 Acc: 0.3639\n",
      "val Loss: 4.5253 Acc: 0.0716\n",
      "\n",
      "Epoch 92/199\n",
      "----------\n",
      "train Loss: 3.0534 Acc: 0.3325\n",
      "val Loss: 4.5391 Acc: 0.0866\n",
      "\n",
      "Epoch 93/199\n",
      "----------\n",
      "train Loss: 3.0426 Acc: 0.3373\n",
      "val Loss: 4.5300 Acc: 0.0836\n",
      "\n",
      "Epoch 94/199\n",
      "----------\n",
      "train Loss: 3.0372 Acc: 0.3542\n",
      "val Loss: 4.5419 Acc: 0.0746\n",
      "\n",
      "Epoch 95/199\n",
      "----------\n",
      "train Loss: 2.9201 Acc: 0.4048\n",
      "val Loss: 4.5508 Acc: 0.0776\n",
      "\n",
      "Epoch 96/199\n",
      "----------\n",
      "train Loss: 2.9910 Acc: 0.3470\n",
      "val Loss: 4.5483 Acc: 0.0836\n",
      "\n",
      "Epoch 97/199\n",
      "----------\n",
      "train Loss: 2.9565 Acc: 0.3783\n",
      "val Loss: 4.5452 Acc: 0.0896\n",
      "\n",
      "Epoch 98/199\n",
      "----------\n",
      "train Loss: 2.9176 Acc: 0.3904\n",
      "val Loss: 4.5356 Acc: 0.0866\n",
      "\n",
      "Epoch 99/199\n",
      "----------\n",
      "train Loss: 2.8724 Acc: 0.3759\n",
      "val Loss: 4.5566 Acc: 0.0836\n",
      "\n",
      "Epoch 100/199\n",
      "----------\n",
      "train Loss: 2.8773 Acc: 0.3855\n",
      "val Loss: 4.5749 Acc: 0.0746\n",
      "\n",
      "Epoch 101/199\n",
      "----------\n",
      "train Loss: 2.8737 Acc: 0.3855\n",
      "val Loss: 4.5451 Acc: 0.0896\n",
      "\n",
      "Epoch 102/199\n",
      "----------\n",
      "train Loss: 2.9491 Acc: 0.3711\n",
      "val Loss: 4.5865 Acc: 0.0866\n",
      "\n",
      "Epoch 103/199\n",
      "----------\n",
      "train Loss: 2.8320 Acc: 0.4169\n",
      "val Loss: 4.5733 Acc: 0.0806\n",
      "\n",
      "Epoch 104/199\n",
      "----------\n",
      "train Loss: 2.8402 Acc: 0.4265\n",
      "val Loss: 4.5657 Acc: 0.0776\n",
      "\n",
      "Epoch 105/199\n",
      "----------\n",
      "train Loss: 2.7868 Acc: 0.4289\n",
      "val Loss: 4.5712 Acc: 0.0896\n",
      "\n",
      "Epoch 106/199\n",
      "----------\n",
      "train Loss: 2.8428 Acc: 0.3880\n",
      "val Loss: 4.5672 Acc: 0.0836\n",
      "\n",
      "Epoch 107/199\n",
      "----------\n",
      "train Loss: 2.7892 Acc: 0.4024\n",
      "val Loss: 4.5529 Acc: 0.0866\n",
      "\n",
      "Epoch 108/199\n",
      "----------\n",
      "train Loss: 2.7498 Acc: 0.4072\n",
      "val Loss: 4.5646 Acc: 0.0806\n",
      "\n",
      "Epoch 109/199\n",
      "----------\n",
      "train Loss: 2.8171 Acc: 0.4072\n",
      "val Loss: 4.5971 Acc: 0.0746\n",
      "\n",
      "Epoch 110/199\n",
      "----------\n",
      "train Loss: 2.7373 Acc: 0.4313\n",
      "val Loss: 4.5339 Acc: 0.0896\n",
      "\n",
      "Epoch 111/199\n",
      "----------\n",
      "train Loss: 2.6971 Acc: 0.4217\n",
      "val Loss: 4.5933 Acc: 0.0866\n",
      "\n",
      "Epoch 112/199\n",
      "----------\n",
      "train Loss: 2.7018 Acc: 0.4289\n",
      "val Loss: 4.5918 Acc: 0.0776\n",
      "\n",
      "Epoch 113/199\n",
      "----------\n",
      "train Loss: 2.6805 Acc: 0.4361\n",
      "val Loss: 4.5931 Acc: 0.0746\n",
      "\n",
      "Epoch 114/199\n",
      "----------\n",
      "train Loss: 2.7090 Acc: 0.4120\n",
      "val Loss: 4.5856 Acc: 0.0806\n",
      "\n",
      "Epoch 115/199\n",
      "----------\n",
      "train Loss: 2.6895 Acc: 0.4096\n",
      "val Loss: 4.6206 Acc: 0.0746\n",
      "\n",
      "Epoch 116/199\n",
      "----------\n",
      "train Loss: 2.6059 Acc: 0.4410\n",
      "val Loss: 4.6078 Acc: 0.0776\n",
      "\n",
      "Epoch 117/199\n",
      "----------\n",
      "train Loss: 2.6231 Acc: 0.4434\n",
      "val Loss: 4.6321 Acc: 0.0896\n",
      "\n",
      "Epoch 118/199\n",
      "----------\n",
      "train Loss: 2.6013 Acc: 0.4313\n",
      "val Loss: 4.5798 Acc: 0.0776\n",
      "\n",
      "Epoch 119/199\n",
      "----------\n",
      "train Loss: 2.5862 Acc: 0.4193\n",
      "val Loss: 4.5802 Acc: 0.0866\n",
      "\n",
      "Epoch 120/199\n",
      "----------\n",
      "train Loss: 2.5649 Acc: 0.4771\n",
      "val Loss: 4.5985 Acc: 0.0746\n",
      "\n",
      "Epoch 121/199\n",
      "----------\n",
      "train Loss: 2.6227 Acc: 0.4554\n",
      "val Loss: 4.6546 Acc: 0.0806\n",
      "\n",
      "Epoch 122/199\n",
      "----------\n",
      "train Loss: 2.5524 Acc: 0.4506\n",
      "val Loss: 4.6324 Acc: 0.0776\n",
      "\n",
      "Epoch 123/199\n",
      "----------\n",
      "train Loss: 2.4887 Acc: 0.4843\n",
      "val Loss: 4.6325 Acc: 0.0866\n",
      "\n",
      "Epoch 124/199\n",
      "----------\n",
      "train Loss: 2.5584 Acc: 0.4819\n",
      "val Loss: 4.6641 Acc: 0.0836\n",
      "\n",
      "Epoch 125/199\n",
      "----------\n",
      "train Loss: 2.5213 Acc: 0.4530\n",
      "val Loss: 4.6808 Acc: 0.0806\n",
      "\n",
      "Epoch 126/199\n",
      "----------\n",
      "train Loss: 2.5436 Acc: 0.4506\n",
      "val Loss: 4.6735 Acc: 0.0776\n",
      "\n",
      "Epoch 127/199\n",
      "----------\n",
      "train Loss: 2.5182 Acc: 0.4916\n",
      "val Loss: 4.6508 Acc: 0.0896\n",
      "\n",
      "Epoch 128/199\n",
      "----------\n",
      "train Loss: 2.4659 Acc: 0.4964\n",
      "val Loss: 4.6619 Acc: 0.0836\n",
      "\n",
      "Epoch 129/199\n",
      "----------\n",
      "train Loss: 2.3878 Acc: 0.5446\n",
      "val Loss: 4.6680 Acc: 0.0925\n",
      "\n",
      "Epoch 130/199\n",
      "----------\n",
      "train Loss: 2.3882 Acc: 0.4940\n",
      "val Loss: 4.6262 Acc: 0.0836\n",
      "\n",
      "Epoch 131/199\n",
      "----------\n",
      "train Loss: 2.3297 Acc: 0.5229\n",
      "val Loss: 4.7137 Acc: 0.0806\n",
      "\n",
      "Epoch 132/199\n",
      "----------\n",
      "train Loss: 2.3984 Acc: 0.4843\n",
      "val Loss: 4.6646 Acc: 0.0836\n",
      "\n",
      "Epoch 133/199\n",
      "----------\n",
      "train Loss: 2.4475 Acc: 0.4843\n",
      "val Loss: 4.6688 Acc: 0.0925\n",
      "\n",
      "Epoch 134/199\n",
      "----------\n",
      "train Loss: 2.3417 Acc: 0.5181\n",
      "val Loss: 4.6868 Acc: 0.0866\n",
      "\n",
      "Epoch 135/199\n",
      "----------\n",
      "train Loss: 2.4283 Acc: 0.5205\n",
      "val Loss: 4.6703 Acc: 0.0776\n",
      "\n",
      "Epoch 136/199\n",
      "----------\n",
      "train Loss: 2.3317 Acc: 0.5349\n",
      "val Loss: 4.6335 Acc: 0.0925\n",
      "\n",
      "Epoch 137/199\n",
      "----------\n",
      "train Loss: 2.3574 Acc: 0.4964\n",
      "val Loss: 4.7150 Acc: 0.0925\n",
      "\n",
      "Epoch 138/199\n",
      "----------\n",
      "train Loss: 2.4163 Acc: 0.4916\n",
      "val Loss: 4.7222 Acc: 0.0866\n",
      "\n",
      "Epoch 139/199\n",
      "----------\n",
      "train Loss: 2.2809 Acc: 0.5084\n",
      "val Loss: 4.6901 Acc: 0.0955\n",
      "\n",
      "Epoch 140/199\n",
      "----------\n",
      "train Loss: 2.2593 Acc: 0.5639\n",
      "val Loss: 4.6485 Acc: 0.1075\n",
      "\n",
      "Epoch 141/199\n",
      "----------\n",
      "train Loss: 2.1873 Acc: 0.5880\n",
      "val Loss: 4.7061 Acc: 0.0925\n",
      "\n",
      "Epoch 142/199\n",
      "----------\n",
      "train Loss: 2.1995 Acc: 0.5518\n",
      "val Loss: 4.7080 Acc: 0.0896\n",
      "\n",
      "Epoch 143/199\n",
      "----------\n",
      "train Loss: 2.1741 Acc: 0.5663\n",
      "val Loss: 4.6872 Acc: 0.0866\n",
      "\n",
      "Epoch 144/199\n",
      "----------\n",
      "train Loss: 2.2247 Acc: 0.5590\n",
      "val Loss: 4.7275 Acc: 0.0955\n",
      "\n",
      "Epoch 145/199\n",
      "----------\n",
      "train Loss: 2.2007 Acc: 0.5639\n",
      "val Loss: 4.7318 Acc: 0.0925\n",
      "\n",
      "Epoch 146/199\n",
      "----------\n",
      "train Loss: 2.1712 Acc: 0.5735\n",
      "val Loss: 4.6921 Acc: 0.0925\n",
      "\n",
      "Epoch 147/199\n",
      "----------\n",
      "train Loss: 2.1844 Acc: 0.5373\n",
      "val Loss: 4.7097 Acc: 0.0985\n",
      "\n",
      "Epoch 148/199\n",
      "----------\n",
      "train Loss: 2.1521 Acc: 0.5663\n",
      "val Loss: 4.7403 Acc: 0.0925\n",
      "\n",
      "Epoch 149/199\n",
      "----------\n",
      "train Loss: 2.0747 Acc: 0.5928\n",
      "val Loss: 4.7109 Acc: 0.1075\n",
      "\n",
      "Epoch 150/199\n",
      "----------\n",
      "train Loss: 2.1141 Acc: 0.5952\n",
      "val Loss: 4.6791 Acc: 0.0985\n",
      "\n",
      "Epoch 151/199\n",
      "----------\n",
      "train Loss: 2.1027 Acc: 0.5663\n",
      "val Loss: 4.6881 Acc: 0.0985\n",
      "\n",
      "Epoch 152/199\n",
      "----------\n",
      "train Loss: 2.0906 Acc: 0.5904\n",
      "val Loss: 4.7426 Acc: 0.0985\n",
      "\n",
      "Epoch 153/199\n",
      "----------\n",
      "train Loss: 2.0987 Acc: 0.6000\n",
      "val Loss: 4.7417 Acc: 0.1015\n",
      "\n",
      "Epoch 154/199\n",
      "----------\n",
      "train Loss: 2.0905 Acc: 0.5663\n",
      "val Loss: 4.7107 Acc: 0.0925\n",
      "\n",
      "Epoch 155/199\n",
      "----------\n",
      "train Loss: 2.0559 Acc: 0.5783\n",
      "val Loss: 4.7725 Acc: 0.0985\n",
      "\n",
      "Epoch 156/199\n",
      "----------\n",
      "train Loss: 2.0309 Acc: 0.6000\n",
      "val Loss: 4.7388 Acc: 0.0955\n",
      "\n",
      "Epoch 157/199\n",
      "----------\n",
      "train Loss: 2.0531 Acc: 0.5735\n",
      "val Loss: 4.7422 Acc: 0.0955\n",
      "\n",
      "Epoch 158/199\n",
      "----------\n",
      "train Loss: 1.9533 Acc: 0.6217\n",
      "val Loss: 4.7413 Acc: 0.1015\n",
      "\n",
      "Epoch 159/199\n",
      "----------\n",
      "train Loss: 1.9230 Acc: 0.6241\n",
      "val Loss: 4.7350 Acc: 0.1045\n",
      "\n",
      "Epoch 160/199\n",
      "----------\n",
      "train Loss: 1.9608 Acc: 0.6241\n",
      "val Loss: 4.7806 Acc: 0.0955\n",
      "\n",
      "Epoch 161/199\n",
      "----------\n",
      "train Loss: 1.9946 Acc: 0.5976\n",
      "val Loss: 4.7708 Acc: 0.0985\n",
      "\n",
      "Epoch 162/199\n",
      "----------\n",
      "train Loss: 2.0191 Acc: 0.5952\n",
      "val Loss: 4.7817 Acc: 0.1075\n",
      "\n",
      "Epoch 163/199\n",
      "----------\n",
      "train Loss: 1.9241 Acc: 0.6361\n",
      "val Loss: 4.7917 Acc: 0.0955\n",
      "\n",
      "Epoch 164/199\n",
      "----------\n",
      "train Loss: 1.9235 Acc: 0.6145\n",
      "val Loss: 4.7469 Acc: 0.1134\n",
      "\n",
      "Epoch 165/199\n",
      "----------\n",
      "train Loss: 1.8912 Acc: 0.6434\n",
      "val Loss: 4.7838 Acc: 0.0866\n",
      "\n",
      "Epoch 166/199\n",
      "----------\n",
      "train Loss: 1.8261 Acc: 0.6651\n",
      "val Loss: 4.7655 Acc: 0.0985\n",
      "\n",
      "Epoch 167/199\n",
      "----------\n",
      "train Loss: 1.8893 Acc: 0.6265\n",
      "val Loss: 4.7278 Acc: 0.1075\n",
      "\n",
      "Epoch 168/199\n",
      "----------\n",
      "train Loss: 1.8382 Acc: 0.6386\n",
      "val Loss: 4.7270 Acc: 0.0985\n",
      "\n",
      "Epoch 169/199\n",
      "----------\n",
      "train Loss: 1.8593 Acc: 0.6771\n",
      "val Loss: 4.7256 Acc: 0.0985\n",
      "\n",
      "Epoch 170/199\n",
      "----------\n",
      "train Loss: 1.8990 Acc: 0.6386\n",
      "val Loss: 4.7967 Acc: 0.0985\n",
      "\n",
      "Epoch 171/199\n",
      "----------\n",
      "train Loss: 1.8542 Acc: 0.6506\n",
      "val Loss: 4.8180 Acc: 0.1045\n",
      "\n",
      "Epoch 172/199\n",
      "----------\n",
      "train Loss: 1.7951 Acc: 0.6795\n",
      "val Loss: 4.7827 Acc: 0.1015\n",
      "\n",
      "Epoch 173/199\n",
      "----------\n",
      "train Loss: 1.7336 Acc: 0.6699\n",
      "val Loss: 4.8006 Acc: 0.1045\n",
      "\n",
      "Epoch 174/199\n",
      "----------\n",
      "train Loss: 1.8514 Acc: 0.6410\n",
      "val Loss: 4.7966 Acc: 0.0985\n",
      "\n",
      "Epoch 175/199\n",
      "----------\n",
      "train Loss: 1.8124 Acc: 0.6386\n",
      "val Loss: 4.7977 Acc: 0.0985\n",
      "\n",
      "Epoch 176/199\n",
      "----------\n",
      "train Loss: 1.7788 Acc: 0.6675\n",
      "val Loss: 4.7722 Acc: 0.1104\n",
      "\n",
      "Epoch 177/199\n",
      "----------\n",
      "train Loss: 1.7665 Acc: 0.6627\n",
      "val Loss: 4.7907 Acc: 0.1015\n",
      "\n",
      "Epoch 178/199\n",
      "----------\n",
      "train Loss: 1.7367 Acc: 0.6651\n",
      "val Loss: 4.8261 Acc: 0.0836\n",
      "\n",
      "Epoch 179/199\n",
      "----------\n",
      "train Loss: 1.7143 Acc: 0.6795\n",
      "val Loss: 4.8319 Acc: 0.1015\n",
      "\n",
      "Epoch 180/199\n",
      "----------\n",
      "train Loss: 1.7464 Acc: 0.6482\n",
      "val Loss: 4.8306 Acc: 0.0925\n",
      "\n",
      "Epoch 181/199\n",
      "----------\n",
      "train Loss: 1.7403 Acc: 0.6458\n",
      "val Loss: 4.8662 Acc: 0.1015\n",
      "\n",
      "Epoch 182/199\n",
      "----------\n",
      "train Loss: 1.7172 Acc: 0.6747\n",
      "val Loss: 4.7871 Acc: 0.1104\n",
      "\n",
      "Epoch 183/199\n",
      "----------\n",
      "train Loss: 1.7336 Acc: 0.6723\n",
      "val Loss: 4.8174 Acc: 0.1075\n",
      "\n",
      "Epoch 184/199\n",
      "----------\n",
      "train Loss: 1.6468 Acc: 0.6988\n",
      "val Loss: 4.8195 Acc: 0.0925\n",
      "\n",
      "Epoch 185/199\n",
      "----------\n",
      "train Loss: 1.6237 Acc: 0.7036\n",
      "val Loss: 4.8251 Acc: 0.1075\n",
      "\n",
      "Epoch 186/199\n",
      "----------\n",
      "train Loss: 1.6224 Acc: 0.6940\n",
      "val Loss: 4.8471 Acc: 0.1015\n",
      "\n",
      "Epoch 187/199\n",
      "----------\n",
      "train Loss: 1.6716 Acc: 0.6843\n",
      "val Loss: 4.8320 Acc: 0.0985\n",
      "\n",
      "Epoch 188/199\n",
      "----------\n",
      "train Loss: 1.5699 Acc: 0.7253\n",
      "val Loss: 4.8244 Acc: 0.1075\n",
      "\n",
      "Epoch 189/199\n",
      "----------\n",
      "train Loss: 1.6643 Acc: 0.6795\n",
      "val Loss: 4.8669 Acc: 0.0955\n",
      "\n",
      "Epoch 190/199\n",
      "----------\n",
      "train Loss: 1.5320 Acc: 0.7301\n",
      "val Loss: 4.8522 Acc: 0.0955\n",
      "\n",
      "Epoch 191/199\n",
      "----------\n",
      "train Loss: 1.5454 Acc: 0.7325\n",
      "val Loss: 4.8224 Acc: 0.1134\n",
      "\n",
      "Epoch 192/199\n",
      "----------\n",
      "train Loss: 1.5033 Acc: 0.7349\n",
      "val Loss: 4.8595 Acc: 0.1104\n",
      "\n",
      "Epoch 193/199\n",
      "----------\n",
      "train Loss: 1.6002 Acc: 0.7133\n",
      "val Loss: 4.8849 Acc: 0.1015\n",
      "\n",
      "Epoch 194/199\n",
      "----------\n",
      "train Loss: 1.6288 Acc: 0.6988\n",
      "val Loss: 4.8462 Acc: 0.1075\n",
      "\n",
      "Epoch 195/199\n",
      "----------\n",
      "train Loss: 1.5100 Acc: 0.7422\n",
      "val Loss: 4.8540 Acc: 0.1045\n",
      "\n",
      "Epoch 196/199\n",
      "----------\n",
      "train Loss: 1.5655 Acc: 0.7084\n",
      "val Loss: 4.8368 Acc: 0.0955\n",
      "\n",
      "Epoch 197/199\n",
      "----------\n",
      "train Loss: 1.5516 Acc: 0.7205\n",
      "val Loss: 4.8601 Acc: 0.0985\n",
      "\n",
      "Epoch 198/199\n",
      "----------\n",
      "train Loss: 1.5259 Acc: 0.7349\n",
      "val Loss: 4.8345 Acc: 0.1224\n",
      "\n",
      "Epoch 199/199\n",
      "----------\n",
      "train Loss: 1.4068 Acc: 0.7590\n",
      "val Loss: 4.8306 Acc: 0.1015\n",
      "\n",
      "Training complete in 26m 1s\n",
      "Best val Acc: 0.122388\n"
     ]
    }
   ],
   "source": [
    "model = CNNEarDetector(model=\"resnet\", num_classes=train_data.get_num_classes(), extract_features=False)\n",
    "\n",
    "# Setup the loss fxn\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "\n",
    "# Train and evaluate\n",
    "num_epochs = 200\n",
    "hist = model.train(dataloaders_dict, criterion, num_epochs=num_epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "rZONhWtUlT-S",
   "metadata": {
    "executionInfo": {
     "elapsed": 4369,
     "status": "ok",
     "timestamp": 1642416516411,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "rZONhWtUlT-S"
   },
   "outputs": [],
   "source": [
    "model.save(os.getcwd(), \"trained_resnet_edge\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ZzTrOYp5lV0K",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "executionInfo": {
     "elapsed": 613,
     "status": "ok",
     "timestamp": 1642416517018,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "ZzTrOYp5lV0K",
    "outputId": "a55ab803-979c-4df2-97e4-9d8e298f48e8"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "val_hist = [h.cpu().numpy() for h in hist]\n",
    "\n",
    "plt.title(\"Validation Accuracy vs. Number of Training Epochs\")\n",
    "plt.xlabel(\"Training Epochs\")\n",
    "plt.ylabel(\"Validation Accuracy\")\n",
    "plt.plot(range(1,num_epochs+1), val_hist)\n",
    "plt.savefig(os.getcwd()+\"/resnet_val_edge.png\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "Jfbjztg8lYgW",
   "metadata": {
    "id": "Jfbjztg8lYgW"
   },
   "source": [
    "### Image sharpening\n",
    "\n",
    "#### Inspect the data loading"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "hf8yInv2ldrX",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 3428,
     "status": "ok",
     "timestamp": 1642406124469,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "hf8yInv2ldrX",
    "outputId": "3674709f-ce51-468d-a86a-b6677a303bda"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(<PIL.Image.Image image mode=RGB size=53x78 at 0x7F32A5839A10>, tensor(99))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data = AWETrainSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), transforms.Compose([ImageSharpening()]))\n",
    "train_data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "VYzweNyXoCn7",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 285
    },
    "executionInfo": {
     "elapsed": 1003,
     "status": "ok",
     "timestamp": 1642406126966,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "VYzweNyXoCn7",
    "outputId": "0d3f2d94-0945-4994-dd07-3280e5e60e83"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(99)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAALUAAAD7CAYAAAAsAdW1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9W7Btx3We943R3XOuy76eO24ESACkSJOmTTqkaF1omZIsyypKVjmyJaesxHGsB6dSeYtfUqm8+SEPTlVSKelBsVSOZclWyaYlWvcrKYkUKZEiDBIUwRsAEsA5B/vss/dea83Zl5GH7rn2AUWAtCCDKhX61K59OWvNNefs0aP/8Y9/jClmxivjlfHnaejX+gReGa+MP+3xilG/Mv7cjVeM+pXx5268YtSvjD934xWjfmX8uRuvGPUr48/deElGLSLfISKPicinROSf/mmd1CvjlfFShvxJeWoRccAngW8DngR+D/h+M3v0T+/0XhmvjP/84V/Ce98GfMrMPg0gIv8a+G7gBY162Xnbn7ePfMHFJIiAIAhgQLECWHuL3fH2Lz3G9N7pV9n+LCLTSxAEwyi5UEoB5pjdjdkecIzIU8AaVUWdIgiqgopggFnBDMyMUgpWDNueH6iAE54/RJB2PiKCqtZzaudTr8nqNRnPO975sOkS/viwL/2l3ov6ffq89tkinH9qHYlCxsCsnod96YHleec7vbfeh/o6s+nr/L0i7TqReq+s3qvp5+2kTPdI5Y5zv/MzpvfZ9n3PrrhhZpe/9Fa8FKO+B3jijt+fBN7+Ym/Yn3t+6BsfgFKwlLc3bzpJVUVUUFE653Gi5JLZpA2pJKwYOeetMeWc25GbwQh0TnGuTqC6NqGiOOcREZyrBlpK4fTkhPV6TclvZhz/d0r5FkR+Duf+V0T+E/PFnJ2dJd455rOOWR8wMzbDQIyJGBOr1RkxJlIuxHZuSw+7gWo8ovX8VPHB45wjeM9iPsd7j6rivAOgpExJCbN6ndPEl3Z/xKzhRdter9RfodAMsmBWEFWCq8cPITCbz/HOEYJnFjqcKE6k3mMKz9mKkzJQSiFtEjlVR0IpAPW8Q0BUcDicODBIKVPavIxjJLV5zSlVAws9oZ8BsNkMDMNILpmzYc2YYps/Vx2Hd/iur46kfQGMYyLGTC6Z9bAhtmP/8w+Wz305O3spRv1VDRH5x8A/Btif+WbI9sd8LFTvhEGhGqyJkUsi50Qp+dxnNM9zx2ew9dJ6buAyecfmFeuXoipgdsffbiH6G4hdR/UjqJ4gqjjn8N7jvcN5j3Mes4JzbusxvA+A4LzhioEJgQcQey3QYyYYgtgZRR8FewrBiKPDcqmLONXdoOSMtYVhZlAMpC4OEWmLOYMVVHTr1TCjbmY9Ob+WnO9HULIqihC9p6S+LagnSOExRDY4qlEXjLUfGTVVo46Zkko7bjVq7z1I9bymVqMxg1IyOZ87mVLqYkw5191GFNN6fTlnitWFWufB1Tk1ARPMIOWEFIEs0OY4xtyObXw1cPmlGPVTwH13/H5v+9vzhpn9CPAjAHfvzyznjLRd9ku30WkyBYhW6o0omZgjxUoz5urVVLXeaGA6kgAqhjRPNhmDim63e6eKU0WM7d9Ev4jqj4F1qG5w/jaqnhACXdfhvaMLgdAFrBRyqfBDROhnBZcciGIiQEDHdyKb/xbKIWbViQqfI9n/iepnsKRItnoeWhcaQvWK1jwj9bxFBNc8eqEwpkjJGacOdQ6QO7b9A+LwN0jj38bMYSVjxXDOsek6VMH79xD8HyLyTDVqU4rCsBTGGVgx0pApqeGKZtQpeHIJqCrmOnDVuHIqW0+dUqrfcyHmuuOkYrg8zWUh5wojRB1OqiGXfA5J8jgthuoMDLYwsZpNeR40+XLjpRj17wEPi8irqcb894Af+GreaNh22/xSKDj58GJ1u51Wdilla9DnHla2WNm2i6Rt1Q1PvsAJ1NdsYcuIyBcRyQ0Dhi0WrIaviCoqQpFpIQhqdbE4V41a1AEdpH1E7ga7cH5dssFstr2W6nENKYJoO6EWO9TFqdTloNsFahMmZ1rAgiIVeQhYcQgHYHeDOUpOWM5gjiQBVcPKHpQKxbIJzgRTSJ0jB20QTynZzgGzGaJCKRXnFCnVW98ZVzToU9qCLyVQDBAPeESglIRZaoifupjNQKY4QilFG5qq1wVUD11aXCUJIb/g1MJLMGozSyLyPwK/ADjgR83sP32l94kIYsLkv6abVr/dETzkBkUsky1T2usn2HEnBJneU6FNxZQlb3evapBtC4xK9RBt3fddRykZodTtWhV1rn7XKYCx8y11O3F1FYU+4C3gnK94UD357HEK/xIrM1IxcjGwG4h8mgkmmZVtMCilBV7TFoYgXnDO4VTpfMXh0nn8coYIBOfpu67i9im+LEvW60cZNz9JTsZmsybFuP1/MMQ2EN8NZMhGKmAq5M5TvK9eN2ZKLpyfklH0E1j+PYqtKCKUXL1oypmcrO6ocay/l9cQ01+l2EVSEurCHEB+D9GPYqS6qz0vKDBKeQ0pvYNih2SDXKTdjckLHSHu/Yj80XRBX3a8JExtZu8F3vuf8x4VbXeqGZZZZQ+2HqxhuVywUic+Sdp6cDNlotfPvfS0MAol57ZlGqmtdaEaNu1TJ3gSukDfd1ucmjXBtFgaFjeZGJhMyvU8p0UmKgTXIap0fc9iscQ5x+g+yZA/TEmZlHNbDBkrZ1DaQrFcPVlzVNPGImr1HMThneBVmQVHFzx98OztLOiCpwuBWT+riw9Qq550dfqHbFa/T4yRk9u32Ww2lJSJw0DJkPO7SemfYOUaKWVyyhRVct+RfaBYM+rSFldzQKo/Rcl/iJRjKuStNyanXI+TCzGNxJTJ5X5i/kGKvY5SrC2QY1yIaPgIImkbuE9GbRi5PEyM/5BcXkMqQkrVnFVBVRD9NF5uIu5x/osZ9Z9kFCvNO5dz6qoBbMMoGJiS8wIrHpMRk9uY1EhZuBMWnO9BZoYVRylzLHuQATgFYtsdJjqsWlKLcxrVpg2fNgrKneNvFWkwRVFRDMO70BZnxfgiikOhGMUyVtZgt4AEFERK/VwF0RpYbgNZQKdFqjXQRYTU98TQ4Zwj9TOCD+x0gcPFglnnCT7QzWaVarQaEFopFE7Aj7hYiBToCjllWNeAjgh5DFjpsZjJmjGFrEKxTCmOXHbIxbfVrEDG5BD0IqoJIaGk5oisBsMt2Jv+abvPdTGXeoxSJkTVrlvafLfJKB2l7FPKBUoWcnMAonXm4BZwCeRye8MfC+GAl9moS86cnd6uv7RFqupwjbqJJZNI5HKZ1fitxPF1eH2cuf+POH0S55RZV6N45x0ueEoprFYrhjRS8jWGzbeT02sRfQznfxbRpyqD0bCwkVv0DbRFpE6Z7ezgGjOizZi98wTvURHmXce871BVuhDwzpNzZlhvSDGx2Ww4evZpUkpYGrExArZlTdQ55vM5XdcRfGC+WBB8IIRA389QV6/Hd4GVKr/W9XwgdBRVvK/X/Drn+Dtd4JJzdeGFsF1YqtWMQs4sSvWci2EgpsgYI6erNTFlTtf3Mdz+ODl/lnFIrMeEpYF4+yny2Q1yvsZq+A5iepicCnGzwUrGua+jc/8Y1SMW/ftY9r+NaiL4Dq8ezOh0RhDDyrMU+UnMDhnLyCZtMDY4+ShOcqUb1ePEkyxTLNUFmYzNWoip7qcFQAUfwHeAXgT9Xoq8uVnUP/mydvbyGrUVVuvTGvFrXa5d1yHaV3wmmZFEtCUn4zex3nwrc/d+vP0u6j6PovS+wwePNqPOJXO2XhFTIqVdVptvYRzfhfO/Spi9D3Wfp1O3pfWKFTITnKs7RnCBfrGgD10zZlc9dDGk0Ix6zs6scsv7O7ssZnPGYeD45i0G25CGNSdP32S9WqHe451HVJnNPX42w2vHcnnIYrlkNptxeHCBWT+jn8/Z2d3HB0+Yzehmc2465d92Pb8ROkZRxHkQxynC3xbhQIQiQnIVHqn3iPc1WPUecZ5sxsIKyYxNjOjZKUOMpNunHN/8NDkmxs3IZhMpmzPSpz5MPv0kMb+Bk/FNDONriUNhfTqSU8bJqwn6IE5P2JvfYly8H+8Su4se6TsUwUvb2eQmIj+PGazHFbmcYqWgajipTiQUh1eHmjGWRgokY9zAmKQiTBWkEjy4AMg+hXdhfEuzqD8DRr0NaGyKficeuUX1OtFvCdFnUPlMYyXG5+Hn6atMfC5Tpiyi+gzOfRbnnsW5hLqaFdQJIxdFqfSg18pDO+eeB2W2KbEWJJlBiolBN6Q4w9kecbhEHCNnZ7uMm4FhFKw8BbLGO89sVr3rbD5nNpsRuo7FfM5yMaf0M67P6/duNmfR93U3CR0heG6p4znnSE7JaFuBwokqn3WePb3TqOXcqFURHxDnKQPEG5BPjCFGbq8hpshtBzd3IpHIauhYDwldC7uHC3ZPZ6SkeH/MEJ9mWA9ouU0cE5b3KekiVjxjvMRm8yDBr5iFNZ0fQWrSC5EW8Fd4YhIRzSCVksul5iEkZ1AhlUoAbGGlA2cGWlmZhvK2bIfhMNyLmtnLjqlhYi5czU65ig+hGbsoIsf04aco+ZcIeoRzX9wyHVMWsWbXanApJnU78zdh/m/I3a+j/gauO0LdvH5GCAiCa1SwqrKcL5jPexTBITU4lcqLmrQArmU9jzdnHMWI2WVy/DpK+oZKf40Jy4WUf4ciTxH6gYODPS5fvEAInvl8wWw2J3SBw4MDlsslj3c9793b5/Gux4UOP5shzlVK0HlGhMeExspkrCRA+GzX8SPzOYddwEQpzYjwvkERRbqu/vyEYP/RYR9WSknkOFLI5HeMxHffhV2ogW/Kxt76lO/YHfkrFztK2mE1PMKYnub28W2+8OQXWK02HD33Vp59+l2kuCAN7+BU7qEL18Hei8oH8c7hZgvUOXIxYh4puZCl4GYdpRTGNBLHWBNKltHYzt9VV+w7x2JBDciFirdFcJ6WxOJ5GdYXGl8Do26BRAvQnDqcOhDBl0btlRXef4jgN3hxOJ01jrpSRwhIrkTC5KmdKiJndPIhSjHUCa5ziOtQdXhXL1VbYOqc0neBeT+rlFVMjYWZMpu25VOtFDarM9ZnK3La4fTkLjbrv4SgeBr+Dsd0sz18CCwWCy4c7tN3XTXq+ZwQAgcH+ywXCx7vej62s8cHQtcMsgNRCpWfhbZJNBZH2hzeMMdvdA76DlMB55pRBwiheuq+r0adFX0kIL+siBVcSYgW3H0FfyEj97ZsgcGV9Snd6ikesBNIhdX4JDE/yXM3b5KHT3NysmJcL7DydmKcEfODWHk9XXiK3eWHWc5HDN+MsN6/2LKLlZ50qCklRYbU5o+COKlQxPWo1Niq68CVatC5XbfqRMHSducXt7CX1ainlQa0NOkd6WsmCDGJh8655crn2jZzZU0DoSotAyk4p6jVv5VG2cnEM5tVqg8oOdWkRFHiEBl0qP+fqlGLVcMXBCtXsXwfpXhWJ2vWpytyuMz6vgPGg5GwWtN94SnC6Snz+ePs7Qd8uMDmwiGPHR4iXaCbzekavFgul8zmMz7uA8ddoHjfaEOr23OpGb3tvWJ7+nWkiK1XkFPbnl1VT91h1OSIhZo8Kg8F5O2uxg5SEDXsYbAepABPCjyprHPgU+WAD1y+BqUQx0IuhguBndWK3Z0VOWWOjx9js36GYX0vw+YevJtR7HUM8RsptiKE6xQ7I6VEbOnyCqsCZh7GeylyWJmSlllRztDyFKanFLmJ6Icr5GTL/NZr1MatSuErWfXL7KnrVFXspNvkgmvQwqlizmFW8N7hs8c1oriUQk6JYRhqFq9kHFPqXHCuqx8hUrllsxpVN/47NqFQyZGcYuU9izFuhsaVT/w2NUDEEcdvYNz8ADkfcnpr5PR4pFwN2N+5RPnWU5af+wz7P/OT7HzyE1w4hPvu3aVb7PG7lw755bsucxYqLBAXKryYz5Cu40SVJ32gqGvUZqM3S8ZS3sYd55NXMbWVCHkAdZjq1lNbF6DrqqF3HQQPTpHv7OGbXPXgoS4AuebRg4BEhd9U5N84bvYz3vt9r+YD7zxEpaCpIKXwl5474vsPL3Df6RmXLpzh3b/j7LTj+vXv4OaNuzA7JJbv4fjsnXj/SWL5F4TwkXPtCjDvO+Y7C2AXid9FWn87xQI5FgqG8x/Hwv+L10coPArh/0Zt1lad1QWv2ozagMx2RbzA+JpgajgX6UwB2vTzBEt0ygI2sRhU9qQ0T41TpLSAz517fRqrYi0DWBomLrkmZ3Iq5FwoRYiawVI1qpKqum1ykWaMwz6bzdeR0xVOzyKnJwk7LMhdG+QvjvT9LdzBY4TwYWazy+zvPUS33GE42Ocz+3scBY85vzXCPOsp3lddg7TgtBiW22K3Ai19XrGIbYNEADGpi09qMsmm9HwOWI5bT02qQSP3J3AenKP0LZAMBp2DlaBPAx+Bccfx+fUuT1wIqBa6lHGlcM055qszLi9OOb59wsGFz+CDslr/V5ycFHKeU+x+xvTqmnL3C7KlrURBRDBVnA8gM9D7KbyVTE+aMrMK2XZaHuEIkaO2hiu/37ZwqrbGkJIw+TNo1NImavqOyJTFbR5qQt5NU11acCCQSktkZEVzDTbNnSdirAlfck4MY6TkQkqvYhy/jmIzSk7klBAR0uDxziFyRHAfQeVpKHdj+U3AIeu738zZA4FsifihBH8QCcOa/cceY/G+L7B/81keXC658ODDnB5e4Pfvvpf1fMFjuzM2s56syiTKoRTKegMiXDLjjcW42HaICocMcvPUAFZDfoPqpab75e5YvBOm7gLShS0UEV+9Mz6AOp72yke7wHNOeXXo+Iv9jGV0yL0d8m0d6zk8cnfms6lQ1CAXXDGecB2/uH/Io/2c60PmyWiklbGzl3hg+RHKIBzdus3Z6RnGMWebB4EDnCreO1SFEJ8lli+ADJh+DPE/gxaPSsZyAf00uTwLuQohchN04diqECeev2aOtWo/AJikq88fL6tRnwuInu+lp1ENu16IojhxNedUatq2YuX2GlU0Z9QEfPV6FV4UihXiGFmvNqRUiOl1bIYfwuwqlksT69A0yOD9o+wsrtOFz2P5fkr8hxR9LaevnXH03/SUEHE/POIeifRnz3Hv7/4G167/NhcXc9549S4u33Mf79vZ5b2XrvDZruO2Rk50JFOwWL8oBdlsYIzclxM/OAz85Zybqi3VAGir0hMQ174AJ+2SFW3pZWkaFUSQzqHBN17XVxZFplyA8H6n/B994JYqb+k7/ufZnLvVwRt6eLDjGef451fmfCp2FM4dx6Oh5/qVu+jNkMPL6L33Mx8if/3xG/zlx3+RdHLKox9/jM+PTzCOr+H2yQ8wjG+iC1UvHrzh5z/PIv9rVG9R9JfR/lehGJoqa4SMpHKrptqtEEtV6flQE1MqirqOruta0iySZTLq1Ze1s5fXU0+p4SnlD3/MsM9f+0KGb82jn1dZ3PlVIUohF8g5kDOkdEDO91DKXVBi3aonzUkG4YBSfDteV9O0doncJcqFDTkYbhlxPhJkxeLsiL2b19kph8zuu5+wt09aLLmxWPJ0CCRbk0uEIngztFReVsYI48BuSlxbb3hVrlLN2Iy6Mi5TyO8RaZi5JSBEFe9qAYCooKX+f03S53rPXG5qwfNA+TPesWeJuSp7VrgswlXnYF7gsJDVMes8llxFq81bnokwhppY6ZfCQh3LYSQe3qA7PEb1NmH+NN4/QcoHFOtJ+RKqM1Leren+fJmUFyhnFG6BHFe8rLnCLZmoOqGYI5unGGjRGlBuOexJ7/NfVnr6Jxo1Ba1b2eQWcjB5crZZPfO+Ru134Etri8KYqkMEGKu3K0Yc688p3sd6/XZyvkLOb4a8gxBx+iGc/xDCCEUQA+e+SO+/SOc8hSdQ91OYXGbn8YL/6QwB9p7u2XugZ9kNvO7hwn33vI6T3R1+/cpVjvb2eSx03EBIuVDGARlO6HPmrZuBNw8RSYnx9ilxvebenOk2G240If22smUrpgLVyeNuN6+qqQ6hVQi5KnlVwSWPJd+0JVUma7AVY11V5XuD5+1a0/0/Pa6rl+87bN1xW5RH5Qyjq/e4ZfNKCKRF5Z5FFZY75FnhkWt3IcXoT07YOzvjQVVOz3p48tc5vv1xzB5iyN/MJl/ATl5NlO9B9TmyvY9SPghScCHge60CsVyaOOo1xPj1FNvF6aPA72M2IuJQCRiGWN4uuhcaLy/8kCqnvNOwZTtr568RUbyrWSPTAibbCoytJ7PGWVPljzLWyophWBNTJKdLjMPfpeQ3InSozBFZEcLv0Pc/jLCqQnyTGhyFiFdP4XOU8ONYUfpPGrufKTh1XN2/xJWHLrKznPOG172a++59PR/r5/zihUt8cDZnY8ZpLqRcYDPAyW3mKfLNqzX/YL3BjZHjW8eszlaQMzqOPJvz88RZtcqm0nznRQDnoi/nHKXraoZUXaUERbAQcJ1vnroKp4yKT4vBXSp8v/MkFf5DF/h/hp4vqsO6UL8QNrGrC0ME6xw4Jc/n2GGBriMvd0g7uwwifAT41HLJpdNT/tZmw5tD4Nat26zGX8LkFqvNu7h18lpi2mGdH+J4/SCix8xmR/T9B3C+MJ95Qt+Tc9kmsEp6PTH+I0q5F+//P8weAxKCx0loi78G9C82XnZP/eUgxZfyjtK2Wp3ghU66L+POctFKGznMlsCcnDM5r1v510WsHGC2j+qA0yNETvDumODPEFnhrO4WtVB2UuYVRFcghqL4WKlHCYnNbkGXcLLwHPczjkPHLRFumeFzYT9ltBQsRhgi+2lkfxxZjiMuJlKMSKyQY2xVIhMXz5RIqhiqkrR3VHuYWQ2kUsSKYq40Ca02aFJvnNi5UZdWIOxU6MwoKiwFeqd0LoM0ysyEPoLF0iBLowS9R8YRRBhjZBMTWZWNACHQdx1nyyWr/T1iyYQdWK4jJitO1jfJtk+Z77LZ2Ue14OwQb5ehDC0AbllbFcQUUUNkBIY7gsEvGZMi8EXGyx4oBvXbymZgyyNvX9M816TJqMFElXXaHZgz50wuhVIWxPhtpPx2SjbGYSTnhHAfTu/Gq9D1H2cx+xWce47l/DGWi0OEHWiCJSu5ppFLxjnB9wtUhL3dPS4cHFK6jkeuXeW3rl5Gup5LFy6yN9/hGeBTpyvWrHhDynzHGLmSM/HkBvHWs4QUeXgYOR4jkjKrszXjMG4LDbCGk9s9CeronK+Jn0koQWVIzDJphHFVDdw5jw++FfQGQtdVh+GaCIiWcZVqNLml1F8fPD8UB85awa8Gj5kyphkx110g9AHnFYkRjQnzjg/MZvzC0XPcViWGgHnPUd/xwQfu54uXL7J/dMRrZz0PXr/OMzdAH/8pbm/m3HrjO7j5tm/HXMf+B95J/v0rdHKT2eI3Efl4/bzOU1WpnyKMP4LmJap/hNm6VdHU5NlW6/NnyqgFvE6gv0WLrSBgGz02T+4bm1FLnRqfC1vDNmji+zkxfT3D8PexIqSh1fA5Rwg9zsGs/yw7O+/B+2fYW87ZXe7Wj04Fy4WcIuuzU+I4ELxn3s/wznHl0hVedd99jLM577vrGu+/eoXRVXWZV8cYIydnp4zjyLWY+K5h4OGU2dw+Yn10RMmRFBOnqdJXcYitytq29LMiVSrgHEHrl0ArmGjlbDlVo06ZcRjIOeO8x/ehFiiEjhK6Gjw6RV3z/FMmTqkV2gKvjp6HUtWYB+/wXjGUs5wYygznlHnpCd4hY8SNA0WF4D2/2XXcco54eEg5PCR1gUfuvotPecfDR0e81Yw3XLrE4sknuHH8y9jpKbfe6Dl+9zeS3BK98VbCB/4KxZ4glydBHkOk4NU35/YEIXwS1YK6WgZmplWWui0Z+8p29jLDj3rTJzdtjcUocqdR11e+0LnfmaxRVcwyIs+i8kcUEdTVGjbvHV3X45zRhWfoQsZ7wbuW4KHmpqxpPFS16acDfdcTvGfWz2p1Sd/jnCeLIiZczoWDZMQYOR0G4jBwNSbcUGWaZRgpY6yceC7kZNvPmbKD5zmVtsCne9HS5KWc1/7VnxtbM+lTJphS063nCZtWO7i9idaSc628RkrBNdjjMvimKfclk0rGY/ic8SqItUpBEfZT5v6UmDnleDbj9nxO8VVJGJ3nLHiens842Flwa3fBbH/OgSscpcDsycIoBXfsEfOAa5KHWBd20687BadGrZYvWyhmE1OFVYmr/hlS6YkIXdeRzUitCUy2dD4JbbIRqXWCIttSL6wWf05qPT8Zo9uQ089i6feqfbiKI/uuY7lcVs87u87uUnC6Rxeq8N9KIZaRGEfACKGn8x3L+ZyLB4f0XceVS5e5dHiJTdcx7xZIVvbN+FvrDd8Qx6qnPjpis15zOWVkHLmRM+PqmOHkrAp6pAregYYbW9GuulaQ4KhxgRAzLfi9o5zNjJRT4+pr7eWWYy/WMHRN3NSVYkylJUIrx5L6EoEGt4zSql1EhTIVvJph3rVquVo1n0qlG19bMv+wJJ5T5VeHNb8W18R+hl67TFl2PDnveM9dl/nN3Tn3zT1vyWv2TlaE61c4+VFjPY50T4AUBc3EuGa9Pmnzs6wBcjasBY9WrCaipGDRVYciwsy3qqMXGV/RqEXkR4HvAp41sze2v10AfhJ4APgs8H1mdvSVj8UWJ1dcOa3C50ezExtg2qqNt25btrWG6hzODCHh9BG8fnj7GQB937NY7NOFjnnXs5gtcDpv2piqR7bGoChSG7+IMuvn7Cx3mfU9O8tdloslLoQqjy3CPBfeOETetd4wbjYc3T5lvVmTU2IcIqelEDdrxvWAWWkS27pAnavMTi0fC1V/0qDYdtdqQWEtfaqGvZ1ka/09aIKgrWeegkva98K2/m8KxszaLleqZZea0MkChqNYBJphdxnLSsmJNNbmM5fTyNePG45F+KPOQx/IKZLSASgcd45H9pZo53l7ibzz9lXuWa65/oe77P2u4VbVQKtirJByJMahqvVY4KpxELzHSSHHTCwJE2sJs9oWwofapOfFxlfjqf8F8H8BP37H3/4p8Ctm9s9aY8h/CvwvX/FIzQOL2RZp1Crt6q2niYQOCioAACAASURBVCiTSHyarJZGrn0+3FYj4p2jiFSDC1UnMPXSmPUz5rMZXejofDVYoRaKxpzIpTBuNozDQOcD3axWviyXS3Z3dpj3M/quq81lEO7OK96yHtgvhZ2zM1YNdgxj1QinJrbKOZNjIplgKF4c6qreA6fbLJ+JwxCKQczWwouC2aT9KOdtAfJ5kmnCZ9YWQs2uG0mbuEusZlmF2hRm8tgTbFOQWA+jd0CgJI6MgM9kr2gpZMtYrtoYKRkphSDC/ePAO1YromXmJ8eE3nOaC5/ZbLiVEkcKj+7scEMDNw5P2Lv4MWb9kiHezzjeg2gP+fWU+NcoskbsBiprTAquMV+Zuyjl7to9wX8RKdcr/nZC371Eozaz3xSRB77kz98N/LX2848Bv85XYdRC5WJLMcQilFKNYRwaE3CeRSwlV/xdDEtpS3N5fAtyfL1gM5xB8h5VoWvFqrO+Z293l+DD1lNZMdabDWcnp5Rcg640RmS5YOfSVQ729jnY3+fua3cxn80QqxUvjJG3rjdcXm/QUrg6jlyPtXPUel3bEIzDyOpsVQNBrZ2Jqi6lwpqt/tk1z0xztjmTYqwGXXJV4lnFvtikrp5i6CbJbegyteSRWTVAEdBklatu9/t8Iuu3gmGtCuVcNenwPqMu4p1jTIkSfPWSruUGcsLlxELgnWenvNky0gV2KMxPjvmEKj8cAted8mmv/PTdV5nlzENnX+QvPPdhwkng+vXv4sZzl2rvkfi9jPFduPJJZPdf4XkUEQMnFPEkezt5/Ltg0Ll/jejP4hQWnWe5nL2onf1JMfVVM/ti+/lp4OpX8ybZpsnv5JrPI1sx2QaAuWkAqiuqQcMURE1FBrXnhWHe151WtRW21p4YfStyrcR+qsLzlBiHgZIzcYzkGLFsdD4wn1XvvpjPmc/m1Vg3G8iFS+sNs9Naa5dTYp0qxh3HSM6JmFKjEzPiBQmKoBRRTF1V5jm/bUYy9RHMVkhTIJhLaydgaNupoXWdmogMpXp9a4rVbWBdGnZuaj+aTz/PVdWYEiPT7i11B1DNWKn3rZRC9q4eo3W5NKF6ajO8GXelyH2D4Eti9+yUhYJ5z+5ijgXPqVdOZzOcGRf31sz3P81CHaenNwkukpmT4wPk7ChBEduZkBLaYirsMlb+QosfLtWFbobTytq82HjJgaKZmch0+//4uLOX3uE8nLcC04rfarp829+QqdMPog1+FEpq3Yw0MQwR5zLBBTrvW7Vxh/P12F5dNaZsrFcbBh3rVt4Cr5JLLe9ygcO9A/quYzFfcOHwArs7u/TdjJQym2Fgs1qzPj0j58TZ2Yr1atUWX73BZUs1wVShLgDiKK6mnMei5LHqFXRMzdDOF2pOkXEcKrWYIzmOYB1S3gjlYUROUfkDVD6P07r1ukbZTU5CVVrlT1Wa1uz6edu17Vwg1ftu3X/92ZkhruBLQjDyOEBOtTLFaiperRCkwh5yJg8DliNnx8I4rgkh8I64w4W+4+lZz8fDDmfOcfPCAY889BpmpwWXnmBx+2fIwz63129hWD2AswOGs2/Ccw8iT+HkYzi3ZjH7I7T8h5oY6j5NFxTvjDRsWL0gN1bHn9SonxGRu8zsiyJyF/DsC73wzl569x3OTRGcVGrPqP3kVLVCkdy6GUltZaDavNe21KriVOcUZhVLqyhdH6oOwqgsQDFyKpxsTs+Zky39Ra0e7zruu+derly+ROcDu8td+sb1DmNkHCOnt29zfPOotkAYNmyGAQE657e0Uu20VA8t1CA0u0D2c0yk9pSLtedFikOFJ22hmhVyjMRhU9mYcSANG8wOIb8FK9+P8ARePovKBwmhbr0heCYKRKhxxNTNNQSH99LET661Iq4VQRVH6zb9Psk7q9w04UWhJHKJJBGcd3TWIU5xKmhTC44pEmNt2LnenGIKXd/x3cMBZT7n/Xt7PLez4HNdxxeuXuW5gz361YY3rj/K6575adLpXdy+8T+xOn4VpGusF/81rox03a/Rzz+Pc7fp5x9ip/9k3b11heKAwrg6Yzg9+S9i1O8BfhD4Z+37v//PPUANFCeXcU6sbluINanptmFK84pT9nFiASav72p1ZmskaLVcK4aG3xNKRDD8HXz0fDZnZ7mDd742gnRTEqAWIsSYanvaWHtnxJhaLkO20GDLq1tN4bdwrPWCk5opS7V3XhwTKUWsCCUqVhzZIGrGpBBdIboKBRgPYLgLYcDrDiod2QTfFYol6mZdG+N4beVsUjtJFas9AUswXNGtN5eWTnfWqNPpfK00bXPBRGrmjoYCSt7SrDI13WmvN2q3qkwhYFyMI513XEyR0LjzIXiGsKR3Qpqt6cIXUC8Iq1YgESj5EjkJ5i4gOJQC7gSvZ+1uVoLASmsy2Vr5vtD4aii9n6AGhZdE5Engf6Ma80+JyH8PfA74vq90nDrxNQuYWquq0vTEU5vW3LTOItLYquapW/fOnGv/ZhrV51Tx6ui6jr7rKTmzHmvWbtjcze3jv0KMh/TdJ1jMPoh3K3Z39riwf0jf91w4vMByvmRKfsSUzzF3yaxWazbj2ILZyDiOAMQx3dFRdapgKaAOFWUswu1NqgY7bEjDhpIzm/WKNIyUchdx+AZKuZvycCb/1Ug5MGy6vrMF+r43ob8PWvYJ/m/i9AGCPUnR9xP8MxivB/tmYB+nNXEhckYIv413H0XUWuuHBksa1eFc64ctUvs/qxJUcH1CQ+veqopJ7adtJaFea5+VJprCMmoFE6vf1ZCcsGGgSO2QOhwfsRlW+C4QZj1BhL3FnKuHByS3x+m+p2wSoauNhkiCuQIpAbF1GWifhyKm1aHl81jjT2zUZvb9L/Bf7/pqDPlLDkZKqbZ8jYlcai+3yahLKrW0SRpvq1Pf48pra5bGLmgNGqSKjWZdx3w+I46RjZ2SYmSzvsato3czbB5id/c/0LmP4fSMvd097r33PvquZ39nh+W8CqGGzUBKiXEcWZ1WFmO1WrNuWpJxGJtu41yDoqIE7driqo0cRYQxwq0hEXNhWK+I63a801PGzUBOD9VOUvlt2NsK+XsMXtM6oJogN4Vwoyf8Lmg6oA/vxrvvxOffJtsncO7zYK/B+AfA/ZXGExB5Bu+u493vgOTnFTZPnto5hwu1Ygj1iHMEFfp+IHSNN3e1iNcHTy4R5x2+CxWKSMXlKlPyq3psyYKNGwqFJLAJjnXXMd/bpe8CXoX9xZxrh4ckv8/t5zx5k7dN3C1R2wOnCFQWpgtVzVm1sEIpUvXvL9Wo/7TH85pm35FUqdSSnJ+x3BkM1dd8adVMrUJv4p8WYBabMnERbAWcIjLiXBVJdV3HrO/pQrft5H+eji7bPstTr+VcqnCqyjht+zqbPNXUxgyqekxqDDikUo06ZsaYyBibnV3G3X1yusg47JPzDnbBKDsGy8r+qCmyAQ2gTZGWbV6N3Q6I+QrFrgGXMfaBXZTJqNeYXSLbNYTcsq9y3mBRwBWHYzJqhxSPKYyqjLTkUOv2ZGQkjjUxo4JLrXyuZfaRmtHc9kdpTFVfMpdS5Fik6VZqqdbae45nM0r00J8w65/BbI5xESsOsx4rl7Byuk2Vq0xsz501rC9uY1+jGsWpvUErsJXWFgxDpBr3pLvGrAZlVpMvIYSaTfRNQG9GiiMbq1qMuBmI44jlT9F3P4bqPns717l4ITCfXeLShQtcvHChZq6mBpBZGMfIZrNhs9lw+/S0HmscGTb1kREx1uYvpXHrOSdUFO9yDVaDQ7VHGvR49iQypMK4OmVcn5EPL7B+17cxvulNFLtKTPdSiiCvEfQAKNVRVY13RTOlZcCxlqRJDzCk/w6R64g9iLDPJI6qZQFLVP8mKg8wOYnJKfipb4wTfGhVMa7WOQaBztX6RFFBfW2S6btPMtv8As4/TT+fMd9Z1C4AweH8lIpXvLV+NKka4us2kX/kzjjyAx/qA7+REqcCH9rd48m772HnVHn49Le4t3yYYXiA546/i83wIJmHie5/wPwRffh9vPw2qmsmfUwRJXv3fMf4ZcbXqJr8zi2xCd6LodqMWqX1opOmt9WtUfuu2/aQhpoYSDFBrnWJaYzkmMCeJITP4hwsFhfY372LxXyH/b199vf3cKrkVAVHEIkpMQwj683A2XpNjJXDTmNsWc/aC6MUY4zVk0vLYNZmK4E+1EW6GkeOztasYyauVsT1mnwpsHrbOxjf/d2Y1LKlKqSqIisx8FYnRG1KbVdDP+/Jfo2SvmsblLZep3fowOYIX8/06J3Jo6mA1+rNnQcfWq2nU/BKwJiTUBocCFW+GvpfI+ZfxfkTUs7gGiNSAkKdH6fVsF2xmviRwquIvFqM5B0yLPntnDlyyseXSx65fIXL81PuufERrqyf5PT0Ldw+fhsWX0OWV5HGV7WSO4eTP8DpAO0qJ/iUy58xo56SvJjdseLOt5Q7++tVAT3bx1k4X1mLqXJm29MjZ5LVBwFVQ3F0XVXpCcL+3g7LnR3ms1ltjRsTRZUUIzm1B/A0Q80T7Mi1c+id/frYnptuNShGTevnXIgxIapNUB8ZihAvXyYvF+R77yUfHlJ8gLUgTwucCSq3cfoMwoijNoTXI9BbrduDA3HSREYLTK5gNkfkFsgz0FidevsCcBnhAhOkA1pWg+nRKq08rIDcxLhJ8Uq8fJFhfw8dBXdTkBWY7uPjg3WxpQ0xrjCrsl5r/atheoxHU9pRCy98Y4g0FWwYazsHA7oO7Tr6PjCfB1L0NclqBYqSc6XucpbWjSuB1KNOSr0/Y/CjpcFKxb6TUH56lNu0V4oIIVTuWRG69nwQ7wNd31dcVwopVS86rjbEYcSp0oeevlF1Vy5fZjabs7uzw+HBAcF7gnpObp9iUz1jTIxx5PbpKcNmaBBkIMXYVHEt+9b43aI1LY+62kE1xZbYiViKGPDc7YEbx2s28wX2ze+E7/h2bG+f/JqHKeqQJwT/44I+CsE9Qu//JU6+WFkMETRCeFwIPYgJztfdoKQ3YOX7KTyEuI+i7l8i8kx7ypYAh2Dfh9m3Uxsp1lvOeayFOalYQSJFf52iPwP7c86+8+9h3/jXkc8r7idAPwphfJjRfgjvbxPt/Zi+Bx+OgQoPvQPtWlveVuDgXGWAXGmP6jsdsGePMO8IfUe/d8Bu6Ll8+Qr3pMQNd4EnnlCkRHL0DKaIQt9H1ssznD9DtUM1nCfmXlyk9zWAHy1fu5WUTl7mPCLcNrLRibJzofLLIdD3fWUYxoFhjOScqyGeraoWer/H+8BiueTK1atVnDSbsbvcQUUZVhs264GSS9V+xESMkWEYGcaRMVZx0tTRaaqHmx4fJ2LIJIvNGaMW/CbOH6O33mw422zYdDPkwdcg3/qt0M+2pSjulqAfBP9+8O5puu43cfqpRs21ekVqZlBFWlsERTiD+J31HupT4H8d0c9WDykCXMPs67di+nK+wZznoFufFKRg8jgmv0SZ7TK+/puQbxHkUcX9rKGDkewS6q/gQ8aFm/TDz1FKInaZko2ql2p6FFGc1PkSaWVyNVODna4w79HQ0c3m9CIslzvs7e6xWS3wqluuPLVFGGMmpkghtpZy085YPfWLweqXt5fenTx1K8dKbasHnretS4vEsSb6KU2F15q5pJTaI9rq49dCqFz1znKXneWStLfPJ/b3YWeHVznPG3zAWy0DW69W5GbUsQmTYoyN6ZgebUY9v0b0q2sC9kbJiKs96kyqjDWnC4zxVWSbsb4aGV4/MB7soK96ABXH1GW/Qq1boH+E6RG4j4FuKjc8tddCzh/LwVStZxSOQH8P4Qj1H8eHjGqHVyVMQqnyGaz8FvWhQGVrBFNGsWZzBdiAfKaWCIxK+bhgvwJ8Hspz1dUokJoQLKbCGOsijjFvM4opFULe0tcV2mgTQYlx1zDytpS54R3PLBdcL7Ay5RNhxmy2ZDObs+4Ds86RTIm5ymtjvpdh/CZcOcKHL+DcM4hYvUdfAX+87E3XN5v6cMdNC8ByKqRUvWEIvu6M7Z9S++qNKYIVfPHbR8/lVshqBt4HvAssFwuuXr3GhYMDPnnhAu+971U8vbvD3xgTD20GdmKVh9587rktJx3jeb89awxKzvXRaGNMbMahKgGDr49+kyqacs5jSbBxJIswpNdydvz3idzL0TcYpz9gDBcc/to1nOsQ0/YYOUA/R/E/ggt/gLkTzB+DTu0J6oRNHUQnhrN2qPoMEn4EsTk+nNLNRpzu0gVHCDXRUvKvYeUj55nR0vpkNJ1IUwZhZLBnavHBmaP8eyH9AsgG5Lo0ZgkkGc4KMib8esTHEedCjW+iJ/iCd1UeW2pjWkyM4uou/Oa44WAcuOUd/26xw88fCDfxvGe+xy8fwqXhAm/ZW3DlrGM9wHBmpCxshq/n5OxhnD+i6/8VoX8Pognf96h/idLTP9XRPGVuou9cJkahpb4nKVlN0laiv9RUc9U41ayiSHvQZK7djKZ2AiG0xubLJWW55AvLJZ9ZLLnOhjImVGoiZxyHCjmaDvrOsS1csDufyGXbej/TlkJuuu2pgWe2Xcb8EAMPMhxCfJ2QLlXjoFAbvTcIYJyCfAqTPwT1oB0mvn7G1gvlrfLOmrpOOEX0NoKhrsP5GU4DLjh8Nz1f8hmsfKFyw/mOdgITRWJThWfFJ6YCyWFPgp215jKVpiZjtQ2YtU5SqTZnTLk6IpGaAS6tn7Rlto3SJ099GBP9ZsOxc1yMCQqMJjzpA6mbcX/X8cZO6QKMaQq8hVwuktI1Ckeov4iUmvDR6d6/iJl9bVok0LAY7RnhLblSvxSz8954pVxh2LyenPbx7rPE8DFUzwiuPqHKOc/e7h7L+Q6LxZxLly5ysLfPg7MZ3z6MXOeMN52tGG4dczTG+tjmbR/smmyp0oYpa1XaE7VSZTE2A8Vq3Z5rBb2iVUIaU2JMiTEnNndlzt5ujEtIO4L7OcX7AcfHUPsEspfhrcCrwTghlbdAfghwtcNSSyzUnWqDl4/i9BNVByFSe+c3GKEiaHD1sdBav4fga+InxdZjUphaK9wpaqtVYAXDk8rXkfLrkbzE9Cba/UQjs6epsNo6rWTG9Lush1NcSvgQcS4SvNH5AREl+FpBlFM+5yOFGqcMmdFBXkdYDyDgcWiYMZut2N/7ABfWj6HhGmfj65C4BDXGlFH7/5l70yDbsqvO77f23me4Q2a+fENNUmlCswQICTQiC0kMxoYWZuqGDqCbptvR3xy2IxzhCIcdDn+wv9hhu6eAYGrCWM3Q0CCEkJBoYc1CY0mipNJQJanGV++9fC/zDuecvffyh7XPubdKqqIQRIlTkZX5MvPevPecdfZe67/+6/9PUFlDzTvrWcjfwuTL3/ohBfoRNdLPjh5pDZmxVLcV+emsVv+QoXsu3v0HKn8Hzp9wdHjEwWJB0zTcdsut3HTpZtqm4aaLF1kuFhwpPHPb0W+2pBs32Fy5yqrvWZ+tpqAe5YFHOEqQYp0WGeLAtu9YrzckNUUhS0E8eI86YYgDm8GKy/WzEqc/p3S3QP8WIfyig5MOl/4Uyb8Kz+iQ/xZ4JmS+nSH9c1J8idlFpL06DkXkKrX/F9Tu02YVUhSSxmB2IoQ6EJq6cF8q6qYGlOiEFIvTllgHR0eFq7Lip6yoVsT4Wvrh53G6Qf2v4me/DCTbjUpjK6nVGNthTcqnpa7o0eyKyVMgJagqT06ZemQQCvZ6onm4dwHSqoOzNc57auehnrOcX+fihbdyCyt88yqub29BtgsiyrZPiEvkKpHrRHBK472J1D/O8U1cqUdMevqJfXpUVavqyWlOSgeItmQp5ZPIhFu3jRH7m6Yx56wQmMWE762YvNH1XO0s5YgTqrHHCNzbkjUXPb7CDkwFfpSci7HR7nt573OqM/ESxFshB5CHwF3OSL6OpHuh6eAasAI2t6OpRjkia0l5EEtBqBFpyHJMlgWiPTqlIqVpVcbCrAlVPsbZTefITqZRLp1O7U6Nf0zwsh6Q9VZET0EGkPuBWBp4JT3RVNh44Ar9d0o/yIWclkvxngoKw/S3UllAUKVKyjwmnMJQVSTn8S5T1zdommvU9TVCOMP5NeSKnCujEes4kmnnaRy0eKzjSZ8mDyGACikpkrMNQY8ds7J6AgwFJ47xi3j3m9T+InV9J22bCeGA4+NjLl28aE5Xx+c4PDjEe0cukFq37bhxcqOYDa24fnJiLgTTlPao3C/lokVUMV+SwSC9pIoEj6ipjY7QWszKth+m1bzre7poc4mJYi7kBQ1iWogOWIH7E+Ar4L5yL+7+NyPufWNqi+BRfTnKdyMyJ7vvQVmS9SGyvpvM5wjO07aN2dTVNc2sLcSfQF0F7Nb0ttuJQiwBpjutlKTGejTPxGJhoSYSGV1j3Z4xZiQBNmSRMUVZJ1D1CS8DVZXNddcF4pBNprIaBcXtooZguiSHwfOaIXHxxporVeA9c+GzjUdcxcHBMeelotMrLK79DtktSdtvY4ivMCKbKj6bilPCkXl8BOTJD+rKOMCpmKgnKPN4MPKpjWsxmDRA+iJe7sJ5pWkqFsuaqjrg3PExFy9dom1azp075vDgoKAXPcMwsDpbcfnyZbabLdvNhtVqRc6ZOlRUflQ4BZOx0mkYtxsG+hiNGquK+OIo6z3eGRUypUzCYMBt39P1HV0ciGrKb+qcxUaRiXYKcgbyJyB/CpLuw3Vvtl/IxcBAG4T/EuU7ETkm6/eQ9VWIfJ7MV8ncaU2JtqVtGgvqdob3njo4quBKcejJye86b7rrehp2vbfLFG1sJBFlDOpQ5L9AixC9kpCccBmcZrxEnAoxZppqwEkgVhkvQo6jjrQtTotFSz2rmVWe18TMa0/XfLmquM83fCaAcxUHh+c438xYx4eZLz9CZGCd/gHD6lvJQMgQ1TqUWRx5lKB6jOOb5iQwHTq2zD2qxjoTTItCnCBpjXMPI2FLVTmaxrSKm6axdKOpJ3NR3aO2jqR+uzkMaZkGDWQ3gjV+GBIzXuxRidSm35VC2fSjCqmUx1VkvWDp6uY8PBjAg56ckdMZWa8gemr3ay5wmYLJ6GwoMni2t8oA2pVzIkADVKgeonoTqk+xgtEHQjBOtAmb71KQ0XVMHJZ+2JPZJ9l9vUtJ2JtxNExcdr9MMby2LV8LHbhcsqw63ShTp6egVOM5HZ3FfVkQKoUqJpbOcSlnbsvW1J/5QKgyIYD3W5zvreOJ2tPqbu3PMDlhP9bx5BsZsZOsGPPWnBI5z4nxe4n5tZZHl8bBrP4CF5dvo63v5fDoiIuXLtG0DTddupmLN91sI0susNkaCelG8ePebras1iv6zkj+kzBMjCY3ptZQSKrElOh6c4e1TpbBji545m1d0qaGUJny5jb2trLrbcT4BobhGaRPPw35V5fw80T+/AfpTv6I1D9MkE8QJNqcigPK7h4oO2iGaUizSHIU5K10BC+Q80+QeQVO7mU2fx/L5YOllmiKZ455mGfNxNTjkp/MnKYzP/p+OKXwOcEbQ1ERoghGHZKip4I1jDRQuik2xIAJ4eSc0WxEs1E0x/yYSvAX1Z3gArN2SQiBIcN6PdAMyg/NIi+rIucVnt20tHVDPTsl1HNCFcBXDGr5dIWWGR/zDvj6/gG745vgTb4j2dtNnst0S0U/vJQ+/iQ5B8tpU8ItP0x14Q4W8xMOj464cKnk0RcucHTuHIJj2G6tzd113Dg9Zb1emZPAdmsyYKNPjGrpGCbrGOqIMWf6YaAfemIsXjFZ8S4Ui2VXgrom5US/MT2MrBeI+ftI6RXkewT3WY9LA8qd9Pw7Eg+hhXLnHYTSbRMZkQ52K6ZjtDK3fzMG9SGaX1fSiU9QN3fTzk6og+XVJtdlMF/OGR+KxXMpFKdnG4N6bJmXdvTYwUwIg1oXUUUmhzKHN5UnTaiOkgy2GueJS10K/2yVqRRetYpNJzW1OQX3fUfXdVQJXtlHZkNCvSNXjen1NS2havEBEBt1iyWw09jhBGLZAR7reNIJTVZYlzedlBwvkoYjcr6I5Et4At4lqvp+VK5zMH+Ao4XnYHHIcrZgVrc0VUMlHkm27uc+MnS9EZT6vtBPh50gzEjiBxi3TSjohhZ5sNJO9kIIVnCG4PGheJV7bBC4FN/OjdPjNgysh0K6HVIFiBIk4VKmugzVlUL6vx3cBZCVAQ2yYod+OSHfBPmCIh6yV5JT6DvyQw+ipzeAexAX8b7Bh4q6agvnfBS+SVikGtPNSShEey3mP7YzBpdRbcmxYsiCaIW4pyDyIsBG57KA6HXIXwFdF6jREoqI0JfkYpNNxrgSwas1TsR5nK8sL/eeJPbahtTTxy1BEuSeoAMrPPd6x5kXTgJkp1ROqUTN+VY9Lt+KpBcjqUOHDamM1T3W8SRL+UJV0g36AY2OuH4F3ekPk/U8VfVc6hAI4WEOjt5K236c48PIs59acbj4FmbzOYeHh/gQaFyN6yI5ZbY3zjg7O2Poe9Yn10uDpUgr5F1maM1hE37JqgwFjlK1C19VjgBUjZ2WUAfqppCJgiDeIL0qZnueZNICSCa/QNj+mCfeZh21Awd6BtVvQ/2HwBHojwCvA/kMyK8DnzbSvvcOqRzDdwvpR0APlKG8Vn/fZfybfwv34Q+hrlgm1sfMZi2HywO8c2y3G7abFaoRtC+Dxp4qgFMTYfSyG8BwIqjOuTosGHqH5kOc/CAiL4Xydw08+wgiv4Zw1VbiIsOQndAh+Ozo+8QJPU0dON80tAFmdc3RfGnY9bxlwwCp56w74cbpCXVTcdh5yMr91Py79pDP1jW3zZQXNonDOrF0mUVWhtTSDN9P1b0El+4n+d+h7z/M12C/e8cTGby9HZMcu7k80y+q6v/5jejpiRaxlcmJSsj97cTudaieo3YVVRAav+Zo/lkWy/dz/vAcF4+eydHy2EaxSsUvCDJkSIm47ejWG4a+p990xG3HpDK6R+cSysgXhgCk4rmyI/1YwSg7dgAAIABJREFU5RaKMmsIwYRTXKFrllTCecV7cN4CBVHyzUJ8NQzPEcSZqxvXoP6QsZxpIb4I8htAWuB3MaowFtjiHPHpAq+HfDzBxOhdK9I7P0nmnSDncf4F+HBMVc1p2gOCczarSl8KugBqGtde6sIoVCqbsbX823tynrNyNT45NLc4ng88H4AC4IF0iGsRsQaVFBOlXHoNDiHHzJZIK44qW7HpnId2hqsDWjl6DDbs4oauP0OlJqUN5Bk3gE+GxIdq5SWV8rygtF5pRKkUyBUhPRuXXoTji2j3DmJePy6n6Yms1BH4b1T1oyJyAHxERN4B/CP+mnp6WQ/YbN9A3w3EuMX01Bvq+j2ItCwWM+aLlqZ5iKOjLcuDIw4WS+qmKe3p0SMk7VCLGK0V2/cTDDikNHEcDMTYE2OYINTCsxCYjMjH0NexWNPJh9FwN5nQEYMkdfd8umssfc3hgA3IJ0AOgDuP4drzUM6jKuTskFyj+nyyVmgP8nlwXxb8fY7qqqeug+HwORFjzxADw9CRnSfGgZRNGdWk1yqyL0oZ0RJ1GeG5LGSBnDtE7iKEPyXnI0Sfg+hTUW4g3An6ECIfReRsavoIChrQ/CxUnwk6kN2XyDxADs5ao3nUIik0Wphcx3La+5jQpp2eizhPCBVVlanqirr2JrJT34UPDyD+PsRdZjQ1eqzjiUyT3w/cX74+FZG/BJ7CN6Cnl9ItnFz/7wozL6G5x/l3sFj+C7zfcuHCeY7OHdI2wqWblYPD25lVDYvFgjrUIEZdBWPT9YPJIZytzliv18Q4sC2ijVMRCsYyKWLmWqC6/cAW2XGPdxwU+17M2QqljGHUOU1dtBjNBcwgJxkHTKbcE+wb4oATkN8G93Zg9XR44J+h8lKDrKJAFGK6QMozOFPCW4Xq3zuqzjPb1MwXLXUTyLmn7zcEB+vKfN27bs0wmMpqFTy1a+2Gr3ojjsWefmPd1VwIUpp7nHsnTfsecroJ0j9D01MQHsDxb0HfB3KGyEOI+D0wsELT95D1H5LlBJ9/Efw9VFI0D1PCqRZXBI8TSwOz7qQxnE8Fbo0MVSKrQbjOVzSzObMhMF/MmB+0DDHh23fimt9H3Ar8VSOBPc7x18qpi1DkdwAf5BvQ01Od0fcvQLMpdcIG7/4M7+4ihFOa9iKz2TFtO2O+uIX5/JA21JNVxti+VgyGG+KwN/m9PwGed0Gt42DqyNNmujwjHVNVv06HSqbf07GDofvi549a9UudJqnsAM6j+N1TdSBfAb4MyiHwHJBvK+fFnmdEFYgK94N8ykhf/pwjLDzei6VPOZGyjZ7hc0F0iqCkYPyUrIgW5p8m4yLv/SF7z9dw7hS4gurDoAOqK4QvgdwBCDLqmoxhrR7Vm03njitkOT+J9tjgh3nRjB7z40IySRHrKEy0O5d5uhjF2jsEfHCEoKhEXLgf8Z8urfx6d14f43jCQS0iS+B3gf9KVW88UuTxsfX09rX0Fv5pbDZx94adUDcNs9kBdSUcHx9z4cIFmrbl3LkjlgcHVBJoQksQTz/09NstKSXWmzWrtZm7n61WbNYbYjKe9sTvGAUlRXdoxTgoqxMB8xHpybjVTiRk7Jfy2GYuOiSmVWKIggDuq+D+ANzNDp7/QtJ3/DTkU3vUGOcypvjPRssasAsXxiV0gslGNagQvHHNi+ur0W4TKZmmdEoDuXhDMtqPFBhRgSSlfoiD5crlZs8pFhrFBnXvL8F+H8r9jEzJycqEPa5MkXUTmRF5FagnxRvk/EVUTwwOHwXycxkISZGhj/R9BJFpyHnrg03ul16Cd57gIyF8lhB+C5Oz/BQ59agkm7L/K+a5nlBQi0iFBfT/o6r/vnz7Cenp7WvpHfuX6tlZT1172rYiiNDOZhyfN8WkSzffxKWL1ly5cOEC88XCLmo2rDSeJbpoetA3VqecnJwQY2RztqHbmJRBXyZYpu7F3gkWEXwVGJ1+Rqz80YE83bAljVAg6Z4uSDIlp1xSIRFwn1P8F0Ebh/yDl5Oe80KQNNo/Tt1DEVBqVA/G88OoOW/e6GauNGoOenHUVSgKriWoY7I8OvagQhp6UhxMrriyItFkce3vxl5JQ0/fb0teHQ0JyhbUwim4t4C+EyWS9RQtOuAixkMdyVv2tRJjQlgi+YfQ9L3EcCc5/jKargBq2L73DKkvcmsm4dZ1A1lhvdlSrzasxVvKWNeklIx1GIS6+gvq+l2oRJKeMsRtceM15au/UVCLXeFfBv5SVf/3vR99A3p644RJQqRHpCOEgboORp8s8ruj0EwIFsyCea1YvWbb7VBa4eM8YSwr55ii7PqrpcsgZUUdV5vxFY09ihLIOxbsyHLbESimdu0jkJWirdEJcqI2KvXwAjlZoEGRvjNUV3bPbc/YoXSMN4UCMgTktEZUccOASMTJCiep8KjdJHe2r+khpQWpjFu8oRdO8+7fRf86l/QFBc0Vqm15DRuQ69Pr41Fffc2VVMVsPc6VtOlhVOupySV7D58oCXmvcTMOh+TELCvLnGm19DEEhDUiDyISQWuyLgujUciTbOvXB9ueyEr9GuBngDtE5OPle/8934Cengk0DjTVF1jMPkRdX+Po4G7On5vRNC2HyyXz2czI71IGW7OYz0mGrhtYrTZ0XcfpjRXXr5/aytkPpCHtXVRbX40Mr4bE5YJOZJ28fsaYN024sQuyi5QdsCETRTRlLYWNkJ1DgkPU44A6QRKl/ZQSfwlUNrQffy+N/xBU8RFUW91xuOw1eM/wiZfhf+nVoErzqffR+I/ShBOa5m7auqJtKxZzIzTNZy1HBwu8c5yixH5DTpmh3zCUFrUrRkTb9Yau25hksGayJtBASt9Fiq9E6KnkvYTqDlRNcmKsXSBjvoVj291uKO9Nc2EEkPYPayiqqRVTpIYLHh+8FbeaIQ6Zm4bEj3YD/0nVc9OQOJ6IVzZRk1LLMHw3ffxOMsFYelNQ/w9fN86eCPrxHh77lv3r6ekJeBepq8+ymP8mTX0vRwcXOT66iaapOVgsmc9aMzEC6zpmIZtDwxTU2+2W07Mzbtw4Jacy3ZFLY6G4EVjdVlh/47YugguBIrQ/tespgV1Aq5Jjyy7w7aWU1WUk1QgqpnAk6vEolWZchuYzSrwDkDWt/3Nq/8tIvZ1WWEudRwQGLBFq8J/6x7hPfgeQqd27qcNv0NRKUx3R1DNmTc1y3jKbtSzmM44OD3FOSMOWzZkQc6YbNvTbDVBGyBT6bkPXbxiGflq1wZPiS8nxn+JkRVPfoK4+R86D4dRlj8+jBDJMQW3pnExBvRcreyvx+NidGpd3Rmxy4qe8/NKQ+ZFuoKkGzobIZS3713i+U2NB3f8TMo2J2E+Q3jcY1H+bhzUaTDy7CgNV1VNVSlXGkcZJ8bFiHr3H05DIicK8GwrZP03FmnEPdnmwOGd01nI8wktS91IPbJUct/+C/5X04FEP+pp3IiA9IpcR+SriMuIyTsFnqAdBuU5oTgjVBqR7xKMfzTQTIi5ewff3AkpoTwh+azZ5wViCVQjUTU3T1KS25dqsBefomtpUSbPl3DHGqYPqsJRtzOcFm6tEBZUKZG6FtFSMIkLlt1CdgR5hdKI1cAO7pa9jArgHwCEwt7OkoxC9TgX1tJgUpYDdQIPd3E6VRpVZVnotfjV7BbqRqmrQOYJH3BnI9nHj7EkNaucc8/mC5cEBR0fnaNsVR0dHHB4eTDm1iEMV+t4URvshsToz9tzJyQnXrp/Qdx1nqzO6Muld+VBkykwbxHtvAQ+TYA4wwdM6rdSjK5ZMYId1PffgvBHXHR+n7BJZuR/xv4HTPwYyGhLZQ6gr5osGE6b8NCE0qLoixzCK41jQjT4uiOLchwlhjRNYzO5i1rY0TeDgYM7BwYKjwwNuu+UmlgcLPrM84M8vXODMOV7olRdvVshmzfrGFdZn1xFMP88L5JSpvBDERq0MRw/E3hNL27yqTEAzZRsESKqQv4WUfgzV2xB5L879AcIa596F8/eC3ozom0BfNd1MpqHSsdlsEM00lTCrayDTti19P5v0xEVstU8xE4dYYF6rqs3bZrTZLlYoco2q+kOk+tjjxtmTu1I7R9M2tK1NfM9mSxaLOYvFbBJ/HHHjGE2oZrMduH66oesjN05vcLYyjse229IPRkL05gdhuVvwJsAughvZebIr7iw4S9CWgLafl8Augb5/A0yrODvlOnEO0as49y7ElXkMtcZDCDNCWJR0QxEJ5OymJ8w5gxg+a+Kjtjc49zlCuAvnHLPZjMWsoa5r5rOW2axhuZxzfP6Io8NDVkdHfODSJR52jsXZCS97aE6lCdFMt10jKNk7M/kE++z8NAAAjkEdLhsEGHxxaGA3qa16K5p/kKwvxLkBeAdwDZFP4NzHQZ8G6VuBV6IKKeUpsPuux4tQ+YoQGkCpqpqqbor9ni/n0266VNyHd+jTKJOx+0DOCOG9uOoPHjfOnuTJlzUhfIKq+jJVnSzlGInuzviQo7HRKKk7DAOb7Ybt1lRJDfEYpgbLiFyMVm6PgORg11yxf03fH1voWS2XLhWjBduYZzN+jCXT3uq990z7vzGuws5bA2L8u5bijMjEAtVbQRcF130QkW6qUb0zbonJD9swQBU8XQjc3TRUbct9bUuetXjnqJrGxMsrPzU87HksUL0TquLlKGLSBlDhuYKXTwMbgj/BO4dqyZkFVG4gcieiPSJfRmQo70YRE1Cw9zylHbuUcCemb4KOIoUdmTPZ2TwjLuL8wNAPOO+Nbjz56Iz05F3xL4z8lUdcyq85nuT04z4Wi/+ZxTKyXHa0swWzWUvV2Kj/uELnnOi6LcPQc3a25vLlhzlbbei2W1argngU3ToRMf7w3s0h7MNeuxQDLFdW2QW0nfBxXbCC0VqD2Kpdzl7MI/9aSqY6er6Mn43zqwi1c1SVacqN41P2t4s3Yb6dlP4Rml+E4z049xuYiY8rKIEZFrWtTfcs2pZl23L/csHbzh/z0PljTs+dI91yiUPvOLjyAAcHc7wOZnTk7IZqmsrSuhBYlHlGm/Azimrff5SuuxfVRE73kpPNYna9I2cBuQuf/y+EBSIPIO4GILi96ZY0BXBZob3x2rtui0Opg9C3NU4KKzJFNNmUunSObuhxVaAZejZDpN/2FtzRhHhyMZHNSfECwQWa2lb+xzqe3KCWM+r6z6nrC1T1U6nrAt956/KlON71qTDoirL/esXZ2Zph6Nn2veGseQ9vHlfqPUkq2Vux91fu/SOzw02dFCkBN66o+2V9gZjY+aEY5CflAbZtjgRXpBjfiyl17ia77b+sh6i+FNXXgF4DZuV1FsloVwQYywptH4FNVfOZtuUL8zmzxYyD5YLWCfW8pWkq3Fhsl/c8diLrqmI2a6irYA4MWHMq+Pvx/kvkDEMfiIMN7Y6mSC5fwclDhvGLK40Yptt6GvgoN+44gGGUhUT0cTdKJ0zi9apKGrJR5Up3UUXoUypDHXmafGL6G5SGqaEof2eCWpxMM4Z1XVFVViyMW1bf2VYUU2S9WtH3HavVivVmY1MsKZpiEHbRRv/ySVZXjROSRe3EpnHKZbfiTqdi10AEKApI7OC20vXfz7/HgNaCWY9woV3Qm4jpW8g6ox8eott+FSfWdMEZ2akufo/oGs0fA+3w/k6qCkQWeEeZZHHUVUVVGlA+WKBddMqrc+K5ceBaHLgv9qhz5GjELlIsvHBP8J7F3KQj6iqwXMyovPkjSnHVDUGoa0/O0G0cvXcEPzD0DQJEn4BoNYAzUhgKQ1GDZQ76LEg3C3kD+X4ln+XdDjpKqJUpmJSSySjnPfqvZupNSyYz5DLtrjsRHlWZehUi1k01I6O/I0HtnGd5sGSxnDObtTYq5YRhMOXQ1eqMzXrNEAfOTk/pui3XTzdcvX7C2aqb8FERs32r/C7lsGAb/WEw88+hNzirrDIirugzjysqjwhqtIR2druboARyEp3mKlOZAE/FXCmlxJCeSzf8c1J+CqTfQ/tfwct16taGgx3CrG2Y0yLuKt79NuKWeHdGXSkiFyYPbueExWxG21qHtanM/+TZTnl56qmHLe/uG36jW3PNOdJ2Q9xuoe/xTmlba6tfPH/E0cEBVfDM2prgnbnqplhWQJsUzwnOzpTt2qQpHEq3rehjIlQdMY07kZtISykOcB7yj0H6AYifh/TrSvq4sbHxggSbak9xADJ917PdmNd5F7fENFBvG/COpmtR78lVZec5qS0uWUgRYm/t+b4bEL8dOwpf93iSC0UIVTDWXfATLj1im8NgDlhDNI5A10W6zkgw/TBY0SS+wDvj5PRua0fZTYPvTYePhU/5rSlfHovKERfZF6jXUobuZikfXTiOK4mWyeolKX8LOT+TyPuJ2ZGdEpLtm+J2lh/ORUJ1H855W5VDQKQhl1VoTB1sqNZNN/MSeEZOHKbEXSkSUkTVlcFl41NbL8BG0prahnOr8nXwjpxcwchLr1Q8KSlDB6myE1CFQK4qFCEEIyBp8Z9RwVRcqaCq0Nsc+gJMR2+uezvhXle2pBCjZWBKmTgofVTEmRyG8w6oyo3wqJBVYwFKFnKyj8erFJ/UoPbOsZjNDX8tKYjhlIbfbjdrTk9vMPQLTk6+i836Jlbbe+j6PyOmM5sNFG/+IsV7ZAT2YUw/YiHwpx3zy8OOQknpGOpOWV/3TqLusAzDQUZd5NGfMRtdMpt4YtREzImU1D6yQ+ILEH4K764ym91FFb5EqEyTuSluBm3TFhVVIZQ8uOs6uu0GFEvPgnFiqroy+zeB9XaNeOHMwaat2Thhc3qd9eqMer0i52iWG8GEbxbzGcELVTAUJIkSC/jui7RCzhB7u3G9F4a+xrmMC0KSMoicxHxXc8C5b0Xk2cAlVJ9DLhPkY6A556iqakoxfQig+25hh6i+iJRuJ6UTUvocOV810ZqCUZvfpSeVhc8WrENyej0pXigX6//+unH2pKcfB4sDlvMFs7alqir6zqa4h35gfXrG9Wsn9N0hVx5+Pauz19Cl97LpP8aQ7rERKqkmqCpUYXrD46o5cqxz0cRTVSrERq8KtDd11Zzlayolj4YpTwZLU0oGv4PjSrFjnuLG2BuyTdsYR96j+WXk/By8P+FQf5OqfoCmgfPnz3N4dERVVSwWS8PlMf08VUu/zs5OyTkxolZ1ZUFd1RWIcrY6pRs6rqeBMy+snLC6do2zGyc0mzUpmhNZXVlOfXQwZ6S9CkoU6/OrQl27klMLOY4ODp6UtnivhGh6DvbelKFXcl6w3b4B+GlgjuZZaYmzgyS9L97wDXVlSJDmwpV2jpyOyflHiPH78fHTpPSviekyVbagFleK48qTtAjcZ4DzxPhfIP1/Vu6fvwNBPW6r+ytrzqn4rSSGITD0c/p+yTCcYxgukPQQqIy6uQfTjUwaa4xo4QoUMZq0KzYeoZNX/l+MlxlDfPxqarQw5tJ7qMX0XOUXUcAXzFlRnaNqOG+u5qSqQXxFrubTc/rgywpmvo/mSWM3m9ECBvp+S05uhGVKmlZSNSe2S6SEi5FZ3zMXCNEm59Fsq2TwluJNEGcZuuWRBCQ7jZaaueKam5PtHCk4slqRagt7JvmM8awPMIGdGaoR1TSduHGw17qGvjgjFJbkRBILmHDRzcADWBue6WpMH0IR7hz/LaBFAeSxU+onP6jrpiZr5uzMMM+z0xU3Tk8Z+oYrV17CyclzicN5uv5pZM0471i0c9QdUHkLCpuGznSd7ZkjF8T8vW2VHi/Ybm7QCsghDkUISXaxCTBhyVpgJ0s5xp9PopGPqMxvIefvJqenktJzGOJFQ4C/TUivE1IQVnd01H95gzg4UjR7urZpuXjxIsvlEqtXLa25dq1FsBzZktRMXVecOzxkuZgTvKdtbKd6Tt/zow8+yEozz9isOZo3+EqoauHoYEFTVzR1IKXR6CiXv5NsMEAFcnE2Uyl+OQ5PILa16ZRUCXUms9Z1JowwOLPWiGlgiIF+ULpeCdEG2rxzVHXFYrFguVziJRNcNptmsN0tp1KxFMaf89MMap74CEXkWIQqCG2j4K4Swrvw7jM8XlQ/6ZBeXdc2rXJ2RkqZaycnXLlylWE45vr1F7M6/UmyBlK0oHJemM9nuGq5E4ABcsz0vaEb2+2Wru8s0IpCjS83gHe2EttYVzYVplxUU10ouhmwXwKmaUUGKet6Vp0w1t1Y1E3k/CZyfiUpBWKsSU6QF4H7OUeqYPNLHeEjp+gQbD7PeZqm5cL5C5w7PrdrSeeMc55YCPVoBE3UVeDo4IDlfG75d0F/vmWz5uLZdUu1+g25raEJLBctkjPeQRW8aeVhHUCxxHTqdI4DB+DwUhl0KJ6Yais2YwJxxJQQ6YgpUuhYxBRtyDlCPwhNArAVuq4q5vM5i8UCdEBzb+dWjEk4nsdxw52aZiK786vGifE46gBNDSoneP82RP6QvztBXT7GKfCYErEIOg7DQI6OrA2mw/Ew6s7w7iGcG6ZtlBF1mIq2MdUYV1XZO1myxxDDCsQpnYCR0K7l+fYykemYOuyqj3wjgDVeGlRn5JmQzjlSI7ibFFmCBk+6eEy8/XYGoK+qyYB025la0VgUjeSmUKwfxpk+cY5UJk1cqQNEgJSoU8YXkc3oPeBw2bwQXXnNRv8s2DSFNTfVFjq9FxEQh4lA7lF4TeByRzfdwar753mUmGDvZwVqzeWq7+c9REQuA3cj3I9Itzf8ML6eXRrjCgJkKUyPyPpx4+ybIBCZSbFnvTqjHwbObpyyunGDGAND35WV5DJV9Ts49zGkOsFVD0DwMA6Yksk6kPJQeAJlNZKioSEGadVNKGqlnuArEOsITsZAxROFkqoYKjJ2IvcCfzoe3a00qCklpX+BsP7PYbgVmmcLs4UDN6d7/RvhGbcRr17hvg9+gO1dn+PayQniPOfOnWPethwdHBizcEjMZnNDbvotcehQ4MbpirPVCi9C401mTAuEB+Z5U4dy06cIOaE5kfqOmAY7b7EzBEIMBRmLwqpyE0Vg7IlWgQnG7It7SCgwoaoVcE0TcE1gaCPaZNpGp8IzVB7nxfwfdZSJHM+bQ+Q64n4X59+H82d491WDakseriImT9w04GrqPtD0joyJ3T+SFvy1xzchqK3b13Vbuq5ju1mz2axJcUZOA6IZJ9ep/HsJ4Y+galB/gDrDTcljqzmVoFYrVMoS60btmSCF4BMKOX3kaORSmNnXxsiz7qQApTbca5Pv2H07eHs3kGpNGIi3Kd0PCv1zzAFLnaDSEr/128nf/gLSl+/h6l9+hnj5Mt22Yz5fsl5vODo4QFRo6hoRaJpxJKroR6fEutuS4kBwQhuMeecdVF4KJGgehqBodGiO5AiDmhRFTgO536I5TbxsEXPadTJaxJWV24wS8Bmyt5a9qjEJvXNkL4Vk5cwttxJSpVS1EirzfPShBPTeQjH2BEQc4lY4eT/OdTiZ4eUizs2mG04LulWFgGooha81xKJYdfB3Jv1QzcRhsI8YC657M+gzgfMgN2NehRhF0hWsoqQXJhIeLZDiLcR4AVU3mSM52aDuPlsJ2G2fMjUbpCAlpRAseLPsNXMeQX4qTRzF4D2VHQ4L4P2aEO4kRE/wF/Du6Xg3m9rvJhdiGqepnrO9/Rn4F38bTuEkZdLVa6SYqItdns1qBhjb/aWBFKM5kWUxqa/sxAJaHepMMUmn9OwR+8nu31+H/7KryZSMebZkTUzUrLLzKYYZV7XdDHX9AE3zCRwL6q9cIn7siPBlwZ2Op7h0WmPZNcYZSXYScDvByj1FzJIO2fUfNQwzzn0F5/6CrA/g5GohnT328aQGdUqZs7NTVqs1m82GYXAM8dVkfgjlHOKebqKGU+6G8TiyqfrnOBD7jpwruu2r6LY/Ss7zKQXw/m4q/6vAhws5aMfbHYtAzWbimcsqFrNtyUFr06LeC2qziC5cBl8MdARC8Ji/8oO07a+j0jLUr6GRX0B4Gl6wYEPI6iFCPLhI/oEf5sZ3vpKTL9zF5vd+l9nn7+L43DE3rp8ya2ccHi45Pj601SpZChFTZL02hmJwQvRWLNaVJzfBFE/rQHBht9qy4144Z7w8l0YVJJnwMlVr9Rus2JeULpOztbWdQF0VHWxvmtgxQde/jzjcQZ8vIW/7cdx7X0s7BNwNB05IpXgPzuEk40nGTtRRxHnnpy6MchWPDOixD6HSE8LbCOH9ZN2i+iWYQNmvfzyRafIW+HNMBTwAv6Oq/6OIPBN4M3AB+AjwM6r6uHKUqrqTB4uRGAM5PQXV14Aclb+3uyDjCpo1lWFO07rI2RPjUxjiq9F8SBE2RbiIqhHIR1zzERrNY8pR7DFStk5mFodY23EqNm0lt91BKatz4V5Luel8WOHDVwlpS/Dn8LLC606i11ZCAfXkek5+1nOQZz2LAcElpXn4CjkpbTNnPusQB7NZUzqMVtDlOKoZDWTBOoBOEM1UDrRgypq1jGkxnbfpPOoO29+hvjv8HWxXotA8KQEohQorKlTlcT5l2uZemuYEN9zG5ovfw7AyKNGdF5iVuiVGYhzwZcfNU0CPN96+fvYj4m2HiHjB5Yh3X8TJKWD9XX38hfoJrdQd8AZVPSv6H+8RkT8G/mvg/1DVN4vIvwH+CfCvH++JVDPbrmOIsWz3SlXfzWz2TrLOpxa0yAnK7eT8w2S9BnwB5LSc7N2FGHehcUhWOAReAdQoJ6jeA7qaWrhjHm2eJzqNk49IiRvn88pIWY6KpDE4bZy/bB/Fy08I4qicp3rQU/1HId8J7pnAC4CmBJgDE6C2nSAeX2TzXa8injuPOzigvnCBpgr0167QX75MlTNNHahDQHUHXY5zfubfKGYShGmdeMlF9myk0Sag8E1EgRY0TTf7eMRYdAfTACmSFYbRSUGCyfJSoERf4XOmndXM5i2+z7TbTxHjjKq6B+RyoZ4ObLsNTpS68rja7dUvO8SWibaLAAAgAElEQVRlglHLNXEFzbKbM09dVcrqTV6i+XnkfGt5/NdX5Xgi0+QKnJV/VuVDgbFXCqal9z/xVwR1SokbN26UlU+QSpm5D1HVn0NVTF86Dmh+Cin+fYb4T8nuI2T5l+Cu7ulW2NLsKCSbKXu8BeVnyfwYWT9Ezv+GnE6wgLZATTEypIFRwWk0Gtpvu48ChFkUidEKy3IjSDYITlTwOPOPCZnmLk/7RUFq0J+A/HQsqN2eX2SuyDmQn/pM4k/9HLJec1Z7rreBkCKHb/lDjj/0IerVGcfHRxwdLHFjU8Q5XGU7ii/stz4lJBm1c9gqTmxsKxR4rQ5CqGocFb6tcRTXhmQrcYyD6Q7mROrX5Li1jNd5VBw+1ITKG1riAiJ1kUCOxKz0fSSmP0Z5O951OLlKTom+33J2dp3Yb5jPW4Kb241DMUgtkzBTQKvZZ0s2Dr3g0ZwNNMDel43PXSTnv09M3/s3C2oAMXb4R4BnA/8S+AJwooalAXwVE438eo+dZMeOW1+A/HGYUvH+BJGrZZXuUB2KvMEC1Wehei9gHoElf7DnHf8vslOylxZ4asm37isYsjH/NY9sO8H04EbUxLbEEZtl6kJKwU7Zqeczdh3HlyOIWIfTrSvc9VIHXMH0z/egWVGB5CAqOSzItz0NyKiHXIEfOmS5xHU99WZL1dSFT+3w0uAqKbtK4Y+XDqdgCqxCsWsuHSpf1u3xXNus4m50bdzCc1kdU4rkOEzFME5LCqKGX7sy+KCUedKA6kAIDxHCWYGhR0TIbpjBCSlVu2ZVOXuPTjp2gjvjiS476Hidx66w1BZm+jz4mmfZHU8oqNWu/ktE5Bzwe4xCxk/ssZPs2NPO1TpaOYyBI5qtoMLhxVl+K9dJvB3lK4h8CXFXUeeRoJADmgNOTb0pj9BsZm+olOJ/ElFi4YUIqhX98GKG/sWgHSJ/AXwWoMBJbpp0pnS3UjJfmNFyGhHDkZMyxPP0w2vph0v0T30x6fuOyEeKvkpMg3oH0MJ1cO8H+Rxwi8ArHdxqSXCfFQjos19I9/d+nHByjfSFO+m//CVqJwyHB8xnDXnWspzPESn63Iztb9gPlhHZMUTIznUWu3GnvqlqucHdroCTyE67FXtuVUTH4s4WBu+hqsyZuK4DTWPcDT/ySLwUhes8Qa/lRdmI20in9bYIjHOpuTgNj7SBkbPixs8yjt49/vHXQj9U9URE/gx4FXBOREJZrZ8K3PtXPV6Aauw4Fe3inC2gVTLZjUF9FS+/A+JRpzifwQejnAKqJojinBRdECXFsXq2SNJsQZ0Llm3DA46hfzl99wsg1/Fhg/d3YoOyhvf64nE+OuBmDCEYvRWtjR6JAwzxItv+J+iH76R/VkP6hSX5mSAzxc3Hd1yOa+B/X5DfB14u8HTgaZa/bmMmq7B68Uvh6c/FX71C9yv/itUHP0ArQh4G4sESspKPjxFXmVuAZtstwLqEsiNdTZzyKOCksN/2FkLGr42ENQ7TmpSVIUbWWi9LTgl8wWYoq8pSoKap6Puaac6+FLMqNodoxfgwJmCWjqkrnpelPtBMztFouynaQle44bb5mMCnliL88Z3Jnxj6cQkYSkDPgO8D/jfgz4AfxxCQn+MJaOkJhfs8kYzGE5t3xR8gknBug7qISiDLHEqTwAoGwbktzp0AmSwNIhVIRthg/k2nULIj1QrVBs0H5HyM6gXQCs3HZDlXCG6+vA6bah+bBYZJZ1IpJFUp3BBltMlTvWTFTbA6Ad9j9TUoLdDa16PUbzcgpxu4PiBJzRFBFQ0V6dyxdfIOz9EvljgVOpaFudiSkm3vrnQ8dyiP22HAY5pUVmylIDnsZizH4nlCQ0oejsg4P7HD90s+zB7jcfz5KCm2L0E2Dm/sxuzGOfy9v/iItvgeC7LcqOPfkelvCTYJv0LkGn/T9ONW4NdLXu2A31LVt4jIZ4A3i8j/AnwME5F83MN7x8HhIX3Xsd5sysiVwVX7NFEzwawRarNCE0iYApJXRWVA8/vJeYuTC2h4I+h3InIZJ29D5E6cfBnHA2UV+1Y0vZGcL6DpJeQUUJ3T999P0tupqwcQeTd9fzdVVZPnc5wPqFKURg118JU3yGuISErW+XKWNlVfUNpfU+L5RPyujxBf/x/RpZJ5HVlfCecq/JvAPV9w8X6a974d/+578DnhYyR7z/DyV9C/9nXI4QHxdW/k7OIluofBvW/G+gs1283DHCw+TRquUVeeWcGp67pi1gQL6hSRotOW40CXjTMSJ+MluyEsWIUQbFDDVQ1Cj6qQtLjmOsg6TIX5WFsMxbkga8J5oaotHTJcWmjbhuViYbl3mU2kNGCcN7k2nx2KUWotzc9l1Kwo1upO9Cd4R1N5ortOyn+AuE8/bpw9EfTjk5jQ+qO//0Xg5X/V4/cP5z2LhfGP82plNm9dz3ZrMlLB+0mWqqlrQvD0OZuvYc6WMyugkew+QvAfIcst5Hwrml+GcBUvb8HxJ7aKiGKko+ej6WfQfBuaFE2Q85xt91r64bvpq0/i/SeJwx3UbYN4hw9VUV41gR3nFa8W1DaFruQS0E6EcDc0d2V8iDDcwfCqXyMvM1EPSfIy5NCjP+Bw3wvVhx/E/69vofrwB/A5EVKP1rVpyL361XDumPiq17J62SsJn1f0M8LZZSHGT3Lh+POg15nPGiq3RCqP84GmnZlWx9DBIAWF6ElDRMgMjL6GQj1Ok4ibRuqCrwmutimYWKZcgKFg1ipaAlMsoEun0JVUxImjLiNobdswn82tmIw9edhYwKJIae374IFkPut70N7ov6iaSyqkeG9cEOfOSPkdIG//mwX13/axoxiWAVk9JOvNtnD4U9teiuaaL9PiLlkVPhbQxmhLJU1Z492DEO5GuJfgVnhJtrK6gEjAyQB6lZwDppY/kJLHDcfAEliQ822k9HRSbBmGA1QDsMG5FSLjhHTZcidcu8f7B/D+bjRnqpxwaSDmh4ANloM+hHA34hpLnSqB6itovI5utpATko3k765cIdx9D3r+FDk4hMUBegDxFpCnQ9cs2XATzfoGThJ9E0EHUuGLTKXitLXvFeR72/5+y3xKYRi3+An8n4q7IoiwV2AyPWhi0j2CUeem5lnW0R9nb0q8XMVHph6j3PDe2Bzj2xlTQcW5iHORxzue9CEB7z0iFP8PiPHVxPifIiSq+o/w4f8jeGjbGU1T44aBTc6kWJgDZed03hX10jXBvw3Ru3B+xay9TAiX8M4RfIUTz9B/ha7/FXJq2XYd221PjEfotR8k5Zfj3E0Mw0+R8hsZhoohtjgPbfsBZou34fwNXAhFksHcu1Q9woPM2t8k+LeRm8R8FkkhweJLrJxpTzt5F07uQXDmTAY4vUrKd0Mai6uAS0L1gQ/ir11FL1wg/9CbyK9/A3qzZ/OTmc2rFb3nVr76zjdx7Z7rHB/eQe7eRdtcx2lm1gSCc3jR6RyHOuC95aVOd3yOMaRNHlpxkql8xovBghYyY147pg6Y4mjJnccVPlUBCvxZlWmbUJm9hfNClyPr7YacImnoSDlOUzpjvp41otmRY5lKco6olBqAspPYEEGtIGFk1X99ochvwkptp9Q6T46UnkPOfw87lZ/DuQ/gg1LVNXXdkJAC7CfropXnsZTAAT1BPoGTTxFCYLFcUtcHeOepfYUTxzBcoeveTUqZarslhC39cAur9QvZbF8KHBDTq5EspOSJqcK5jOoNfPV2vB/M/LKcTOds5Ak9oa7ei5OM+bAO5EpZN6WBI+Dk0wif2oPf1KDI7EDd5BggGcJdn4e77oQLFxle+GLiG95AOnIMr1RiBPnAMVfe9TLWlxWNHcv6HcTZKYt5w9AfQPC4MG0l1qSR0sBAJh71tOKOwudSAr+E8W4lH/sCAhOxzLDosfsavLecXIwL7qapeSvqs2a6oSfHaNIME8Fp3FcM+ZHS4TUtciHh0FE8R4xFiZphqD5aMvZRx5Mb1DrK845Tmhnn7sf5jyKS8P7ypA40bZ3O8q+gofAwdOpIjfBdCIHgKqoycdE0JrHVVG0J6oF625FSJoSK4AN9X7HdPEBKdxgsWBh7yHnQZ5BzU5ytzDjYJWeeg7CTX1BDY1UK/OScDfi6EiID+Acc4TLmHZkNHtOvHpCfdjs5HSBiWteSM/6Be/Ff/TL0Pe6eLxE+9EHcckm69ano0TEcQf98O3XbeMy2fyGyuUS3ifTdgOYBL4HgPILatHpwOJRQAjcXe4+smTRYB08xnkYUs65IyZPy2K0dTYOksBgFhyuCjVosow2/H1ssWZWYdl7wcRisU5gtqG3Zz6W5g20GaqnPeFNl8agr6PseMubE4f8KoPpJDeqsyqaIPFqOmvHhfTRyNyLQNJepG28KnMEjXvDqjSwegnmVD8OUe+VkfIi2mVNXJo5z8cIlFvMFVVUzny3wPjD0A92mI8XMer1ivVrTD462+RCHyzsYhsjp2Yqu6xnSy9n2/5isT2Hoe7rtGu9XBmmJneQhmXaFKQwlMqYtUlU11FZUqiq6Eao/rZm/tYJe0WwjWsNTnsf6+36a+NQX7PLGoWf2H36f+f/7m7jVivqtb0U+/lHy7U9Hfvbnia9+DfoM5cbPZriekb94HvM/+Hnm11a01Z+zmL2Lpu5whwtqNyMEz6JtaaqAd9B4a4xsu471ZkWKiWHb0203CAM+r9DqDEXoc0VSB1LcvWQk/QeQwn3JzopmtZsSHUXSFZHItutwAuvNhvVmTY4DHjUheAzdkLHtmkYsUKyzI5BdIPlgKafYbaSlM+r+pj6Kf5vHeOfmSRA949y9wFcs3w5zvDcngVGySpwzC7KSDBpQb8WOtdodVVVTNw1N2zKfz1kul1RVw2J5YKtyF6mrjhQT3gecCyVgH0Dzdbqup+9P0LxFOUD1rDRsbKVWjfi4s7LLhfCT8uhzriDOtkg/rjgZBsHfU1F9sIEt5NyjmkivOE9+xrcxvPK7LBgAui3VJz6OhgBDxH3pi/i7v0h+7sOEH3qTpV6HSvdCs49eXz5mwyVk07PZfI5uOyC6JcUatEEwkn1TN1ajVG4S9Bn67v9n7l2Dbcuu+r7fmI+19t7ncR/9UHerpQZJLSEQIGMhJCNc4lkgMJjgKBhcuBJSqeRDiqp8iJ2vqUqVU5WHU6m4IA62qWCwcRJAtiJAIAOChIdAENATqdVSq1ut27f7vs7Ze6+15pwjH8aca+17JXW3iNKt3bX7nHvOPvvss/ZYY43xH//x/89TmJwToonkJjwjKq4a3IflPeBwxmAB5mTJ1G624avP2cbuIrNybckTKhZwttne/M4aAGBYtzVMtVATV8f2y5qXm9vaL3x7YYO6FPbbLXkc8YJZDM/dsmPVd/O2ODkbhzorMiiuCE69LdM7xQfb/nDOcbRZs+pX9H3HpbVjs4JrnfKBVeYsCDnCtIpoCQwnjnFaE6aJu48c993o2W53jDJRbinTNjLthTELOX0lef/dOP80SR4l86k6ADAIyyn4On1zKvgxQRHWn4CLvwWJSHz0VWh+GN04xldn0v1KeviVlMuXjRL6hBI+BHJd8B9yUDwaPEmqvcfZOekPft805IpCyogq45WXc+2NX8t26+luTfQ3B1b7EdcrYe3o8MTsoXi8GhXFCYzZk6SjOI90iW6tmMXujixjrfyDIR4a8OoRPF5s48aaQvD1hC4lIzmx+CRCROhrCTl6Tx86clv+qPV80erAy8IKQQSpFtrqA+qtCc1Ql3WPGdNXkcsD9Qf+2eeNsxd4SSBxdt2mQVEEiYHVasVqtbptwbLRLacxkZND9p6QheBNscg7YbUKrNcR7x3Hxys2644YAxcuBNZr+GRX+N+OMh8NgAuIX5kNtIuo77grJf7W0xf4xpu3uHn9JlsZSU8p+xsrdjcd2+Jw4xvw06twcp3N+E/Z7D9K8CNHqw2r2NtVosJZmhK6G1AKF/7IkT/hSdqx+/SbGaYfIt3Ts/1+Yf/tgq7X6N2mMhQ+KPQ/KbhPOsItD6Wj9J3V7aroteuMv/AL5H/9ryvJR/Ei7L7le3j8b9+Pv3gXu3ft2f7iGZu0R1cXcUeBvgRk7Jh8Z340yRrGUiJFNqhX/Dqylg1aRso4kaaMKqTKqQku4NSa7Uigl4p41LIjl4KkxDgN8/gdhaiw8r1JHnQ9Y78x9VPqlU0LUxXwVJi392d0zInx273V8ykls5orp2zHv8k4fRuWrf/Z542zF778mCacM7EVqdJSXYwVv7bHlaqxW3LGtos8km3hNOLrAmpkHS2oNzGwjoHYeVZB6CxBcdUXnvAOF8B33pQ7Y4d0K3zK6LTnCNO6jquI6zzivV02VRB3gewv4eQmUe9iSgFKQqMggapSbdoUh3yLeEPot+BVmG6eMOh9aFhT7upID9Y3TRwyqNnUPZrxnxgRlytyYXO/UgplmtDPPIFm400QjH5adleZ7k7IvcruqLBNGYbMfoIxC64IU703yWLbMBeMQVMQF3DBTIuKBKMksODd1QAOI89ahq7l8xzBXos5m7Wgxiw5fC0PvbiqPV4HVlQvRgp5Bg0tezuqkGc9Pk5qYy6Gkhf15HKZVB7k/+uY/Et2U4VpSrYlUclDZtkbWBT/q/ZGFfNWNTYXdUwdOgvk2HXErjO739jhY7Sath6IBwr8YII303i5GZyRdpxkNjnzcCqUZC5RKWdSLqRcyBPkDLwGeKONkMOffRP+gwMlX2Oc/oTgP0Fwjugj4BHn8bHDabHNa+9AE333R5QVxPEuyu++CX/t4Zl7oQr+zz6Gv/l7iF6F8nsUzoBkpLliVniNH81LH0Df/Cb0JffhN68nvPMUh2P1QcdaPevO03nTn5YilHFPkmyBKhWn1lwRCFMhHYcRygh5AjUfdie2hOBE6g6kmlLpmBAgT4U8WbM+5UyuWzLOBVuNiz1dZ5qBYTfU+GvLAFVIPSdKTjPDw25tp1Tm4ZsdK48RRW/h3a/j3bNz517wmnqcRgtk7+sYOsxaFyba3S5tmSlnHLad7MQRu0DsTZK2W/V0q97I/V2sSqpmR1ZU+IoCfzfV5sUXXDkMahORvDUVtlMhJbtPOTOlQhohT4K+Tij/iUPWa/xPvRX3Z99Enh6jD/8dzn2YLnhWMVit6E0qGFWCs7voiPbvxenvkvcP4d59if7dD1MUUp1BMH4Idv8IeAR0pGQTfqE0nrguRKOHXk76sR+jfP1fIbyrJ/6PG/wTI5tjz+bEs+kcnQOvGZehjDumMlrZIca4oxREExQ1EaD9AExEN+BdsSGRs2xuGzaWKadSyMMECnlKtVTRuazwUD3kA6Fb0a/W9nncLpPDUoO5SgGnugGFq2Q1lsmkd5bwAFRzHXbdwPtfwudfetY4e2GDGmhi5Y0B18bmBlsvoH+7XC0MLVksy7xbMn39XA4ODAgBYaPV9LOoWdApuFKQulC7r3rQ8xi3FNQregH0CMpdgt4F9EJZHVF0Qy5n5NLdJkFWDQjmEbTUgUdBcbLD+xHyKWH7DGV8xjDgZAMX5WmUp4GnmW035kl1/c/GjhTXXKoEN4646wP+mQHnznHHdezcRs0CWlwVwFSa4y9StftoXI46MRRbpWuad228PoMN7f1QDhwEZkDS/u6qm+daeSXLjuh8DValqEP1uA5yMqJ7dJ5jLjOKNqizMbyzE1POEHnWVdgXfqJYquRXjItEbQi2pW22vnXqVRXlTcPDE5xJw67WPSF41psVm6M13gurPtBFXwc19SSpeLKUgjqtojXOLOWSMqbErVtn3Lhxk1s3b7I9P2e/3TK8bGB6m5IuCrxO4AjIQlZlrNrIQyr4MVHU08cCzsIx1gC0+tUYhXiHi5Hiz3H8Kn38ODnDMOY68PkIqVyr65J1t1AE5/x8grbFinLlSfK/+jnKb/4G/iOK3ym+SwT3MTTdIk/KuN+x355RYmDl10RncmKbvjOtjpLQPFnWTCPj3sqClDJZrd+JXVzKgMbVadPHAzYlYCWHk+XK603KzbRZ6oKFmvVyKmoZP19knL6NcXod4h/Dx3ch8gkAvGtX8UiIscbCkghDCKTc5sqfnwPyggd1y75NAbQJsMMyQtfDTCDUjByIXaRfGXtvte5Zr1cmW9t5YnAz4ciamdpo1pOlSEZxuAySlHFKnJ+dc+vWGWdn5+y2W4bdjvH+kfS3IL/SmQpCFLgJWYWpGMNvTIr32QYVWXFZCTWzzC1WlUFzTvAxorrDu/dS8u+QcsY5s9RLOcE0VDqEVk0LC2pXs10TYNennyK/8x3WyiXFTwWJ4KSj5J4yBaZxx34X0RzIvYdoTMJ139F3wZYmsrH4hp1JGBRKnR8kvIdQxdelzgFsc8a2WBp0Vx9QSWP2+kxp1ZiWWq9UdkUTw76LmkV0OSalbyelv4nT9yH+/Tj3SH0+S2DBB2ItP5pbl3FAAt63YP4yCerbmFztEqcHfIT5gB2ys9zn3H0rReqlqW3SiBysdFE77qLYDLNYgZhhmBK73Z7tdstupwzDSxjHi0zlfkrs0AhyJrjrIDcK7sY1vFzDuU+D3JwbvaaMj0glLJlQuUrDYJdAF3ItTQpOJrRtldTyQqtWZRt0uHopD67p0iVIVTr36Ai9eNHkgseeNKxwCuNYmMYBR2vGzIe82ddpLhQ8WapEG0siybkgkucgatm4IQ1Sywzn2qusJ18V2mwlhyqWTWtJ1+qpxVEAinpUO4rGeRRuh+Gg/BA7OdzB126Xvfj8txdFyrfruvlShTC7wOZcZm6IEzdf0mLX0YW+GiB1pgHXdXSxMz5v8IRQD7zYQcylULJdMoeU2Y6TCS2qkNSxH0Yee/zTPHHlKrfO7uezV36IazdexfbmA4zTfaTk6N8n9P8G3NN7uo/9GjH8nzh3Dec+NGfu3ZBISSsxqVJ/fKKEmkXEQ3Waddo2U9Qma/aCwbmaoSukhdRAqfJgzja6S1ZbsFJH+rqvY/yBf4d8ejfn74b8b4Q43iTor8P4+xytIyedo3dA79n0gaNNj5ZAyeYIvD87I3grLXKaGIY9OQdC6BExFqR30VawxNtgBGrwNlPRiHdhRnRQk+vdnY9kLez2u8qRaetlxjfJKZNSxpNrM7/g1KHpa3uPCvic8ZWSarqIzx62L3hQx2iXllmTGJkzQ8llUftsXbA3p6lDpGT+o5t3ondz6WI3I8rYwq3tF+73ZsgzZGVIym4/8PQz13jq6lXOt/dz49Y3cnb+1xn2JmVbikMehfBr4K8kov8AnfslxO3sNWM6elPKlUXZUF1FQkZIVgrV2hSoLDmtrLm6xCrMkmYifg5kV43pm1uXwWyOVKw+1Ze/DP3O76Lc/RDjIxN5SMSzK5x3f0rvbkGOjMMpOZneRxdN21qLq1YXiRhCg8UNbZomVFk836vfZBtjW3klRpTyFb4LNahVmSZrunNR9oNJ/xrP56BPqo5prTGXA6MdoRqaugODKjm4uutypb5duPP224sS1K2GLlraOh3NVaB12K7WaKGeue3yabRGP5cvTcbAnh8qRsRCQrdMNE3GThvGxHZM7Paw3b6M7fnL2fpXsXvVRXbBMT3sSCszAm1TLlfH8S4ERH1TzrLBQbHxgRjIYjivuw9fHgAXoQTL1uzR8gmUpyjlBC0PgZ4gehWRTyLsaGr+0gLItQ1Cq3XK8THTfa8kHx2RXvkw2ptJpnPtauVnT0onmFZJMu3CcRgYop8VUZvfoVblVyduhlqFJvzTmsO2+V+51JXn0n4m+GAlWCrz+zol09luPoq5OIo+hOpF20DSu2fKx3z2N+HDg/eUO4L3+ZQgL6yRkfecnJwQY6zlRp7tmg0qKrO0boyRSKTvVqxWPTH09JUbMmdqV3m7B3+fLaAuq1wipu282+4Yp4kb2y3Xz7dsdxd58sm38eST38T2oVOu/uAruPU1Eb0slLsdqMkK+AAhVrQmrC24pkRJNuKdSiJlaEa5aKSTN+LC3wG9hFZnbdVPU8o/pugTlPIAJf0HlPK1IL9FcD8N8jihKn3Oi7RiGalUQ6bpwVdz9iN/l+nhV8PlS8jpKVIyMTg2q0DIPX0XiNHjPZQ8MQ47dlvHzevXSeN+zsylZMbdjjyNlDwRvKt0BV/9Eq3cGceESMaJKceKq4pUscd5b4SpEElpYpwSBRO72e52jOPEmCaGaaKUDVP6TlL6PnI5oXD/bCQlPuB8xPmABG8OqVJxxPr/lpkt2Xme7fa8g7ou3r4PeFxVv+8vo6VnipldFW1szYkF9eE2OUjNONU3ZM7Wfv76nSLdS4vVvrBAe6btlucyxOiQJ2y3D3J+/lfZSWT3ip79G9y8ciWlHnBH3Yy2XUA0oKn6e9cmFDWlJwtqpeSXoPoG4B5mK2i9q26yF7Qco+WrUX0TIk8gYkqp3pm8ri0YtAaOuR7V42Omr/pqxtd/A45CJCG7UssARwyOUDP1Ij1gTMNxHBY3XVc1TVLVuK6SZsGF+WSiZmuzGrEzwSTIsGaxXTGrYlRRa5ZbY97YeSZTkS1Tl5dTyhtR7Q2XlkUbXCoD0PBumf/+hVSyxFBTnf1Cty8mU/8E8CHgtP77v+aL1NJb1rlkpp+WbLzo1uVygHjMjWKMC/znq1XxzBWx4tkkf1ttyuzR0phjWRe6aErFJocZSnFkdWQnZm3RBNwMJJlXnoo3U1C0rhm1069KE+DaqCTMDdXhTTlG9c1VzPLVdvkttQZ3bfOEuZyaYc1yTEqvJk33MU0Po+UiWpT4mcc4+YsPEG7uuOvJE+554ALdeMbF04njzYNEr2xioAtWKgzjiKouI3wUFYhdj1dPlFIrfQE1oZpWX6s2bpEzhKboAbWuZdKFkmscGCvHjMJbaQ91UFXYIvInOPdJvP8EPjyz+GpCTRY2IGuNZyrLckmThftCt+crO/Yg8L3AfwX8Z2Ip8IvW0hMRutiZemnKVZ7KzmhoY9a2A2cNYVcNLnoEYysAACAASURBVLtuVX35zI/EN5VSWm2e5xOC2vgU1fmeVUmqjKkwjIlhzExJSNlszZI4UpAZWhOFSSEnxSUleRN0tzKpyuLCzFEWsZ08oatk+jaKaxnnLlTfjpbvpZSeXC7TFoCCGnHeizVKgtibmOugYvghxvGtTENHTqdozqw+9AHu+l//Z9ZXnuTlpy/jK1/zEH1wXDjacbz6Gkqe2J/dZBrN+H6727Pb7YjB03eWWBRYbTZAscu/Mynf/S4ZdaBKWJRSCN5G987rnJCEJZGa+LySsiWOdk9ZmZLV8Ckb+qRcQ9z/TpB3EkIidiNd3xFiVcaCuZlUVcY0MeXabyF4+dKgH/8Q+M+Bk/rvu/hLaOndcxRnbPrQBfX2DeM2Gm73g81ydzCCPSg05hFuLUEs/y2Y94EES510HV7ValNUEYgm5KKyZCC94/Gtw5mTVQ3g20fDrSZMLN3wReBizYZ+fu2o4eu0V99+VNUguHIPOX8FJSkMCfYZd+Oc7onH6T/7OEed5/TIFpXXmxXd+oQ8jWavkQ3Xztn4yyIQ8qIb6H1ApOB9wXmQVHCuzK/ncBCmy0G87dZKpLm5VD04Zi1zt0kkdkzkCiKPGFPQneBcV9/7+nej89W2lM/dLn+22/NRaPo+4Iqq/pGIvPW5Hn/n7VBL71V3r7VN0Q7r6FZjN8Zea/4OA9NeS7vb1wxtqCbsuZJjqppmVtuTy0UxgypB1OFdJIbO7tETIvho5dz8mmlZSMkKrr1BYiSjJqI4DwnasMR5nBhrzwaKe0T+EOFPQJKVjKqIvBTkLag+YOeIc3aXxa3g8IxtlFYeV+IvCuEPHJvHOo71hKPTC1y4dIm77r6bvOr586M1j6xXHA97vqYk7nfCNI3st9n8vnHGL3euNr/RBlZsgT1OMnufIZW5tlekshDDwqys4/BS6+4pZcZpYhgnpimZC3AzT211dZ0Q23tt8GWIwZ43hipXxjyzmHKq+45pHo07WSjKX+j2fDL1NwPfLyJvw/SzToH/gb+Elp5NmirwXrclDvkFwVv9vDy+5Vi5LaANk1YLHMXguoqHKgZF2Gi2NTsg6hA8wQVi6OliR+xc9Sqp1E7sd7Su3HT0zLGq1EC7A2CqgVzHu8HkbsnetqLZ4fy/Rdw/QdjTFESFN4I8BPKA1ajOzfK5sJw083Go+5jyKeh+HlwQNhd6ju895fjCjkuX7+Kee+/l5qrnT06O+T/Wa16223L3uOOVKPvdlv1uV/cpncnyek+3OuJofQRS0OQoxeFkxLuBZugJthUv3hNirBBrxFdjqKyYcE4yuHQYJsa6aFtKJpXEWL3Tc/WqbMvUPlTuT4yELs4afAWbSI6jxUjSZFK/Irjw3DX1s3/XAuu/UNUHVfUrgB8G3qOqP8qipQfPU0uP2sUfutFC635lFkCZmV0Vs9aDy9lBCqvPqAeXuGbonivIv1z2GjLSrNBmBGUBGuZp33LfA08DVzFxmnbpa+UGy9dmjNxuhXbJ3gLPoPIMyDVEnkHkaZCrIE8Bt6AJP7Qz97ZbBm4CT0G6jtxKyDWBc7GRf5Mmr6WaF0f0dXnZVTqsWANof9Us5Wjfc43h6Gio+NIPLB9cO24V9ZB6DOfyTSuB6WDT3vqZFVouo3oZI4lfBZ7BrOPc/DyHqEt772fPnXJQArV4+f8Jp/57fJFaeqUo4zDaH18vJ76xwsTNCAeqpJLQg7G5nQxN0GY5kFrM8HMYB5rlRamMPHXBDClzsdLBmZxY36/JuiLGQPCO4MBL1SGaY7og8sd4/2uE8DTefxgDNSJ4axYb/NRiWuuQxEoee6I4d5Nu9kOHzyDun+N4DyKPANeZIciK6kgJNoThBj78MiH8IcqroPwNlJcxFWW7m0AGbp5tuXHjJmUYeYtzfIVzbPYjD6XCpLbFk6gfVZhqgpgQJnGL5URu1IJGlpT5xOi6js16QwiR1WZDv97YdXEYSVMiFyOJ7aeR/TiyG0emqTBO30jK30bRNcg14L9F3E18+LDx4+NSDoqAmQzbdDlNU2X6GZnKOQcVx3+22xcr5fubwG/Wz79oLT0bpU63kWUandDV1a7gK+aZWYK5cZ1LzdgtO9RMnHNiSuOMpuScLQMFNbWfBjPUoI6dEHNf2W+G3R4gefZaKYh8CO9/Dueu4N0K73pUTV64ZXip6Ef9AyumrEx1dC6e2zxX7Aeewrl3onqYdSyjzlcr87qnlDOcfw/OjyjfQpE3ozxEUmU3TKiMnJ/vuXVryzoVvqFf8dauI02Ja1nZKTPfJeHscxOrtq+JYc9SQHLlYpflquicx6HEGFn1K0Ls6PsV3WpFKcowZVSTJaxkY/G5tk4wpdeQy4+g2iHuvwf5GZAB5/taegSCjwQX0apnbSVjVR6oFOTCoTLulzCovxS3+Sxr3RiVY83SaduYXAA/49pt3at5kIuXapG2YJciOgvN2FhXkVq3ex9MrciZzNZcRrQmkNtREvt/FSKXhLjF7zzXR85d//yzTZi8Oc1WnsZ8uW/BqxRaD+DIxYYeORumK9Iw35r9JSMugd5C5FGMYP84Ke9IaWI/DJxvtxQtrFc9q+CZxoFhP7AfJqYpz1c5asmBs13MhlS4YsdqLtfUjmkb3jV+c0tArdxIU2IYBtNMqc7FuXLm7XjdAvkUEEGeARkxNatAqYKbuWTbTJqPaw1e53AFVHRODDOv+1luL4qWXjsgqAX0NE3kg0zmnNB1Pc4Jfb8idhEfPblkzs5uIcCqj6z6Zj9jQjLmO75jGEecD3R1idO7wHrdkxWmsmM/Gbldi8yBNKky1WebSxCRuY703uGjg6IkCpptJJywTZLgvSkHiZF6mlOsCxDmet6Ct4ngpFLqeN08v007zs1Z3YZICs6aqlI+jer/Alwk5Suc7x5nnEauXL1K10X6ruPateucHG8oObPbnTNNY+V7YMEcIq43bZUsnn0qSMmEMeEGa+JrP2dNdd/hnOf4+ITjkwu2YiXOBljTyPXr17l27Tq7/Z5r16+z3e1R1YqOCMj7Sfrf1HLto1VcXxizLRUnVdw+EFMyA9JgjSkOYhcpWqpwTkKQ2oB+OQU1LAyrWhE0gRvnnFEMvUdkoR/G2Jlak3PmlLsfKt5aTKKsArvOB6RCeFPOeIRQtKrQe7qwIgMhJMSNuKrXUTK1oalT7jte8dxY+qp+L43NVOaTU1CKgO3RLZNLpW591OrYtaZNzEra8PJ2WRVkyiDJmjJfR/YKOGcjeq5R5L2AkotjGAMpO26enfH0M9eIMTIOe87P1va7k9Wk3kk1Q/LgA+I7JHgKjinXVbdcTLAnW0Ft75WvC7SBVb+mX6/x3hvalDLTlNlud9y8eZNhHNlud+yGAe+90SFESPopJH+0rpDZzbb1belZgbFOO32pcGNLbsEj6kxIU/zBXOPLKKjvvDWK6Rcq/BsQn3IGWTTgmt9fqWIodW3EstpcPxxWyMw1PJ8LVBizDgjtURX/tgnfwUaLasWZl/LYlTmx0CwmnHuC0P0BiuDDExXGq4qtKKWcovogNjp/CjO83FkpUJTibKtbtApSikP9Urqpms50QwxKgSnbRHVKmSllZqSp2DJtUW5zMytayzcMe08pU6apUkfLXIYUNU3wrGbuClL9LPM8tZ176zbAqsO1plra2ujD43/bgK32EbMGgxOkLOM1+zub/uLnmf7ccXsRFm9vh/PabVn2vD0cSykM48iUFM1T9Q+BVBJTDva4Utf1U55VM+fR4EHrt0QsB5FtYoc9YiYWtdYXhZXI7IvtncyqoV6E4u21erHhjHOKq/bHMf4+2n0SRAjuSZN40GAk/1zI+eUU/Q8pvA7lN1D9KdDHrNHNCVeM2dcktnwwTxVViF01FVIzGBIRUilsh5GYCyHGCtFh369/UKgnSca4LqKOoh7UtPEYJjjfUlSZil1JxFXzJmeDrHHK+KK1dk6M40gqxWpzEdQ4uhQxhEVsalB7CuslTE64mqsebjPVherG0BMBXxEuY9tUR7UZqvnCtxdnR/GLfLzBejJf8hdsutTmoQpGHkhfzSezLEE9/+ZDb+sa/Japa0ap3/ZUcfWDkw0sU7sa5Kbnt3BAIOPcZ/HhigEujRaAQ8n1NR6j+lXAN4I+iupq/lstwy24fMtkbp6duyp7W0zoBytj0mGmznVhua6K2fO2DGr3UpvkxgfQbJna/oIl27bXMU81ReYNpTyvfbXAbcFbMfH2XrTDTM3Q87E8yNSu9RH14+Gj5kzdctGXE/pxkKFbiDmWXUUrj9sltgZ0sWkhld8l1bU1JStHBDFRlEphTdlGtGTHlBLGwzFNvqIw7AeG/ciwH5lGI9nkrKaLRy0rTHiTakcylx6zI69qfawcwIEmFCOqOG/LDQjQBgi6KOqrXsHJexD3GCp/gPd7IB7ww+tiQr0a+1np02p5agPZLCdymkiTmZiWJpUstoNYD+ocLG1aVzBUoeQC04gOIwyj/cI6YhUWLovBsQlJhXE0U9FxNA718nfNb3O156tDqDkGl9dz+1JAVYKS5T4L7FMhvIpuPVeWhheh/GjjTmCexLWN5dsJQTpn6TEPZDVd4uAs+/kp43xTDJrQbAE9TEZML0olyyiZRGIiq3J+vuX8bFubmsQwKSkZHzpSnzsYDh39PK8zjDwZgOu10UQt4EQEKRmS4SdejCehCGkaK22yLUUUhEfx7p8AK0R2mJbFanm/kcp7bhvUxis5AJCN/ZZHGyOPnskHCIG0aji9oMH2CsWUNDHKrzJOGcmFpAaukQbYbpHtDuc8cWMnmAnbGIms5MKwN7r8bj+yGwbjp1fIMKUyzxBQ5n4na6E0slaDUZ3MsKKxqCpGLwtPvp4T888Znd9oul9WNTUs5cedzeHthP8lU1u2brVduwwzs7ekNUR3iNIUsYYHClkhVQ2KlDJpMoJM8UpZgXbUYYfgiuIGcBnc9Lmv/dAvsF1ZrPQwNMWOd6RxJtBIKc1ht3XuipMrNAVVcb5epwJSAUA0oNo+j6DeutJg3BZ1juZWW7zZ5c2Mx4amyOGd+bJtdtQWcEkVUjZfnVxlebRlR5mfB2pWxzL9VK+UZT7my4xhbhw5KD+4rTc/KCHsd9yWze+MmbnRPGg4n+X2Agf1wR8uC5lp+e78qHnxM+XCMBZyEbpghCSkEpWq6tI0TUzjaFl9rMMGD6ZO7xlzYT8VUlFu3Dzjxq0zdrFn+OYRvU/Q+x3+ISEE8I9C93vgrkL/ISHszL7CNctnVURKLQ+E4AwlUX0FRb4J1QuU8kFy+mNbbZreSBq/FlVPTnYievcEEt6Lk8ctsAEhktPrSOkbgTWTLpf/4Kvux4OF/JaM3qe2PjXsYZpwH/kw/Qf/HJcnvFP6zth0674jVvJ97Cp1V6hli1nHDdMEeaQvmVgXM0xoqK/EpYBznlIg5dGE83c7znd7UkrshoFhnEyLsJTqcrvodmctdd/TyhCpTbjTQvNsbwOsUssWWwS2krHtruYqtWC+il9GmbqNtrV2wCp3oiD1z1MomiELKcE4ClMG6Oi7iNZJWPujxzEx7FtQG601u0JO1tjsxsTZLpFS5vrNW1y7eYvhnhXDWyfK20zwxXVC8BA/Jax+VgkfgT5ACLVRbMKK1BJFTBKhM511ij5MSj+OykOU8rOk6U8pZNL415jGfx/VVdNlRPwfIOHjePdYa8mASEqvZ9r/R2i5RJXNQIBYX0P5amV8W6L8Fa2k+wnZ7Yi/8PMcffjPCZPpfvedCcFsVj1drBIHdW/BEFGj/U7Dnv1+j+SJLhe8C5UpaX47C8XUM6bEMJoAz/lux9n51oJ6t2cYR3JFbnKFVnOVNDCFJnuPS9vwUTHvlnr1qn5cJn5TLPgbUjQHdeUKNQmMZ7u9KOWHNQKLzMvyzeUTLRVtKq3zlrnTPnx4Q0HuZP+ZbJnURtL0JlJSUl6Rc0dOl9FxhQwgo+DO7KRzVwf8jVu4W3vc+hZyVJhZmBUSbHVvyxwt04qeYA64l9FyT0U7LlPKRbT0NaBA3SUo94C7d4G76NH1ZfToAoVTk9fNtdcMAg7KSW0Od/aRnGEfccm8bWzH0fYU/cHdXu/hddCOT9PdE21UgroHeiBRsNAJtG7ilHlhuiEgbY1rQTyW7Ku6RnVTf/UWZHubFt9tP3MAJBw+ZmFaao2fL7OgtgmcVBDt9pqr1MZQYFbbTBOMkzAlzGwSqcqmTZ/CSOhjJTTZ2n+5beXKFm73TCmw338Tu/2bmK5eQt/xdYT3Gx+aKjQen/gY/bV34OMTRP8IwZ3bJbsGgmAik66y+iRXTkhORgoqAZ3ewJR+wjDf/etJY6hXKWxal15K0h9Fw7ctJ0vwTG/6atJbI2Wd7CqU7Q00lQVFzqF7N/DLMk9lpQgnjwcuuJ7VRjk9WnF60oycAtHXsXzLermQqyyC5gmKeYGvVitOXMSHQL8+wnd9Vb0ywco0FYb9wJQyu92O7facnG2G0LbIjXLaFCTs+Bd9Ayl9B4pH3LsR99s4l/DFBkyUUtEqMcEatdM8ZzNXstds3GxLKIsC1xe6vTg1df18bgoOJmWlBkkjB6UEaRKmbDwNrWy2dolrA4spTZXskusZ7ubslJK5v46TYxhexzj+CNPuGH4rEiY3Y7ZFIYbHiKtfIoQPEr25fonIXOrM1nFiikuurkhJKTWoHUW/mlxeY1O40ZEG1y4roJDLvSR5m20gqBGJtFfya4T87wnlckaTiaGLU0os5vr1O0L8JY+8T2jWdU6EzWXP0d2R1apwtOk43vR454jeBOFzNs2/XAqUBHW1i2KOWU6g73o2K8vSYbXG+WiJ5uB9GUfbJx32A/v9nlzq1+r+YBNRL9goXPEUfS25/DCqHaKfQfS9aKWTenVWc5cq5ClygO4k0mQsPZ2D2oZcX1ZB3RANgFn9Z76k2AO0PtAkZq3sKCq1/Ci3NcfzYm2penZtWEBtSFr3DgaRilbZsEfweoqUl0C6dNiug9ibrW6yH9JKmqolDUSUu9B8ajWgNFbdg5RiNsj2XL6iCODa4KDK6HoH0TeFqhuIXkHCiD9XwhOg5/YyUINyu6jWxN709H2Hv+DmgZPIwIXjLSdHK1Yrx2bVs+rsRKx4jNWs1fdbS7vXZeU6lfTe2WrXgTBl47a0BLKUHWVZytBmvcdSHhqwX98joZSAao9wP8JrETlDyzVUz5eyowZ1qcCBHpSTrRQB43p/+ZUfpe6+UWbctJlOtqGBycxOlJIZsjBMnikLqaxalBjvINtu4jBM7IexDkyWLYq2UdFUhIpmuu69rFcfZ0r3sh/ezjC8lQZjUWTO9lLhqnoUjemWJ9CLJP0Bir7FihtnDWNJ95HGy6Z+VD0eHRClGO6NEiTjRAnBsVp1VvO6DxPDv0D8k7iPOvw/9Ujv6KMJ93gnrLpIDJ7VWc/FBy/SXeoYx4n9foeWxMnxFU5P7id2yqXLlzg9PUFLYdjtSZNpe5RxR06V21F3/zRNuJLwzrPqTzk+Oa3Ysh2OlIohGyWz2+/Z73aWqYeRYbQsOmWTLzhEjy1bW+uXszImq62d+y68ezWqnyHmXyDn/xuRQk5WchbBRvi08qOaHdUyc+HcfJkFNXN2rv88aO5KG/9qqRhoItWyw2SrClVef4b0cq7WFlMyPDZW80hZdh+dk2qNlvH+43ThA8BL8e6vMXue19fVMrLOXV19vTWzqXZo/lqKfjeeRe9Nk/VtWkpdImimQ6Uq9CvRZ7wUkyKOkRCF4J+ki7+Dcx8nPhXorkW8c2zWa1Yrs5jYdCu6GNmsN9xz9z2s7lux3+85Ozsn58xmvWKzPiGEwPHJEZvNygj244hOVhppnsjTYBm2yQ1UuzmHWTD3/YqiRva3RVdbep2qVELjS7dFjFwnvjOxrGLbS6NY6ggfVAOB1yJ8FSKPUsrvzCCAHe+F8AXMs4c5i9cYcWbW8KwlyIvH/dDP/fphxm4ehja48AePr4hGqV4tybDpYRhoZkjqAwFBvTILl4ujOCV6Txc7INZJnaInCg9BuQh64254/E2wfwnCM3j3WZyMuOApGix65SNI+U2cODrfvFHuJvAKVNfE8Bm68CjINBv6OIE+CtHb0ul6ZQaeFtQbnLubLhon2jvHem02eoNzPN71XPeB1XrNXZcvsup7TrYdd6Mc5UwXA32sGtFCrfWbv4sZktoI3xwOgmuXcxu/d8Fqb6jKTblUaqmRlsw2xDK2CdJYVIlWBKuWWQ2iXUqQlnxAi9QrWN3Ib4F/mOIbqrSMEg3Jlvo75gn0s8fYi8T9ULvWiMxTMaRuidQzdJpGGzFrpGhA64q3UTKhpMI0WOY4P99y89Yt4zuXI7ouokEJvq8bMp7oA4JDe4eTjiFsuHUeiF7hZQo/rpS/CuX3Xw0//RPw6Dlefo0+/DzeXUOdGhlEd5B/Gcq/xTnovY3KKd8Cq/8Y9KW41R/h1z+LczdYrczfMXjP0XpN33UEH1j1vQ1V/C26eA9OLrJa9axXJiS/Xq3ou57HxPErPvA7zpx/T05P6bvIm2/e4t+92nPvNNaoMR5M5yCUjJaEKyOUPVIGpEy4MlmG8y0IBYjEUAXO1aNFGYeJYbJR+Pn5GeM0sRtGhilVxKnRbxtnxzRctWLTyBLoxh4whViLWsOhs6qhHU7QOc/LHLhtCqpUsUtXA7oOvZ7t9nwVmh5lWXtOqvoGEbkM/EvgK4BHgber6rXnEdkWoO301KVRbKyuom3snSn45dK2nMKWUcqSVcYpEYon5YzPDu9KqybqKLvxKQIhKEVDNRctuGOBhwW+Abh6AV1dqkHyYZzrLGjxtVTJqHwSymRclGB8FPQBnNsBhdBfJa4/gPfPsNmsWW82BB84Pj62PT/vWXV91VqOxLjGuWPWq571em1B3a/ou46nxfGM83xczJHs5OSYPkYeLpm4PWczOso0ktOCeIuaI5k1gi1TlxlTb9JmTcMu+OpgizH4TJrN+pVWdjSn4tnht70lNaxnoprqnHEb4FNqa1LUGsfGS7JsfUeAyh0fLQDnmYbxn74EQV1v36qqVw/+/feB31DVfyAif7/+++8957McTBBbx4zasGW+ZM2LtsXEyJ0Y6aV9FPNgmVJiTIFx/AaG/SVyOKOLH8HpU4h6UrQ1KfMRXzakjbU24MP7iF1Cb17Gvefrkcdegnzw08iNPwN3DR/eT+wg+I05XlFHgjlYUDthHQTbQLqB099G5EP0Rx9jfXpCiIHT0xOOj0/Yx8gjpxe4sl7XTRzb9XN4nNi2RxfDslnvrJx4EniiXtzuSRNvvHWLl4jw2hs3SM88zfVxpKSJnKxR7rqOEKOJQg7bqtI04Si2NR8cMYZK3DKNb+8CMfh63HN1zpqYxnqfUi0FTSB9bsi1kF3lkgjLtFhve4dnROq24fEM/smSr+Tgn7UnUrRumNvXvZcvaVDfefsB4K3185/BtsyfO6jnZoyKKiyAu9blwNJcs1Km+FINMV11QTVvv1QKQxoZxwts99/Fdvtt1dvwH6H502hx9F0L6sXKTqnoiNsSul+hW/8q+sxr8T/7nyLcg+SPwPST4D+Oj4F+FYjhhM4VoqsTlDJAmYhOOAqe6ASRq3j/LxAXOT4VLly+TOw8l++6i0uXL/GZruddly7zm0fHB00saAZNOjc/M7Q5ZTQlxqJcnUYkZx4cJ95+4wZflzPT9evsn3ySK+NAzhbUToT1ek3f95hr8H5e6QpS8B66GFg1r5y+o+87Q4hyoJRMSYlpGpmGgXEYGPcDY0VAtM4AHNShjpA120JBa/JroNbOpwIAdXWtvv32SaEFtVWW9se3eHVGgbSIqRITLajdlyioFfg1MSzlp9SkxF6iqp+p338SeMnn+8FDLb1L/SFH9uAx8y85HJfOs4rbGGZLN9xG46B6hOpdqF6r2nNtdJ4pxc94521diSS8OyP4LXm8QLh+nbjfErobhKPPEMJjhHCJGO4mBk/0js7r/OahZkXdxxbUCe+fRpwQVqewuYh2HWzWsNmQu46bmzVXN+vZeFMEXDUkatu+DQ3KqQlcGqS1yonjNHHPOHDflLg+7Dkf9ozjSM6WraW6ZNke6IIetCbrUE3We0fwgRCilQpFFo2NprfSxuIzDeFweHZQQnIA9deU2j5y+PXDTK13pOeDBrE9s7RhTP14yDp8ttvzDeq3qOrjInIv8G4R+fDhN1VV5QuAh3qgpffgidcRkyuI7fK76nG9bX4M2y37/b5OCW3HzTvHqjONPa+J7c0biIjJaOUJ1Vus1v8Xp5euIvIULn6a4mDUiVu7M/xgU6vUYKFqDxxc4bjfEySh689y19G7KOUjxO4RVpuBEE44Oeq5dGrsvS6arZ0TwfuN1eMidHVq1zgoWeCTF0/46N0XGWNgvYmsYuKGg49MN9lvh3ZgAHhNznzLmLhYzOJ6mhIlF/a7HeNuNKx+GEgp8VApMGaulMLZ2Rm3zm9YaVD7DxFHlsBEs4FbI6EOoCq3s8TA6DvzaZeeoj2okocdebBN/Ovb83mJ9uZuzzilmbagDbarU919UYZcedrJjrVTweNAImMJJpms4EvTJXQUjaArjODTgQaTs2gD9izWDijYakM2ViQO9xzCYs8rqFX18frxioj8IiZi81kRuV9VPyMi9wNXnvN5gEEdvYvVEdXj+x6/WlFKYX+25fpuWIovcazFs+4CfRfJOXF24zptI8YMkCZWm9/Gx/dSdCLlW5QCoybG3U3LUm7RyvNeTCnAKyerzHEseP8k69U7CKEndo7NxuHDCasusu5Nm7lfBbp+hXPOnA26OL9GkFn7bUT5vUunvOOeS9zwvo6ZM4VGi7WsU+oE9OU58beniVeUzH47sD3fkVPm1tk55+c7Ss7sR9PUCGaXxGdx7PYDt8621bbOlgbEeZLrmEj4EDg6WhH7Fab+ZNZJxTmGKuxe6EnaoyUxjuekd9E7fwAAIABJREFU7Tn7YeDa2Tnb3Y7dMHFjOzClZI6+1e/SIIhgNNSsDMmRS2HK9tF5RxCPSGQs0XxqilhQFyrNocPIToJohxRvNFw1rxyDJNpYooDUoBbwdzaXX2xQi8gR4FT1Vv38u4D/EngHpqH3D3jeWnpLSVGf/LZ7Y2Sht8/3nTS/7GbO3sbONjoXN+B89UUuyZoL7SjlyDKALlcxaE1HxrszvNvhvdD3IzEUYtfR92uCD9a4Nfw3mFej854YA7GrBP7auPrc9OhgHSOnzqHOkIRGvdRsYiB6MAC5mBJH48iqZBj2lHEgp8Q0DPZ5zjCOhJRAVyQ9YaJjHG+S0rmtb9X9wxkfkrpF5Ez3zhYWbGhuENpyL7VNyJVYlEvzaDlUx6rDpIZLH76XB2Vioy1Qli30w+8f3qEtWjdZ5oP/DoDuw56z0ZLv6Dg/5/Z8MvVLgF+sdUwAfk5Vf0VE/hD4BRH5ceCTwNuf64lUqfuEgVIKrmHUtU4uxepH41tTazPT7QhVQ60hJrdp7BU1HBnmYUApX0lKfwMtD1LpcfXYZJBC8De5fPF3OT3+M3wQw42DKXuuq75FDJ6uIgWx6/CheaGviDGaVFrsauBgcBjw1s5xv8KQylyXqhqxv9E1p3Gk5MxDUyIPI0/lwn4/1LKqsNvt2e9tAjjV8XZKD7Lbfi8pvYyif0jJvwzcJMQVXVzhQ2C16tmsjOC/7jvD7BFKZYI0AXpU0ZyqamliGkeGoTaFOdXAzpgMWIYiRryqltDQFj4MI7dF4MqQLDAVB1JIaaLIMpBp7VFdPK93qQvOzqztVM2M1LrmmU8jYEJCzyFr+pxBraaZ9/Wf5+tPA9/+XD9/x0/NsFEL5MOPVHIMumxJoDprP9dnoHGyrY4UvM+U4jgUOdHyUkr+fkr+utowZhvblkQuib67gr/7Okcnj1l50ZsXTYhhNvQxYyHDcEOI+BBrg9XNWyE2yq79QYiIwOU88fo0gNqiQkllro2nNJGTUWFtlJ2Y9hPXcmEY9gyD9RTDfmIcp+UEVmW/v8iNG3+dYfh6vEsE/06cG/C+I1Y7vj52rHoj+PddnOm6udldlGIMQIA2IUzLCHyqJ97MT69Nucz7kUu2bFNgKiY+k6WgLgkpuaSDVa3l1nDyOaDrArPD15rD9MYPwRKFuu/45SRmUwF3WD62A6XqUX0A9AEOv2Xm7H4Zv9bb7ft37vZ/I4icA59AiIhcQOQujHHXOm3TWI4xmnVxiDU7h/nz4IRQg7oR6BfXXc/eOZ50nn21LvbeGsmLeeLuUvDFFlwp2WimuSC5kvutEKYk25Yv2cQt25Aj51R5zwvFVvUckUdxboUPTxIDOG+GqX1vJ1pXA9n7MOtqlJpctR7UxqWw5StHThbMU0pmFVevLqXofLVfysYWyDIjUXYCrCjlAXLZHLxLAdUHEMLnyrDcgU8viMYhPLZcESyRdRS9H8qF+tj3fd4we+G5H205FOaz3UqII3L+DnJ+I6rVXEUg50+Q0rvJ+XFaNWxTMQfOPmZfKsZt+m/qC+gjBP8/UbiA6FsRfhjlrvlN8s4kfY+Ojgy/7ft5P89wXI8X42pYVlkCuutWxBD5lPf8fOz5QKg2yTHSifCt057vHwZObqNqFkMxJpMz2G33NqkbJ/a7vQmXjyPjsLflgslwelUWnL18jBj/Md6f0nfX2GyUEC5zfHzKyekFvI+s1kd0qxUm3RZxztu+32SZtpREmmyhYpwyU3WgPbt5k932jJQy2/3AOE2m2lSo5UtdscOYjFpKHYC17P4yxunvMOWvpWHP4MjlZTiO28yryhkviF7Dp5FlOVjAVG/FTqAm5St6iaJvR+Sb7Yf5ns8bYy+O7NiMGVPPdkU1ovpaSvluzORY65vwPlR/H9XHbnuKhrk2y+fmiCquTqLkaUQ+Y4/jbtQNoC2bN72+UC3wZLGWjpG+b0FtI/AFG11wXuc8N33gD53nvdUxLMZAL3CvCN+REuuUSK3pmjOxBZIhIbYwPA7WEE7TyDBO9kYmo202gpFlq6t4f7WehBvW64vEuOHoaMPx8ZFBpf2KEPv576SOq1vm03oVMP3nkWEYyTmxHwb2w2goTs3WBoNSn6NmTZr8QTMrqkaj5YSU30DK31off1BzNAj6Dlh6eUxTEjjAoKWWmmBloxZgjerXI3z3lw9Lr9XDX+CbyyczcG/wl3ktNrX92z7Y57XZ8Jhwu4iJuXvn0SLAdZT306S9SlH61U3W62dmR7BmZ9c2t5veRdPROyx12m8/LYXXp0RQtfKjZKLAQ9sd6XzHtvKWm0beNFm5UaomndTGyMQfDesNoWoCmoRknbba9+xKYrX/arXh+PiUEAKboyNWvZl1uloCQSvxtFrEVQmyxk/OuV41RttMHwf247gMXBovw1WMWxrCIrU2V3LpSfmVTOlucnk1pVy2431HqWHHb0LkUUQeA54E+ezBY9qEuQXvYl2nZZlICueo/D/A6edyRg5uL5pA5Jz9bps6HdZPdbVVsaGEq76j9RLlZJFXcE6QzuGL4PyKUjr7eiXPOP8pfPgZRDqankUMyuXLA+t1j/OuZmyzuYuV/umlKenbL55JQBVKfCArPzbsObM/qL7pSv/MdYann2FKqdpPuKUhbvJdaib23itd7Ci+mP6H81ZTu0RyNn3IyRq21WrN5ct30a9W9P2azdGpvd5odbU1zxZwqiYaafrPbQpq+355Go0Ett+y3W6Z0sStszNunZ/beyFLVLoQ6vtT9fKw1z+lQi4n7McfZDd8J0WPSemllCKzBDZUGN+ByB7xv4pz/wrnzhC5QSOJNO61TVMbj/tQx0Urb+dpkH+O8M5nja0XVfUUWoVcb3NV0oQEAbVOPBer5+amwlVNubkkszrOSzAuMUKoTaMPW2J8ZC5ZzCHW0/enhHBUS4owlxUtU7ftGfu19jpnrFuVNcorMnMTa42tcmsYuL4fGHOePVKgIWG3X63sNVmZ4J2SnW0GadFZRoBa18ZgcON6vaHrV2w26/lEDCHUY1f3Bevvatoo88FtPcxc70+1USxMyV6X8zIf12bYau8VtUE0mnAuPbm8nFxeTymOorbv6axIqTVyxeJkRORx4E8x4fU10C1vfbuqHDSHd26dw4DyKM+SpIEXI6jveEH6OS8c4AYi78e5xxD3KMjVBXyvPyNUOkydf4BpX/tgwRicp4+Wcbsuslr1M0LRXHOPjzasV7X+dAuC0shGAE0cvqEq9jcsl/d5X08LqWaW3W7PdjdQcpqZcMgyZmgBpu25qiyvx9NZxUX0fg7Itla2Wm/o+94QjhCMtFjr5FTle9OUmSarhUspM7RWR5tWflDvJdnaXO6Y0hvZT/cj7gbR/SlenrQD6+t8oCJXhUgqr2ZMD5PzvUzplaTcJIIBUcQ/hovvx7lrtSFURLaI+wDUfc2CVoUoqTLE9tXaktaavvVbS0XaHI1r9HzeEHuBRddrXeqWuqtdzmasGhB5Cu//Jd7/uokcVnvh1lQCC/JTs4mrZJ6u6w2v7TpONsfEEFhvVhwfH+ODoRtdNO5DFyzLqdraUiNJNY9HLc1wSecMr2qiljjbMJ+qQGLKi9jLrbNzbt46p+RE7COhCwssKL6iN7U+FartsxourmE+ViBzdkeVvu/ZbNYmNON9Vb1djp8WTMxnzAsUR0NNpZ5L9TJfMfuSJvJ0yn76Hs7H78X7vwD/D8F91q5aVSyy2d7lsmYqb2U//Sg5HzOmE1Jqipp2Hkj4KL77Sbz/CzumzfjanZvNhxNzYVA76YZkeLkYwEgrP8zGuV1pDlGvFtWfH69+8TxfnvWWEHkakccx4cRLHF6q7HYnY8xuCwutjrNjpOt6+r63jNn1pFp/4j3Ze6SY5K0rTQVI5qtH0eVka6VPUcUVmZuZ0t7wOyC82fKuKMVVMo9YzlFdSplW3ojUPmBuIu0SJPXK1PDzxaJtwXa1Ev1mezbaeX/QNGq7MrZN7WJLuUUpZU3RC4geo8TZy1FbD0GzrHOonlLK/ZSyOQCylikwssO5K4h7fBncAIfG7zq/lmW8fvhbDr8/Q9Zz//XsMfTCBnVrtA66Q62wUNM6bvWxryaUoRlIhkiZSUw6b8fYgTQfksMhTEMKbDDRs1qtySHwx33H+/qOVN3AvPPcmzNv2cPLk2XdtumRpolpNIKVcy2YzMTUeUcpOg9LppSYBpPfQo2sr6Xgo036BNuTdC14K0+k8R8AvLi6DW/1c/BhNh514uZj0oK6zYvbyTefe/W5l2lcXWQuxVCO3ZZpmtidn7PbbhnTU6TyLnCfQt3TqHwKxVnerAE3lVzx6JEhJcakdSF6KQOdtTrmHR+szDN34zy/TuObLOhGkcVzcek3qiRD41zXr1WI+zlvL3ymbiXIQQZpypnahitS6amx3oNN/lKSWiIoRXOVBa6XvBJxlf/RGr/YRbquo+tX9Ks1+xB4/6rjp1c9W2fB6Z3ndSnxylJ4BTKPi8dxZKxcDC1lzv4iMvvQcIAhN+y51cFd7FBVs6X2oV46Z/Bmht2cLG+UkaZs5L7qe/qut8XYuCL4gGorjWxBWYurZXIzp6/H+PBqWEuQXHW7x3HPfnvOOI7studst+dM5ZyUfxV176n7YAZZzMLs9aQYp4lcJsaUGZO9D7nV0vXHxCnOi9n5hVAhuqV3MqKT3Jals2azAqlDHVDD96v2y1yWIjjlOfL0i9goHjaI9Qu31cnL+PvAW/DwvdKmjMoBsanc9rxzddJ+jyqdKieqhILJFyD/b3vvHnzZVd13ftbe5977+/26W2pJLSGhRoiHQAJshC1jXGBHBjvh4RhsQ2I8OImTVFKVccrzqJpUUlPz59RM1ZQzrprJTBG7Uo6DPSYEYhvztME2D0OELMC8JJBA6C11t9SP3733nLP3XvPHWvucc3/qblom7haq3tLt333fc85ee+21vmut72KnWIeqoYp9akqkjE4mxjq0puGQ62/WhVnzNIaDHUL3k/OC4XyG58E57GxHqPdLCBxvjH96VjI7WpgVKGUzqDK9fvgmPvRtcUL6PqcxFJ4KKe+j5B0XvFNIWBrywgLFF20xLpZUtknlUnK+hFL2G9Ix+UmRHuQ4It4lWJLr1v2oXu6OpGUMKhnVFapdVcKDmTGZMvtLAPaD7seQnV2KtmcV7AsC6Y0COR5azlV4DI6rIWszPSJN42mSfR40Vtd3/tlC02SLEC6MA65pOvq+I4rQNo1p3Kbhh0vmYMpkqc4aHMyJq5ZLTvZGELNcLum7nq6tmlq9W5bZuf0sEZs4sfvMDGk7Cz8brZfvRQOpjrEgWeLOhJMEY00VsLwTT46azxcs5nMeDYH3LxZ8sWm4oRTe3AWu9wBJTxkquIduBcWuUymFvrMCgnXbcfykVYWfOnmKJ46fIPWR1fLVtOsfJUsL8z+mmd1udeHBOuOWlOk7E7yufznr9RvI5RDr7qV0aT6co0SQ+AAS34c0X4XwIKk8SkDo8y305SdBD0AxyrjIEZL+IUFvBw1Uik3F3oJatzPDDnZAX4/qrQQ5Rir/GeUvOBPyARdQqIftqGq6mhPixxrdhGhiJEQhREESOF5ibTMqSjH5fN/3NE1jxOp9IoU4mBMxZ25U5aXO11ejbDln2tWSlWfPrdetmSHe9FKLa+hogphLJqZmEGrAFlnXucNTnG1k4vTVUqsYnKE0up3tLaRFaEJk1sws1dX9id0Y+eR8xvubhtfkwt8o0Ih1cCQ7o9Vw/tXpso4NXW8tlterNbsnd2m7jpO7u+zuLunTgn59I33/VooskcX9NLOvAM7opEqfC+u+pRSh7Z7Lev1mSrmGLkf67H5AdDAiHEXiRyF83KpiSoPq3DgFy9tQvQoptYztWxT9Elk/Z4vIt5wBpRYcdBRUF6C3gL6DzLeRcjvIbWeVrwsWfNkwPWT0dO2x+DY/bt3+qUn94djfvP6tuSDTXOuc8xBoAJC+H75Pi3FOW36zEYfnNOY74AKJk7CMnvu0yU9dVCOvH2pwmYg7hyGg6s3qh+8V56wbbW1CmLRtM/NhJoXn5MJLpXB9LiyqqeX0xLmMOSXGGmufy7l4folBjW3XsW47+v4AKV9DydtAIIQ7ISwJ8QTifgLuuBUuIZfnkvIWKV9HyluUEslFXEt3WJ/1Y0i4E5ET1IZDQzBFhVIiEBmaKGkcrkGd7yHawwhnjgiOXT9hm6LPh/IKf+Xzp5WtC6qpqwkiKqOjuIF+WJSvnmwp6t50GrSvqg5bdilK13bEGOliR9eZ7SVBvKWZcy9Ha1Datx3JO1KNzYYyJSt4F4KmmbtZJPV/QHxrHIsVcs6k3sPSqSV1awQlLbbIaeFIRiCGxvj3ggWAAlaHKVjzIjwymopCShzMgbeWwK1ROVgKV6WethS6tmO5Wg7C27bWHzwVS4RKObPcXdF1PcvViqPHn2DdJrr25azWb0PLpQQ+RzP/VTSu0O3jsL1tVG+q5NLTlxtYdW+jT9fRtteyWl+CFndrBSQ8QRP+E2H2MUI4jsRvIsEwa3PyPDEq4QEsu0VxW1nMpKu7oOUGeV68WAcxrYEfQPQQWt5OKK91SXrjaeXrwgg10yiiJ+yoTjQkQ/hYwghbjWmqoyNXTZaaFlqJBa12L9NUJ6lPhFJIYt11tSjr1Zq+85ySisZQSb4ZwuS6N8t9EtSoWjHncRdJfaJrrcA2SqQJRptmmDVDBDGE6KVqNb/EcWxxVCAXtgS+j0zwcHxtQ9GnRO/Bnq5NtK13B/Dq79q6out6VuuW5WrNuk2k/kpS/xpUL2PW3E5oPoHGFXF2FTq7zNJCW0M+sh6iz6+iSy+hS0LXm6atncOCrEG+hISPWLmXQ6tD4lpFqgqoWo1iUeuENqHeYWBlYtwN6+KeOqPKDpSbJw7y6cd5jyhu0MSWWmKlqC4R+TJN/AgxPkopD5FSD6qk0KDBsNgYLY+i8UqPaR6F6kg52/eep5xqzrAVhM68aTxq5VUppY1jrCZEjSiqH+PYy7BOglAraobfW6/tfrumXZumDh5EKUVZ9JmmqWyvhVJqT7viIEblAYYi1mTTFnWgCWJmUrKdIeVS6w5IRb0DWKHtkqWOpsxqbf1Y1l0/tJVGHoDwJ4gegPANQyIwhtOus9zqtutJbUfbZfqkRvCI2c+qYxmWhG3g+ynlDfbdyVrmIY8Q5GtACxipI9ojcjfCvSAPgT4AlTtRa8JSPf1RuEUSytdQPoRwEOXFINecVc7OlXbsIPDrwMv8Z/8hcCdPlXZM8OqQMAj0kKykxxD+gPn8k4TQk9IxVqsVTWygmDmiahlt2niUbo/WLo6K9H0/cF01seZNLwghsLNvh50dq85InZVWbTqsOphChi2b/phFXwwyXvQBhek61us1J06ctKT5dk27XiFgNnpWZrM5s/kWoTEy83kuBCkevfZrkartr8POE0OAuaCNael11/mWnugz5CJ0KbP2pP7d5YrVuvX7a7rWWvF1qXig5PNI8wAW034UpSMXWLeJVTEqhuWpFX3b0qeW1VrdHxWaxjVqrFl4B1H5OVJ6nTncnqYb4p8Qm/8LkW8DBdMha2L4MCK/g3AK1cddKXjzIg86WlcHXKAjsEbkgyh/jnI9yC+jXM3ZVPW5aupfAz6kqm8VkTmwA/wrniLtmDCmi1YNW50e6IAHCPEBy7EokZzN1sxldASHdFNPTqpar35nFXLBunaNreh0qEGczy3kPjVhpjh3jW7W6N7gwAw5BxPXVZ2qKxvEaIn/lvwvCH1vWlMkbTiwg6Osk5LgYlzNYMHCEhSNVlcYSxk0tKVlWgFt8Wr1lC0Y1ac0VK10XfKAySR9Ux5HwrEhSmd5INGjo8Vb8hnjad0NShHjHJz08rQIfmAowQMkeyKYfIug5ogyNBjNiDwCfBVYg87cfmZAbURGfHrU1gV4COR+YI1yDJHEdyXUInIp8GPAP3DB6YBORP5KtGN7WXZqAGWMV1SPePxMKUrwxJ2BX20PTDlGKn04VmuP8yD8fWcJ8SBDwv5YXeKs+bXdg1gXrgH58GT5GhHNyZxKdZuxBotisNZ4FgCfTJxXnYRg/H4xOC9dNWscbanQVrXD+5RRNW6R3ntA9n2mbS2pf7lqWS5bUk4sVx3LVWcOZN/T5+w0DWOiXuUUDKqeLae0fWKdO9t5Ut6ofCkowl0gd4DsUgk3N6fgEMgPAc9GuQ4tPwPhEVRfhaWZNii3AL9oWlq/QCn3Diacx3iYVqoPKkSC87bsIuFTHuA58zgXTf084DHg34nIy4HbgV/hHGnHNoYwRAdrso6hFh5LklGgq9ArWLYclt2n/uwUEtyoShk+V+G1+l5bEG1ruLeI4Pnog1M2ZumNnHHRYbdSoOTa+9FsR2MGdedPBSESRM1GDnM3X4IJveeJ9A4p5pRJEhzyGxe0BF88WvOJCy2JPhsV2bo157DteparNTkXdneXnDy1a/WFy5bV2msQnVPaBNmCHFZraFdInDEpUVilxK5aumzXJ7JnzmW/lip/AeFXQR5xR9eTaAft+nLgEPBs4MUU/WfW3El2QPb7fP4kqj+Clnsp+Vcp4R6KqPs11j9RxQotKq9ilRHjQDxBjL+HyB+dVczORagbjOT2n6vqZ0Xk1zBTYxiqZ6Ydm3LpHVyMfchP8756Z7i/Idha2TRHbFvc/hqT/2WisacmwgQ5ccEVjD+OijWXspF1ZwtGqFo+5uIJ/HVbDEPAo4Z6B33rx1OFevyMHb+Ounh6BYabVK9/MNFqwKm2fBtvKeXxlqsDOeLmNYnIDnHIgXOceYFqtIZLHrDJRchlm6JbGINSdKFdovIIIg8OO4pdW1uQyrWYYyjANuj24POFYRYPoHoAZIXVG24G4kYxrn8nSg6wDsHHCeHk6URtGOci1PcD96vqZ/3xezChPifaMZ1y6V3S6N7U07HyY5zmwY6tQq2171O06JVYF9j53JOGcqRkQ0bmC+uVYgRd4qbNiInnVEh9Nsgo1yT0qcDroKnFbyCDAEkIzBw/rySK6hCVhEiIEGVuuDjCfGthpWJNtAQqaTyn2tMwJTi7Z10MTqKpzlVSFLIRylgzztYRCjM1Us4s255VZwhPV6znuIqgYYaE2bDowAoOkkLRGdr/ECW9kqQdy/JnLMsXKeVKcn8rJT0f5YWoXmmf88CJirE91WoeipMPDb6B/VTRyTIdp3J8vS68YmkO4hvUOOc6mklSd3cZYxdnGedCZvOwiNwnIi9W1TsxApuv+O3v8xRox6oWq2dmqaYyBEOE8bmp8KtOM/gM041NZGafGEyHGKMx9c9m3irYbG9remSN362vttnaJVeyFmteVNGPPrmmLgXN9t7sWjCEQJkVGs9AM6TEzy5EgmILbmYdsmbzmdU8RiNYjxIdl65CLQYliNgCUq9nzKPDaumtBkGuVmt3BhPLVU8umdW6N7OkKF0u9H48xAa8WXI1w0rJJFVKWdCnV5G6f0oqJ1jmh1nmz6F6KSW9GS0/YWpB5oiMNMhKpTPzrmVAKcmPf3T06joK6o4lk9d0vJlguxnnQm2KqFCtavHGp8HThZvm7GJ7rujHPwfe5cjHPcAvYbvKU6IdO9uo2PVUoE/n31bbuebtgppz5ULdzCxNFVWCI/eSRwrh2s1VMDt3QDY27PmJF163xbpF7kEups76WA4mAytq1SxxUvtYTZJxgduCHX9jRGOG4E4pnlmX/ebmSA3PI8PCH2BHXzgVXagaNPln0oGGfr5jNGOPN5QTZpaoLoAdRua8vHcWNownm7seeAzk22Z66KUIM5RToLXI9hLQ/aPWngi3P9wc1QLc89vfaZwr6+nngVtO89JToh0bIk2TYZlq5vFPtfPpBNu46yxJfuav2uK2UGuMke3tbRoX6tp8szblUZT5fMF8bvS1XdvSdz2gnoCjxu9Hbx2tksNqWrdM6/VXm1nWqy6C05RFigS2ZpF9c6PTtWDPjBAjWwvn4PNsPb+2Zg5RczosC3Ht/WxyMdKZlAp96lnurhyjVrrenLmuQJGGEkw/xtpBz8vgcs50vgut+8xy3ZGbhvaHM+2tSl7B+vegfHJ0nsGSRKnRvgGZGvn4hmsvIPowlN9C9aMoN6P8LKrPAv0sWv4ApEd4E/A6Bjq+PfNrP1GfsRTVigBV76qaLWcbF6w719SRO52dfbpRc0IqvdeY0jkjRCtw3dnZMU1d1AkLdejgpcpQODDW9ZlNPLQyc1wbTKDHaB8jpDix0f3I/HisL8xsNrM+ib4Ia7XKfDEfiC6neH09hqEjmaoRNrpQrzvLf+69oVBKiVSUPpsGTgpFoimGaFt+TfBXNaajVIzut8v+faFn9eLC+megnID+C1A+gRFB4icsgsiTe4GbxjdcOUqt3HkckY+52j0O+lrgSrTciYb3AC3C9Qi31nUy9fkZ703UeD0SF+xN1OvMGvv8c+lVDwJ8p6wauWqEPQdbHT0drGrDf0OkaSz40syMF88Kbp07uiiardtUjJnqg8/mZp6oKiXlQaDqNp1ypqDEEsmSht7htWB4MDHcUBxKpxRitAveRMuLjjEM4fyhdtITftSRHKP1rTnZeSgP6/reAidZLYMwFee5c9PDEYvRLrXrkwfyl0LvuRZ93kfbPZdUFrSXdHTPWZP2L0hXX0NqIiUw1CISVojejXCHO3g1cHIf4In/NRfGdzBz7F34Q4HyOCJfBk4hsgJe7G981oYvMRRKSDXJqt8F6oXCFUCoeLk5lWc3QS5Ix1uYmBehwl+eSeeaEBjsVos4mkarWm42n7FYWOh7sdgeqsj3799nPU9KQR3FsH6AFg43obb6wVkzZz1b4QcEWB/zGtRIvZV01ZYVxW3+Js5cK0+EPCohWvnR9taM7W1nSHXH1WoMZ0Qnq+l76+ib+kTbtpb0n6zHeilmfnR9Gvp/52Qpsm3vtjRC9oXYZ6W4z4+bAAAgAElEQVRz57ZLxZOalDZZRXafbmC5fgdJX0D3isz672TyswL94evo5gtQmHkNJDyKyG8hfGCiiRUJDwG7WBRQNhpPDcib+zsqX0Pk11DdT5CbEf0VhEsReS4iMwb63mgOoPWAj4OZY55Sxf9HOgURU0ADW9cZxgWjHRucwrrqJ47W1GmsEb+avzwtrK1O2GIxZz5fMHNu6Q2hLlYVE6KZH3PX1Ca0Cc268fupT4hDeMGjLoPT5jM5reYe7yt1N6icfLHa1LOZefDSIBKxvus9JVfTqLNEqJRo+95s366n69IQBc0uqH0Z2yxX8zc7C1NN7Lc8j2LfkQt9f4BVezO93kx/BbSvVMph07u9b5xN3Yl0F5EvUoMrlg1rkmY7ZaWr8DmVsTJ+9IOOUOQRQ6r0BtAfBnkWoys4UQbOSx1rqweqbh7Np8HkEEVScd/hzOO8mx97y7hgFNR6f3i7A6zTnOVphcw0kli7NtVtfsjDFStvaqIFPGKMxMYgtdlsTp6XcUFhec4iwcPYxgI1JE05S5RRkcVBqEOw/OoK3c7nRtYeg3jR8JwBjiSMmr/mYXvPwpRMIxfPAalUC6k4na9W4kUZkAxFPRV1zNJb94m8bz+rG19Kd+hKUv5+2vYSskK6TCmfBt2XEO4hcg/y+KOE++83E8BRm6mjPk7JCFNotSJHX3LDdLDXqrHIBE7yOcczNhnTC+zL2MC8qxzYDi9kvEj3LOO8a2qrABcnhvHav8nKrWOIJqqabZkS8/mcnPOQ/1ETnJpGnFPO6hpns5n3YHT8O2Tw+7PZjNl85nS0JuT1AiMypJEa8XnHLM4mjKWOAE/YnMZuWEbFBcLOXNi3ZdpnNt9iNl+YXVg5OTzfIeXkDewtO673NNlS1Gl2Hd1w51Ul2JSJ9WlPakUJXUqsu56UC6dWa3Mmr7qa5c+8le41P0Yu++nLVbbr/Zmg/xY42iPyceby75F8guaJ47aTeAdfS7PV4aYT7M3OAYpgbaOxHVeCOdUjF98w8RuOIVikNRKdr9CKBLA9wKDEPDqMFdYEN1+fbjb1uNgnEN5ZjlGVsXXGEE4dAeJRWzNsZ1V7l2CkFBu2ry8E09rBgfzamiEMi60UpTSF1MwoUoZtENhwGKdCXaOETVO764aBfRUgU0P0o/Ypk9B3qbf6WsWsJxAaVWB0dKBzEVIOllutgR4hLbborj1M96IbKQVSLmjq4bOCfEvgvg6RR5HwNYKcIsy2EK+PHHvsGDAIZoZV59ac6imljMcLfD6qEJ/OSJgYIBvKbArgDjyKgwyMpX519z6bzFywyhfxCzM983qiG5Ux3+lWyhjq1klFTCmGbuhYAlZRjop0lEocVIXd8xlmWAmXFiUnHUhzfL4GrW7FtI3bnJ6PKcIsmFMjKKrZWkRopSjLxgW9WrFurci3a9eeOqomfHUBudtfgzFDWizGP9emRC6R1fpGlquXkRczVrck1i9I5GuvIT/nuSgBHryf8Nnb0EcfQ26LhHWApiOEL7kZtQCJjLrCrGlzECvAJ44bC0HUK19wRVDtj1rq9WQ6sOpxTHk7RmivwijTncE/tyHMZ4Z7p+MCka7v0dR7hiDUhp4D95vW1NDJ86WgIQzhass3zoSURqEutXd5P6z44lislsoTXTPvopOxNL4bBMiMMNJg95tNXivEa/BHooXwZ6W3Zvfq4ffcmfPXWr51u16zXJ5ivV7Ttt1QMeOMc26qjHNsRbYm9F1vf3vHm3PZYrm+mVOrf0LeuoTlrdD+XaFsNeT9B0x/futewm/+JnzxLwltJK7mMMfgthCBbWtS5BFWGZxfBt5Dldr1TK0CnsnOGNyirni2hlED2IRO5tYEO0yfrqaNO4NjalP9gvF+hXXPNi44le+Txpn2rXMZE095XBBPLgAITi2wGVOoE1OPwZENz+cYCCJhD/rh0F6wwl4BpHa1ZNxNhj4u2SpRUp/I7ivY82YzFzEceJznGpwplBDJ23OSBMOtZ5lctuiby0jxKvKBS8iXC/mqgIYM67XdHn8ceewxePQRJDQQDI3RbYEt/72lwtoUTXVG63WZuIz1/xF+2+sL4YbI1ATRydcNcrqpik8XT9EzycF3UNYXhPW0Ptr7fEUgNutcRzu4ljeFUKuQK9Tk3Hqayeqt0krxJvFlqAYx+xOSt7yInq5qQmuTO/LxQQhKjAXByNFjU7fAEfmYzbwzVzSGUBEoq562s+5c9OIOaOHU7pK2ba3064knaD05ad11qBYkNgPJuVejmZbuWnKf6A5fz4nX/ATdNYfpkrLuIZeGdX4Fq7RN2Rfov0/oEeTRo4SPfoR459fg2/chjx1BGqs2yYqVh72ykF5dkBZmfwSz262sKkbzDyp3ic2CM5t6JNFSm6oJUpHlMgizuDMZhll000Mm8j00RLKbPel/yggUwJjqKtMvOMO4QJp6ItgTYR4N17psnTDSYTqjOZi0rRi2PgaSwVKy99fO9N7Xr0+JLvXuRXvRqFQSxkhAiVonsJaMiZsiZhTEWGjcdhd/3RxNC7LE2NA4m+p6fYq+7yi1GSgWqdw9ddL6I67WnDx+gna1JlVSHtSSsRZzEHGh9mSmriV3Pd3lV3Dqb/00q++/hS7Dug9kFTppWDOz6uu5mS/hsWPE9/8h4UMfHHYU4oxkBFKUqLSvSKx/KSEnYefegPwXS9fNBacsmE5ZdOXjwlznRnCzQ/1/E1TcfKo6367siIKMu+mIeVdtNrVGqpycDvI907hAQu1L1m20opa0L9Ntt27djBfAODWyccwNdAMmjFGi6ZaC5UnnMpRrFW/vZlokD/ho0OATBKI94jal1otYeozvzZKcLBMN3KK048fpx2p7B9hANCp6kTwH2qq9i3WcU+tvmMRb6mmgZJOUlCt/h7JKSpthpYH1fMF6a5t0QuiPBUqvJDmBynETqKrJHrwPjj8B6xXE/ej8WUjcdoTCDYxlIj6WkVNKaJ2zUCMiWygzYAk8jtWP7gMutfNlzH2vnW3N1PLILydAHkK9q4EJsl0vR6OH+S2+y+YyvsLGL9SF45o6fCeL+kLg1JpcMzsurAmSN/ssDsLX5QxOL2bMSetVx4kTS2aznkbm7MwULcJsa8FOs98EtQtInyldottdW7V4SUjqiMB8MWOxyEgQZrTW71shZhk5p7WGY5NzyVlRanbHMsockQZVo6oNpSGVQJc7QFitliyXZjK0XU/rtYInTlkT+5QLyzyjl0gOZg4p0BelX0NWZdX2rFsLja86ocuRbhU5kQOdCvplgfcI+liPhk+h4Q+BpTmvIsjxJ+Deu9CFoOF5SPx5RJ4HUWncjIifKiyOKJJg/hVhtsDMjti4EH0BeDdwFLgZDW9B9TKbR9wnSSDZTZKqufkcNL8L+jC1ZSTATGEmVm9qXdszWTOaIaiVdNWk1qKFYiuQECIRv79hw1fOls1xnjX1qJWj28NFjUJARIxrrVaLu2mhauHhnI3wcLVsSU2m20qU3oD+hhmLsGVaP2UoSmkTaWkMTKKJUHpElCYq85nlNMyoOlehZJulSaZczmlyH6+lDISwjcgc1UgpoOL8Fd7t1RANcwR3V2uW65Y+F04s16za3kwGIoWGHAJZbIqXObHrVdwnlz27q0JSZZ2Evgh9J6xUSATCt4XmA4J8qyDhq4T4XoQnPJm+arPi/VCvxrKEf8DMh2ghu/g1ha+aIDZBiLM6T9Up3AL+M+gKeDbI30J59oDAbWD3IvYdQRDmhPAB4GFygW4CZgim65NjPUFBS+8WfCVenwwP39cC6Br4Ots4z0I9cQjPcKvDdjRF9QCqL6DoFkVPUsrRgYGoTz0iMmS0AZZDvVFn6NCTw23W+Kdxe9wyyyx5xto8WxW2O5gOEdrkGQ2tiPWAsKpv8dIvy8foPUOu9xbNtQ95nkCRfhmQwds3P6Awpp6mHEjpMCkfJC+EfJ1SDipcey3x2/dZ9O4eiK1gKZ33Ib5oB0ojqSaql6RNUI0hxb9KptNTwBFG01CAuxBWbmgdAb6AyAMDoiGu8SvMNhoGR4AbEC5D5HqCzFDJhPAAQR5C5AGCHJto9gn8NzU83GTZCGecERIZx3m3qQcUw+vMTifQ1csGUH0+pbyDXF5ASn9K3/8OWo6wWq84uXuK+WzObL6AEIa6xCBC7tMgRJZrbf0S59tbLLYXHifJINbQp+s7481w7ZqzV1MbWYY3JFp4KdMCCTNyVlarlj4V2q7n5K4FUWLXE50tte06+mRUYVqdK0Ao5kMU64qV1XgA22VHKgdoV6+jbd9APtTQvw3Sq0Huu4+tD32Q8NCD8IiiJxWajPIAysrLwUaUN4iVSMlQLuA+3OB0VcfrFPA+4MMIk1wYOYaVngpGInAEmPvxu/CJ3R+/SxBehPALCFcRwrNpmv3Akqb5ELF5H1FOEMN91IZR0+VQF1r1pcDqIysgEFQH+/pM4/xCelIx3k1B3hTqcSWao3iQoj+A6ssp+jClzMiSSckqQwDavqPtOqPADcYsqqU6hh5caZwSuGmIzdxjA8mcGa/Bs5zlQusENKMHXrdAq/lzQmYUy89ona/u5KlTpD4xL8qWlYM7I37N8PNzHd19at1ghSBT6klZyPl6Uno1ZTEn3ySUH4X4qU/S3HM38fbbRhxexjItwHuQ6yh0yqiZJ6hZ9dNNQjvgLtA/BfGWI7UpqoNywkPAg6OmnPg941dX4bwE4aXAS7xlhgK7BLmHIJ9AwhqRxlWQMNVnDKkQ408VBSmOoRdFTxOxnI4Lgn48yezw5zc3lrpFPgb8Gcq3Uf0LctkFiguTReratqOd95a+2EATrDiguIVWHKZCLAeiS73hyZpQTeRiqEQa0jZrH0LXEuJE5rlAEDQVJGT6zhzBru1Zr3vW646+H6urRXVI7M+eeNT1xZOUnEQmXc2qexGZHVaHE+vDiRT303XPJ/eRcq2gV4/XLUYrPq2Yuyjk2mzJ0VBRRbkEeAlwNYVrEPk8eCtsU3oN8EK/uWwPGN0wU2yCwlPUCoQd4EaU5+x53ytRDrgG/xYxfA3kGCF8gxDUCwI26zRr2sTEChk09WB+qFUjfdfoh4i8GOPMq+P5wP8C/HueKpce0/TQsHFCwKQZT91AQbkb5P8FZmQ9RZ+OkcXID0NYetbdDiHMTZi3tigNvoUJNR9BPQ859BmCMZLm0lFKMoLyrifnnj4nOuczp35eBckK4qX8qUdF6bqe3d017bpjd7XmxMklfZ/pRcjRUJU+p4EladX1tCmRsrLu7W/bP59T7T8mNdex/H449Q7Il0a6coheZ4ZeXOW+QbBC3tl8ZkhMqm34BNU8XDvT2s9C9BeBH0fkC4j+B0TuHru1sQP8AxNIhUqrW2PzdWY2hXWPgHMZwtuAv00Nq9vYD1wFZEL4DDH8P8BjzOLjxDA2haqgwIaQVl2mk9z7ot4sSXyhfJeOotMi3Ix9acQ8ivdh3B9PiUsPPwU7pmp/faexBJ4A8ehTsVBydidOJAxkLkRLmI+T3ckKDkZWIqM6AKicdJmieWgbPJRDDRcA1+p4BylvrClW3Z0HMpkyctGFQGV7G4hnCqQspNw4waNh0H25hJ7rSfIC+suE/vlCuUzI2apLVBSVDH1GUhps16GYYu98DfcWKIcRbgR9AOQIcLdjyRlhH/CYrfaJZrapkQ1Jq0lmMmwFHiRhBlyF8ILRTNmY0A44gsjXEXkUi8RG9pqeT9qjN5zBTUdR1HtPnkFa4KmbH68D7lbVe+WvyKX3pGhRjSYyvSAjBhSCekjcnqrptKkvtKEnZ5wpv6c0sLUwiN98FksxzSWTugxaaLtC2K19Y9ak3LlCVpBqzwlonNiqFjyRziJgqXSW7pkyu6uOvkv0fUHEGnYWzLxAq8nRk8ol7K5ewbp/nmnqTklFaa9/OctXHCRfHmh/SMhbgrbA5yB8CdDjhPgZQvgm8d5vwiOPjEXDrs0MaDQnsN5gnHjhMCI/h3ALIl8DPsuoEodpcEaF0cYd0heGX7CLP9AbcxLl48BJ0/Rs/q6xW/05IbQMSV+Dcwin07jVALFg1vTgePLGcYbxVIX654Hf8ftPnUsPhiy5UA8wjCdn5CmwIdSCZb8VY1Qq7oAJiaLCLCrr7Zb5rEPnzoqAIyvuJKdU6NoezYU+rUj9mlIyXb+kTy2xCWztLIY8Dqv4Dp7KWlvKdfTJophtr6Rk2HXbWuQvZyVIhCagObH2TMF117PuWlLe4dTy9azbN5JyYNULqQjdcxesfmk/+XmBvC3kHdDHgY9C+C0I6Riz2ftomg9ZfLtdUbwDWNHN64Ra4v5mYYgSeB6Bf4Roh/BukK8BT2xgx9XSGjQ1lcY4DLI0xAPrDqEnEH4f5cODkI6yVxGRdlOoq5Y+g0DXYoTqCANeHHEO0uzjnIVajMjmp4F/+aSD0XPj0rt0USm1RntJ9uwlpy/1Co4nz1EWoIFcIpIahEDKM+8cUClrbausCTSKJwepB3H67EJtWHejkSY1lmOCEIpp7uK1f5aTbeZFKdD3Zj7kAjnbrTijkOL2u+d593lhPVP0UtLWFaTFIcehg/HW7YMyE0oUtAeeAB5X5NgKjrRIOkaYHUWaI3adYkCDjJMPg2AOqAaCGVCngMdN4yoY0/82wkF/z/a4Sw4XfHMehPEailbAxH5NJQMnEU5M5muvAARqiOuMYrkXKdhjkUwX2ml/Y894Kpr6DcBfqOoj/vgpc+k9e3/QkvNge+wt3xIZ7e3g93MjNBFKaSj5+0jlVkq5FMyaoIlLYnM7qneytejY2tpBvNpkPpsZbW4REuYotklpO3MO152VU8WmoCEyy1YiFp3tP5fiSVFK12W6zoU6QSr2+5XfvdrtqkaS062X5Lxg3b6GdftK8mWHWP6Nm2hvihSEvli2XF4B7xEk1S5XwGoJd/wxyGdMmONdtbU4uc/jPuZwXpheP8BYP44Q5L0IX7DrS83R2EJ5B7Bl5siAO4/CsiE0p9NV4iKqo7lwxjHN7SkW8BqwgAFTtC8dyvzqzlNrDipSwpPM/dOOpyLUb2c0PQB+n6fIpWc2sdVAny6SOP0bopkoTRFrC1yEPryEXn+RUq4l9ZnUJWJ4jKY5BtxB3/fs378iNjNr3RYbIpbJZslDGJlLb4Wu6zbR9okmKxJ6Uq6t5UzFpyHTT2nbRNsawpCKeIcqyFm8gsY3XYVV2xmTUp6xXv8g6/Yfkq/cZv2jc7q3GG9ecZ9Lfw/4X0Humdi1rEE/icqvI6EbykUqMXqp3CkDx6R9cHAiEUQeJ8oHzSSibuUC+vNo+ReoHkaoWPGIdtQAis3DVOSmwljfS0X3JmNTzZq5aZHbEIxpVkM1NSzrj2ouIuO2EAK1aHw8p+8oYcC5t8fYB/wk8E8nT/9vPEUuvbFMyz1c8efO/uuMF7tBdUFhi3RJpt9OZHZo0yFm7WFElHXbMJ93lIJ11oriRIuMEJ9Y8r/ESCxmR9eqc4qSLV3HhdqYR1OtHXTs2mgCnJrA2cyLpRPT7exnfellZA7StYfou23Kc7ZIlwllS6AFPQasQY7uwvpxpOuHJkZwHIvKX4POEuVQRPcHylIpDxfKSoFTiB4FrBGpX2AT8gH66hExWjWhEi4WkAXCll3dEcw446xt3B9kdu+sLYArUN0ZrYm6KBSgoHoS5NRwoHVXlqmd7hp5Lz3dIA1iB3023OxcufR2gSv2PHeUp8ilB5aBplrMAVEGutzqXIA4sx6DjSrepUsJZtvGzPJVyu5rA9IeoHv/6zj56ReyPXsE+FNO7d7N9tYWBw8Wa4Xh+dmqEQlzmsUOqpkwa8i6g2G7mR5Lbe1TS21v16dkx6LB8j9UPA8kGgtSbYmcEu26I0tg90d+jOM/+Sby/itIeiOZObpPyC810vPwbWjeDXKXEh78MuHUe5DtR7zAFpQ5fXkhKf/PcDBQfi6gPwT6FSi/rejXFZFPI/wuIo+Zv+FKImA+iEXyJqCBhQnR2kvc0MnRUZdxp6nuUZ2D0d4eQOQRQ67ywGFUfx542WiXw6DARJcEeT/KR1F6Bjt7QF0YWFPNpBKy1v1DJ7909hA5nPfU04r1mjNXWyyMMI79ZyJf7caqXaPpm6KkRuleCMs3CJzaJv+X72N9/CbWs6+ztfU5cj5Bt5OYzbdJBWcBqhfRwuRKIc4iypxcMm23JpdEX7BMOsewq1CHMLfFhafNYr25U3YCmS6xXK5JIXLymus4/rqfIh+6EnXHbrqXxyMQ/wTipyE099HMP4zM7rYFX5SiV0C5mVJ+Ft03I/0w6JtBrwD9AAP6E/QPELyF29DrcfRN7CfdaZ7iEr7Da2CDaalC0DqYI4wvTsR3Ol/jey5H9VZEXjuFsk2fFAV5HOUulD/a3JurH7WHMdKc3qkpM/17djvkgofJTz+8QBZLnK/RrhCOEuLnCfog8uA1hNuejaZA3gn0t0RCv4/l7guQ3VMU7VhsL+nSLrOm8kVDzkb3hahbIh5GJ5C9iXy9WSs39dBzIVQeLDU0IaXiCEqmO3Ap3fU3kLb30V//PPJiTgliSEUYbVS7cxLk28AJ4Ouorjz/RAfid3M6BV0FuBP0M8BXQHaxZS8TLTy5bgNE59ewQqQqvu9NnXTd+wWjRqxox1TIbDTANahezdAjQBThOcADqH5mItCCyFXAYfaU2k7MpCrU1TRyc5OCBdymMiGObJ3dQT3PNYqMHM0DV/OIbdYiU8QXtyiqDRIDQSKx+RLN7H7I+5h94meJn38b+cAO3Y/B+l8F1vdfTXnXz7C448fZ2f4yq/Y9LOYPsdias72zbbkhgzDDYh6YzYJRE5RAyoEuBwuTZ6XvlbYvjgeLw4piudcahx4rfZ9ob3oFu7/w90mHr2N19TV0lxygNFYIoW4GRHUxkHsh/gbM/pIiR1GOgS+iPnmljNco6iOg7wI+DBwHud852rEGBIJl41WNLJRBcMFlYkI9YESOPh91OwSs/bJnEg7C7NAhU8hhC/QnUP05YHvy/APAR4DfGJxHK6T42wh/j1qMW+WgNl4dK9LxcH8N/dtxqniVu9tMA03cWcYFz6fe8FImENHUqUTEssbiEWK4m5LnhPtvITyRKVdCeX2gfxWUO/exu3MD3a6Sy4r5XNia75qzJ7XhvSMrQYiNYdlZZbgVNYcxu5ZOuQY5qqPlHrxaM6Kut05Z3aWX0b78B0g3vNhyP8QxhY3NqE7ICZAvovJpd5Asv6SoGD9etS1V0aWiX2G0U4Mt9qEMbeMn6gdHo3YalBunoeZYD28brnn1azZj0ZN5IKJcD7waZd/kSz+H8hugn5isgRlwE6odwmKPJPi/Uyh30NgTh3Bqu1fB/g5yfX6FWoycPEgYqAUqzzRs2mFjcwT7DGo9U2bzOSJz5ovIYi7EhZAj9AHKwUD3I5B3CsyvYnbgNayb69h65GHSvd8i7K5oZoHYWPFuyjPmc6u5SyUZlZfTlUls0OSUuaUhpReT0g3odkO+KVCuDUbC2Bp5Y/+KH6Q7cMDMhm9CvNPC3WPI+RSRLyNyH3LXXeiRI95o6FpKeRmE/fTPV9ILC6pz9Cun4O7fQzQya2o7a8XIJQsityEs9yyaLZSXovp8wLqCiXtrYyr/IYQ/AbkE0ReBvGiYG7v47PnO6XCNuZGMUZ++DOXHEC4DfRDlS8BqtO333kItgMYZr9gj1NWIWoC+GMrzQSuEUMd7TnuU550iIcbGas5CZOTO8BOoYd8aAlbbr41kZsZsltne2iI3C1I/I3VCvyN0CyAK5Rpl9Xah/HRgFV9AO/9FGl2x9aEPsO+uXyc+dpTZvGE+t5ZxW1tz5rPGegnNxEiKVNHgfHWS6QukvGB3eSur3V+gbG3R3Sqk1xvWnCrMt38f+dAhs7dvE5p3ghwb5UM4hsi7EfkgrFeUI8esr4u+hD7/Mjp/LukWJf3jAnqC8M7/SLjvXcRc2NraomlmlJLpk7WTEz2O8ET9cldiByjlp0D+DoJ15w3+uhUVF5CPI+HfAruI/BLw3PHzMIj+ZNKGV4ao/GQ4ywHCYVR/icISkT8m8K+Be6syHv7WFtAh1FY3XgIWZMhrV9UB+YBttLyekn8edD6sKRtPA6G2sRfgn3oq49aiU40wbFGm4TUGQijE2FLCglAa6CPaCOUayCJI2KGLh63Q97IraUpD7GTIsQ4xwEwoM+/6FAOxGWdX1cjQc5/IaYsUr6ILL6DMt2mvhHS9UILT6YpYnaJk6DvicSHcI8gRJk79LoQHQL7umtP7GuoORa9B9TrKAaUcVtAjhEtW9l7JhLBNjHMsQagF8sShWiBkRDNIBC4HrgNmo9YdhCBjePLDiDwOPIHh12cbunHfNGXG2stV8alzeJW//2qs+nzu79nEUiYKmYFH4bSjmjuXgV4HLFw+nmboR/YWZYq55lIqw2bNp3ZHZ0jxDJTUolLo+46274zTo9xGIaLLywmffBXzEy+jXCmEVwnlenNEcEczlRtZd28nrI/TpcC6C8glQnNzILwsWiFLY5mAtpgsftZ1yQjYy4Kuu5mum6EHBb3BVGBF6uw6fxPkz4FjiBhHSa0esS89AvIts5+lakNBwreI/EdUDsGXQd6loKcIX/nKEP5OOaO0xvRkiSYgNwCvAnZA7kDkcwhLgnyCIGnAqjdQElGC7CLyFiAi/BAwcxMgeDcvO95q01e0oQ6hBbmN0Pw7nLvMZUxtN1A1s4i/iYUxfgRkByQNtkdtKKqqQ9s/YOh5M4bWAWmJ8TNmr9CMftZZxnmnSEglI1ow5jhMsCsIVfnxUFA/OSCxopBJfT9Uu/TpE2Q+Rdl9FvEjC+YfeQnlpoZ4DZQXMLl4kVReTlm/EFlbqYgG4ADojwB/l0qZRyUTt6pxk1YdigW2EeYQBdmuDludUwW+jvBOkK+j0qBhhsaqqU27FXY9cDNB0+ROmnAfEAl3eIU3Cstdu84+oJ4AAAgzSURBVGgipGQptn5w/uGbQP4ZyFUE3knkLxE5RZAPEcKf+lZv4WmhAmoBkbcg8svAtVhUcWHnECOhMaHJ2RorGfGmKZnR9l0Rwp8R5XbATDapcFw2e1/Laynlv0d5ISJb2AU7PlGwhnCUSuxZDHkxoXaqhArjsiY0HyPwqcGBfpINtGdcMC49dRC1Zn5R0Q7G9MNqjpiYlwkiUlDWQAtljqzXSAvhKJSj9rf25NOslqxWNwBwiChQyjaabUvLKt6729EN8C4HbiZVzuYMcsqFVXpElggdcBThKPRHYbdB1fkGhu3fz6cuYamuW0flr5A1yKo+X5hcmUGb1WalTxqCnaTsArtYxG4H08Q9whIzvnrgICKXA0uEYyDHGPt9VwfN/91Qi/X+LrBmyDURsBVsHXSRBXCQPUHoje85Heehlhmq+1B1IGGILS+xHXCUibON8y7UFtsaY1xTV8r+nRqBZXAkCkJQq9ETsP7itTjUa2E5BuG9UG6vsmT7XPjql5DlR5DmuFNuKrq+lPLRv0l54Aehsk74HNZkpSH0xmTy/EhNAz8E8kHgbkTvAY5DnqFfFcrSFaqOcl1D1CaWo8COgmSLF38k9X4NmKAMcJl8GZF/g8g+RO4wW1vG9yBXYomVLwXuQuQDiDzAuEw7hE9itR3HUP08aGZsxAoVphi+sx6K919XYZI6/GyU1wM3YBV/h4Y1MeDeOliWVmgMw85QtIDeRMmvB66kaJWUJcgfgXyKzX6OZ7arzztOjW+GsjeGryAO2UyD57YNWcZeKAy9QazW0QUkCiFCeQL40PhLQ5aX3kUovw3xQSQqNAW6w+RPPQ/59CuBmhGmVmhbs2MH82PzDMb7jyH8AegnGGcsosGS9TdOr+ZcjKf7pEc6iHwNptjWUheFbT7Vwf46yD3uL2aGKvDhdy8HeSPwJpCPQbgNeHB4g2gPfA74deDk5Bg3j2yIPNobRuFmjOxZStKVWLr9j7sTvKfVcrWhfTe23oiVfjm5pr4eLW8HXjTmbstRRB7Bcl18dz+7n3j+e76MiknZQ286vsfHRplXtV9lQm8VxIhd6uPSEvJRKLuTTxdEHkbCGqSmcSqwQtJDiH4DL4bCEm/EOMPrsRbsdb0c5SBKRjiCCcK9oCd9SxfvRRIsdNjoeHrKRFyn57f33EfarenrVYAqamDbcEYkD9fnybeewBHDxXkUoXf7/wSi3wR5HHgUYcWIZJyu5/eE1V82nx92mBrw0Rm6EWRRRI4jcpQgjyPyhB37xvdsAQdR3QKuBfajLCZvWQDNHll4mqEfphDVaw3rtjq5OHumXsTCwd6ym5IVyUoTqykSaKIFYEJ4GPJvI+EOjJez8cVwHxJ3QZpBqJVdRH6foJ9308CcFVHDUKFukwA7aHkz6JuA48B/Aj6JkSfeC0Q7QN8wVTKZNGy94xRU1GfynNqRDg/qNalPDe814WpCpThg+JKK+wojo6vwKFH+A0E+BvIoITzkV/rzIP8HonPgLsyen5B96Tgf1ZwZeUOetP34tRvNqWHeAKsm/yyz5r1IeIwm3kWMtedPXXzXYVnLN6Jci3Jo4zuEsfEqEgcH/WzjwrTHGIR379Hpxh+wt9RcDZs8+2wIln1n5Ogm+EWPE/VTUD4AEikyN5swCBLtS6pQmXa6A9Hb3bs3exJhcE/q7qB6KXAj6OtBl8BtoO/175lj+ZvGD2XDHR8/l+FMZXT+aiGhVaT49ZjY7Huvi2LCHKdCPTnGMPwV/3uCyJ8P2h2ponovxmhRL7ROf2Zj2Mf2miP+fRM7efrRjfUmSpC7CfEPCeER76Bmiy5QQc0rgFtRfXX9xSed/UZqRWCy4k8/nn6dBPaMyQ5+7p8ZzMAnf8poBfY+X4XqzBO8+cTU+5ke5V99nP4bRqmRvU+f09gratOv0Ce9a+N1W82n+fT00ZM/efp5OpfZkzPcP/17z6at5UwVBn8dQ0Qew/CgI+ftR8//OMTF8ztf47mqeuXeJ8+rUAOIyOdU9Zbz+qPncVw8vws/vnNtzMVxcXyPjYtCfXE848aFEOp3XoDfPJ/j4vld4HHebeqL4+L46x4XzY+L4xk3zqtQi8jrReROEfmG0/9+zw4ReY6IfFxEviIiXxaRX/HnLxeRj4rI1/3vZRf6WL+bISJRRO4Qkff74+eJyGd9Dn/XORafVuO8CbVzW//fWOrYS4C3i8hLztfv/zWMBPyPqvoSLFv/v/XzqbzdNwB/7I+/l8evAF+dPP7fgX+tqi/E8gT+0QU5qrOM86mpXwl8Q1XvUdUO+P+AN5/H3/+vOlT1IVX9C79/Epv4a7Fz+k1/228Cb7kwR/jdDxE5DLwJS+VDLEXwtYzFgU/L8zufQn0tcN/k8f3+3Pf8EJHrgVdgbOZ/Jd7up+n4P4H/ibGxwhXAE6qa/PHTcg4vOorf5RCR/Vja3n+nqiemr+mYufU9N0Tkp4BHVfX2C30sT3Wcz4SmB4DnTB4f9ue+Z4eIzDCBfpeqvtefPife7u+B8Wrgp0XkjVjS8yXArwEHRaRxbf20nMPzqalvA25w73mOtdr4/fP4+/9Vh9uXvwF8VVV/dfJS5e2Gc+XtfhoOVf2XqnpYVa/H5upjqvrfAB8H3upve1qe33kTal/Zv4yxwn0VeLeqfvl8/f5fw3g18IvAa0Xk8357I8bb/ZMi8nXgJ/zxM2n8C+B/EJFvYDb2b1zg43nSuBhRvDieceOio3hxPOPGRaG+OJ5x46JQXxzPuHFRqC+OZ9y4KNQXxzNuXBTqi+MZNy4K9cXxjBsXhfrieMaN/x+Krgd8KnYPaAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "image, label = train_data[0]\n",
    "plt.imshow(image)\n",
    "print(label)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "GBaOZQ_MoHp9",
   "metadata": {
    "id": "GBaOZQ_MoHp9"
   },
   "source": [
    "#### Load train and val data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "d6-49mthoJ94",
   "metadata": {
    "executionInfo": {
     "elapsed": 367,
     "status": "ok",
     "timestamp": 1642416578698,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "d6-49mthoJ94"
   },
   "outputs": [],
   "source": [
    "train_data = AWETrainSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), data_transforms_sharpening[\"train\"])\n",
    "val_data = AWEValSet(os.path.join(os.getcwd(), \"data/perfectly_detected_ears\"), data_transforms_sharpening[\"val\"])\n",
    "\n",
    "# Create training and validation dataloaders\n",
    "dataloaders_dict = {'train': torch.utils.data.DataLoader(train_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=4),\n",
    "                   'val': torch.utils.data.DataLoader(val_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=4)}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "rIVsMbuEottv",
   "metadata": {
    "id": "rIVsMbuEottv"
   },
   "source": [
    "#### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "essvD7f7ov-w",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "essvD7f7ov-w",
    "outputId": "3c415ff5-8c45-473a-c840-012cd36b1e47"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training model on GPUs\n",
      "Epoch 0/199\n",
      "----------\n",
      "train Loss: 4.6750 Acc: 0.0072\n",
      "val Loss: 4.6328 Acc: 0.0119\n",
      "\n",
      "Epoch 1/199\n",
      "----------\n",
      "train Loss: 4.6435 Acc: 0.0072\n",
      "val Loss: 4.6331 Acc: 0.0060\n",
      "\n",
      "Epoch 2/199\n",
      "----------\n",
      "train Loss: 4.6173 Acc: 0.0072\n",
      "val Loss: 4.6164 Acc: 0.0149\n",
      "\n",
      "Epoch 3/199\n",
      "----------\n",
      "train Loss: 4.6014 Acc: 0.0169\n",
      "val Loss: 4.6087 Acc: 0.0090\n",
      "\n",
      "Epoch 4/199\n",
      "----------\n",
      "train Loss: 4.6027 Acc: 0.0169\n",
      "val Loss: 4.6049 Acc: 0.0209\n",
      "\n",
      "Epoch 5/199\n",
      "----------\n",
      "train Loss: 4.5870 Acc: 0.0169\n",
      "val Loss: 4.5972 Acc: 0.0060\n",
      "\n",
      "Epoch 6/199\n",
      "----------\n",
      "train Loss: 4.5741 Acc: 0.0217\n",
      "val Loss: 4.5881 Acc: 0.0090\n",
      "\n",
      "Epoch 7/199\n",
      "----------\n",
      "train Loss: 4.5586 Acc: 0.0217\n",
      "val Loss: 4.5834 Acc: 0.0209\n",
      "\n",
      "Epoch 8/199\n",
      "----------\n",
      "train Loss: 4.5357 Acc: 0.0217\n",
      "val Loss: 4.5846 Acc: 0.0239\n",
      "\n",
      "Epoch 9/199\n",
      "----------\n",
      "train Loss: 4.5169 Acc: 0.0434\n",
      "val Loss: 4.5705 Acc: 0.0269\n",
      "\n",
      "Epoch 10/199\n",
      "----------\n",
      "train Loss: 4.5106 Acc: 0.0410\n",
      "val Loss: 4.5637 Acc: 0.0090\n",
      "\n",
      "Epoch 11/199\n",
      "----------\n",
      "train Loss: 4.4936 Acc: 0.0361\n",
      "val Loss: 4.5665 Acc: 0.0209\n",
      "\n",
      "Epoch 12/199\n",
      "----------\n",
      "train Loss: 4.4861 Acc: 0.0434\n",
      "val Loss: 4.5595 Acc: 0.0149\n",
      "\n",
      "Epoch 13/199\n",
      "----------\n",
      "train Loss: 4.4638 Acc: 0.0578\n",
      "val Loss: 4.5553 Acc: 0.0239\n",
      "\n",
      "Epoch 14/199\n",
      "----------\n",
      "train Loss: 4.4381 Acc: 0.0554\n",
      "val Loss: 4.5534 Acc: 0.0209\n",
      "\n",
      "Epoch 15/199\n",
      "----------\n",
      "train Loss: 4.4341 Acc: 0.0699\n",
      "val Loss: 4.5482 Acc: 0.0209\n",
      "\n",
      "Epoch 16/199\n",
      "----------\n",
      "train Loss: 4.4216 Acc: 0.0651\n",
      "val Loss: 4.5485 Acc: 0.0239\n",
      "\n",
      "Epoch 17/199\n",
      "----------\n",
      "train Loss: 4.3904 Acc: 0.0771\n",
      "val Loss: 4.5475 Acc: 0.0269\n",
      "\n",
      "Epoch 18/199\n",
      "----------\n",
      "train Loss: 4.3795 Acc: 0.0843\n",
      "val Loss: 4.5410 Acc: 0.0239\n",
      "\n",
      "Epoch 19/199\n",
      "----------\n",
      "train Loss: 4.3650 Acc: 0.0940\n",
      "val Loss: 4.5397 Acc: 0.0209\n",
      "\n",
      "Epoch 20/199\n",
      "----------\n",
      "train Loss: 4.3399 Acc: 0.0892\n",
      "val Loss: 4.5310 Acc: 0.0239\n",
      "\n",
      "Epoch 21/199\n",
      "----------\n",
      "train Loss: 4.3402 Acc: 0.0988\n",
      "val Loss: 4.5233 Acc: 0.0388\n",
      "\n",
      "Epoch 22/199\n",
      "----------\n",
      "train Loss: 4.3250 Acc: 0.1036\n",
      "val Loss: 4.5168 Acc: 0.0328\n",
      "\n",
      "Epoch 23/199\n",
      "----------\n",
      "train Loss: 4.3044 Acc: 0.1036\n",
      "val Loss: 4.5111 Acc: 0.0328\n",
      "\n",
      "Epoch 24/199\n",
      "----------\n",
      "train Loss: 4.2762 Acc: 0.1277\n",
      "val Loss: 4.5085 Acc: 0.0328\n",
      "\n",
      "Epoch 25/199\n",
      "----------\n",
      "train Loss: 4.2709 Acc: 0.1205\n",
      "val Loss: 4.5070 Acc: 0.0418\n",
      "\n",
      "Epoch 26/199\n",
      "----------\n",
      "train Loss: 4.2601 Acc: 0.1108\n",
      "val Loss: 4.4948 Acc: 0.0358\n",
      "\n",
      "Epoch 27/199\n",
      "----------\n",
      "train Loss: 4.2254 Acc: 0.1590\n",
      "val Loss: 4.5002 Acc: 0.0418\n",
      "\n",
      "Epoch 28/199\n",
      "----------\n",
      "train Loss: 4.2157 Acc: 0.1373\n",
      "val Loss: 4.4871 Acc: 0.0358\n",
      "\n",
      "Epoch 29/199\n",
      "----------\n",
      "train Loss: 4.1807 Acc: 0.1277\n",
      "val Loss: 4.4837 Acc: 0.0388\n",
      "\n",
      "Epoch 30/199\n",
      "----------\n",
      "train Loss: 4.1764 Acc: 0.1422\n",
      "val Loss: 4.4790 Acc: 0.0507\n",
      "\n",
      "Epoch 31/199\n",
      "----------\n",
      "train Loss: 4.1429 Acc: 0.1735\n",
      "val Loss: 4.4717 Acc: 0.0418\n",
      "\n",
      "Epoch 32/199\n",
      "----------\n",
      "train Loss: 4.1282 Acc: 0.1759\n",
      "val Loss: 4.4736 Acc: 0.0567\n",
      "\n",
      "Epoch 33/199\n",
      "----------\n",
      "train Loss: 4.1161 Acc: 0.1542\n",
      "val Loss: 4.4639 Acc: 0.0507\n",
      "\n",
      "Epoch 34/199\n",
      "----------\n",
      "train Loss: 4.1047 Acc: 0.1253\n",
      "val Loss: 4.4631 Acc: 0.0627\n",
      "\n",
      "Epoch 35/199\n",
      "----------\n",
      "train Loss: 4.0785 Acc: 0.1759\n",
      "val Loss: 4.4515 Acc: 0.0478\n",
      "\n",
      "Epoch 36/199\n",
      "----------\n",
      "train Loss: 4.0561 Acc: 0.1855\n",
      "val Loss: 4.4481 Acc: 0.0597\n",
      "\n",
      "Epoch 37/199\n",
      "----------\n",
      "train Loss: 4.0708 Acc: 0.1614\n",
      "val Loss: 4.4439 Acc: 0.0657\n",
      "\n",
      "Epoch 38/199\n",
      "----------\n",
      "train Loss: 4.0063 Acc: 0.2145\n",
      "val Loss: 4.4441 Acc: 0.0567\n",
      "\n",
      "Epoch 39/199\n",
      "----------\n",
      "train Loss: 3.9732 Acc: 0.2072\n",
      "val Loss: 4.4316 Acc: 0.0478\n",
      "\n",
      "Epoch 40/199\n",
      "----------\n",
      "train Loss: 3.9437 Acc: 0.2169\n",
      "val Loss: 4.4354 Acc: 0.0567\n",
      "\n",
      "Epoch 41/199\n",
      "----------\n",
      "train Loss: 3.9495 Acc: 0.1711\n",
      "val Loss: 4.4326 Acc: 0.0478\n",
      "\n",
      "Epoch 42/199\n",
      "----------\n",
      "train Loss: 3.9260 Acc: 0.1904\n",
      "val Loss: 4.4232 Acc: 0.0567\n",
      "\n",
      "Epoch 43/199\n",
      "----------\n",
      "train Loss: 3.8990 Acc: 0.1639\n",
      "val Loss: 4.4175 Acc: 0.0478\n",
      "\n",
      "Epoch 44/199\n",
      "----------\n",
      "train Loss: 3.9255 Acc: 0.1976\n",
      "val Loss: 4.4266 Acc: 0.0507\n",
      "\n",
      "Epoch 45/199\n",
      "----------\n",
      "train Loss: 3.8569 Acc: 0.2072\n",
      "val Loss: 4.4160 Acc: 0.0537\n",
      "\n",
      "Epoch 46/199\n",
      "----------\n",
      "train Loss: 3.8669 Acc: 0.2145\n",
      "val Loss: 4.4175 Acc: 0.0836\n",
      "\n",
      "Epoch 47/199\n",
      "----------\n",
      "train Loss: 3.8438 Acc: 0.2096\n",
      "val Loss: 4.4138 Acc: 0.0746\n",
      "\n",
      "Epoch 48/199\n",
      "----------\n",
      "train Loss: 3.8372 Acc: 0.2217\n",
      "val Loss: 4.4160 Acc: 0.0507\n",
      "\n",
      "Epoch 49/199\n",
      "----------\n",
      "train Loss: 3.8139 Acc: 0.2193\n",
      "val Loss: 4.4091 Acc: 0.0776\n",
      "\n",
      "Epoch 50/199\n",
      "----------\n",
      "train Loss: 3.7843 Acc: 0.2120\n",
      "val Loss: 4.4040 Acc: 0.0746\n",
      "\n",
      "Epoch 51/199\n",
      "----------\n",
      "train Loss: 3.8049 Acc: 0.2241\n",
      "val Loss: 4.4069 Acc: 0.0716\n",
      "\n",
      "Epoch 52/199\n",
      "----------\n",
      "train Loss: 3.7491 Acc: 0.2072\n",
      "val Loss: 4.4089 Acc: 0.0716\n",
      "\n",
      "Epoch 53/199\n",
      "----------\n",
      "train Loss: 3.6825 Acc: 0.2506\n",
      "val Loss: 4.4008 Acc: 0.0776\n",
      "\n",
      "Epoch 54/199\n",
      "----------\n",
      "train Loss: 3.7038 Acc: 0.2554\n",
      "val Loss: 4.3966 Acc: 0.0806\n",
      "\n",
      "Epoch 55/199\n",
      "----------\n",
      "train Loss: 3.6994 Acc: 0.2361\n",
      "val Loss: 4.3977 Acc: 0.0716\n",
      "\n",
      "Epoch 56/199\n",
      "----------\n",
      "train Loss: 3.6297 Acc: 0.2795\n",
      "val Loss: 4.3920 Acc: 0.0806\n",
      "\n",
      "Epoch 57/199\n",
      "----------\n",
      "train Loss: 3.6646 Acc: 0.2265\n",
      "val Loss: 4.3935 Acc: 0.0776\n",
      "\n",
      "Epoch 58/199\n",
      "----------\n",
      "train Loss: 3.6392 Acc: 0.2578\n",
      "val Loss: 4.3906 Acc: 0.0806\n",
      "\n",
      "Epoch 59/199\n",
      "----------\n",
      "train Loss: 3.6026 Acc: 0.2867\n",
      "val Loss: 4.4040 Acc: 0.0657\n",
      "\n",
      "Epoch 60/199\n",
      "----------\n",
      "train Loss: 3.5965 Acc: 0.2602\n",
      "val Loss: 4.3926 Acc: 0.0746\n",
      "\n",
      "Epoch 61/199\n",
      "----------\n",
      "train Loss: 3.5923 Acc: 0.2651\n",
      "val Loss: 4.3926 Acc: 0.0776\n",
      "\n",
      "Epoch 62/199\n",
      "----------\n",
      "train Loss: 3.5549 Acc: 0.2627\n",
      "val Loss: 4.3910 Acc: 0.0836\n",
      "\n",
      "Epoch 63/199\n",
      "----------\n",
      "train Loss: 3.5421 Acc: 0.2627\n",
      "val Loss: 4.3900 Acc: 0.0896\n",
      "\n",
      "Epoch 64/199\n",
      "----------\n",
      "train Loss: 3.4998 Acc: 0.2916\n",
      "val Loss: 4.3857 Acc: 0.0866\n",
      "\n",
      "Epoch 65/199\n",
      "----------\n",
      "train Loss: 3.4854 Acc: 0.2843\n",
      "val Loss: 4.3971 Acc: 0.0925\n",
      "\n",
      "Epoch 66/199\n",
      "----------\n",
      "train Loss: 3.5007 Acc: 0.2651\n",
      "val Loss: 4.3859 Acc: 0.0955\n",
      "\n",
      "Epoch 67/199\n",
      "----------\n",
      "train Loss: 3.4443 Acc: 0.3157\n",
      "val Loss: 4.3867 Acc: 0.0836\n",
      "\n",
      "Epoch 68/199\n",
      "----------\n",
      "train Loss: 3.4383 Acc: 0.3060\n",
      "val Loss: 4.3852 Acc: 0.0746\n",
      "\n",
      "Epoch 69/199\n",
      "----------\n",
      "train Loss: 3.3921 Acc: 0.3301\n",
      "val Loss: 4.3788 Acc: 0.0985\n",
      "\n",
      "Epoch 70/199\n",
      "----------\n",
      "train Loss: 3.3683 Acc: 0.3205\n",
      "val Loss: 4.3909 Acc: 0.0925\n",
      "\n",
      "Epoch 71/199\n",
      "----------\n",
      "train Loss: 3.4109 Acc: 0.3036\n",
      "val Loss: 4.3992 Acc: 0.0836\n",
      "\n",
      "Epoch 72/199\n",
      "----------\n",
      "train Loss: 3.3379 Acc: 0.3253\n",
      "val Loss: 4.3829 Acc: 0.1015\n",
      "\n",
      "Epoch 73/199\n",
      "----------\n",
      "train Loss: 3.3569 Acc: 0.3494\n",
      "val Loss: 4.3886 Acc: 0.0687\n",
      "\n",
      "Epoch 74/199\n",
      "----------\n",
      "train Loss: 3.3204 Acc: 0.3181\n",
      "val Loss: 4.3944 Acc: 0.1015\n",
      "\n",
      "Epoch 75/199\n",
      "----------\n",
      "train Loss: 3.3357 Acc: 0.3012\n",
      "val Loss: 4.3860 Acc: 0.0985\n",
      "\n",
      "Epoch 76/199\n",
      "----------\n",
      "train Loss: 3.3359 Acc: 0.2988\n",
      "val Loss: 4.4018 Acc: 0.0836\n",
      "\n",
      "Epoch 77/199\n",
      "----------\n",
      "train Loss: 3.3420 Acc: 0.2843\n",
      "val Loss: 4.4008 Acc: 0.0866\n",
      "\n",
      "Epoch 78/199\n",
      "----------\n",
      "train Loss: 3.2705 Acc: 0.3253\n",
      "val Loss: 4.3835 Acc: 0.0866\n",
      "\n",
      "Epoch 79/199\n",
      "----------\n",
      "train Loss: 3.2443 Acc: 0.3566\n",
      "val Loss: 4.3796 Acc: 0.0925\n",
      "\n",
      "Epoch 80/199\n",
      "----------\n",
      "train Loss: 3.2140 Acc: 0.3446\n",
      "val Loss: 4.3941 Acc: 0.0925\n",
      "\n",
      "Epoch 81/199\n",
      "----------\n",
      "train Loss: 3.1815 Acc: 0.3542\n",
      "val Loss: 4.4122 Acc: 0.0836\n",
      "\n",
      "Epoch 82/199\n",
      "----------\n",
      "train Loss: 3.1743 Acc: 0.3783\n",
      "val Loss: 4.3928 Acc: 0.0925\n",
      "\n",
      "Epoch 83/199\n",
      "----------\n",
      "train Loss: 3.1918 Acc: 0.3373\n",
      "val Loss: 4.4032 Acc: 0.0866\n",
      "\n",
      "Epoch 84/199\n",
      "----------\n",
      "train Loss: 3.1403 Acc: 0.3639\n",
      "val Loss: 4.4097 Acc: 0.0925\n",
      "\n",
      "Epoch 85/199\n",
      "----------\n",
      "train Loss: 3.1536 Acc: 0.3373\n",
      "val Loss: 4.4129 Acc: 0.0866\n",
      "\n",
      "Epoch 86/199\n",
      "----------\n",
      "train Loss: 3.1160 Acc: 0.3759\n",
      "val Loss: 4.3909 Acc: 0.0896\n",
      "\n",
      "Epoch 87/199\n",
      "----------\n",
      "train Loss: 3.0694 Acc: 0.3663\n",
      "val Loss: 4.4191 Acc: 0.0866\n",
      "\n",
      "Epoch 88/199\n",
      "----------\n",
      "train Loss: 3.0672 Acc: 0.3904\n",
      "val Loss: 4.4062 Acc: 0.0866\n",
      "\n",
      "Epoch 89/199\n",
      "----------\n",
      "train Loss: 3.0863 Acc: 0.3904\n",
      "val Loss: 4.4149 Acc: 0.0866\n",
      "\n",
      "Epoch 90/199\n",
      "----------\n",
      "train Loss: 3.0775 Acc: 0.3807\n",
      "val Loss: 4.4223 Acc: 0.0866\n",
      "\n",
      "Epoch 91/199\n",
      "----------\n",
      "train Loss: 2.9864 Acc: 0.3904\n",
      "val Loss: 4.4246 Acc: 0.0806\n",
      "\n",
      "Epoch 92/199\n",
      "----------\n",
      "train Loss: 3.0258 Acc: 0.3590\n",
      "val Loss: 4.4358 Acc: 0.0896\n",
      "\n",
      "Epoch 93/199\n",
      "----------\n",
      "train Loss: 2.9291 Acc: 0.3976\n",
      "val Loss: 4.4357 Acc: 0.0746\n",
      "\n",
      "Epoch 94/199\n",
      "----------\n",
      "train Loss: 2.9839 Acc: 0.3952\n",
      "val Loss: 4.4146 Acc: 0.0955\n",
      "\n",
      "Epoch 95/199\n",
      "----------\n",
      "train Loss: 2.9367 Acc: 0.4048\n",
      "val Loss: 4.4139 Acc: 0.0925\n",
      "\n",
      "Epoch 96/199\n",
      "----------\n",
      "train Loss: 2.9114 Acc: 0.4096\n",
      "val Loss: 4.4462 Acc: 0.0806\n",
      "\n",
      "Epoch 97/199\n",
      "----------\n",
      "train Loss: 2.8606 Acc: 0.4361\n",
      "val Loss: 4.4281 Acc: 0.0836\n",
      "\n",
      "Epoch 98/199\n",
      "----------\n",
      "train Loss: 2.9057 Acc: 0.4120\n",
      "val Loss: 4.4319 Acc: 0.0776\n",
      "\n",
      "Epoch 99/199\n",
      "----------\n",
      "train Loss: 2.8388 Acc: 0.4434\n",
      "val Loss: 4.4305 Acc: 0.0746\n",
      "\n",
      "Epoch 100/199\n",
      "----------\n",
      "train Loss: 2.8485 Acc: 0.4434\n",
      "val Loss: 4.4138 Acc: 0.0896\n",
      "\n",
      "Epoch 101/199\n",
      "----------\n",
      "train Loss: 2.8242 Acc: 0.4458\n",
      "val Loss: 4.4340 Acc: 0.0836\n",
      "\n",
      "Epoch 102/199\n",
      "----------\n",
      "train Loss: 2.8419 Acc: 0.4289\n",
      "val Loss: 4.4270 Acc: 0.0896\n",
      "\n",
      "Epoch 103/199\n",
      "----------\n",
      "train Loss: 2.8043 Acc: 0.4217\n",
      "val Loss: 4.4548 Acc: 0.0836\n",
      "\n",
      "Epoch 104/199\n",
      "----------\n",
      "train Loss: 2.7871 Acc: 0.4096\n",
      "val Loss: 4.4391 Acc: 0.0925\n",
      "\n",
      "Epoch 105/199\n",
      "----------\n",
      "train Loss: 2.7978 Acc: 0.4361\n",
      "val Loss: 4.4493 Acc: 0.0806\n",
      "\n",
      "Epoch 106/199\n",
      "----------\n",
      "train Loss: 2.7362 Acc: 0.4723\n",
      "val Loss: 4.4434 Acc: 0.0925\n",
      "\n",
      "Epoch 107/199\n",
      "----------\n",
      "train Loss: 2.7415 Acc: 0.4506\n",
      "val Loss: 4.4615 Acc: 0.0925\n",
      "\n",
      "Epoch 108/199\n",
      "----------\n",
      "train Loss: 2.7262 Acc: 0.4410\n",
      "val Loss: 4.4761 Acc: 0.0866\n",
      "\n",
      "Epoch 109/199\n",
      "----------\n",
      "train Loss: 2.6573 Acc: 0.4771\n",
      "val Loss: 4.4724 Acc: 0.0896\n",
      "\n",
      "Epoch 110/199\n",
      "----------\n",
      "train Loss: 2.6686 Acc: 0.4867\n",
      "val Loss: 4.4859 Acc: 0.0866\n",
      "\n",
      "Epoch 111/199\n",
      "----------\n",
      "train Loss: 2.6910 Acc: 0.4940\n",
      "val Loss: 4.4750 Acc: 0.0896\n",
      "\n",
      "Epoch 112/199\n",
      "----------\n",
      "train Loss: 2.5913 Acc: 0.4964\n",
      "val Loss: 4.4834 Acc: 0.0776\n",
      "\n",
      "Epoch 113/199\n",
      "----------\n",
      "train Loss: 2.5836 Acc: 0.5012\n",
      "val Loss: 4.4787 Acc: 0.0925\n",
      "\n",
      "Epoch 114/199\n",
      "----------\n",
      "train Loss: 2.5413 Acc: 0.5205\n",
      "val Loss: 4.4678 Acc: 0.0985\n",
      "\n",
      "Epoch 115/199\n",
      "----------\n",
      "train Loss: 2.5823 Acc: 0.5012\n",
      "val Loss: 4.5032 Acc: 0.0806\n",
      "\n",
      "Epoch 116/199\n",
      "----------\n",
      "train Loss: 2.6356 Acc: 0.5012\n",
      "val Loss: 4.4992 Acc: 0.0866\n",
      "\n",
      "Epoch 117/199\n",
      "----------\n",
      "train Loss: 2.5159 Acc: 0.5157\n",
      "val Loss: 4.4960 Acc: 0.0925\n",
      "\n",
      "Epoch 118/199\n",
      "----------\n",
      "train Loss: 2.5535 Acc: 0.4819\n",
      "val Loss: 4.4886 Acc: 0.0925\n",
      "\n",
      "Epoch 119/199\n",
      "----------\n",
      "train Loss: 2.5226 Acc: 0.4964\n",
      "val Loss: 4.4886 Acc: 0.0866\n",
      "\n",
      "Epoch 120/199\n",
      "----------\n",
      "train Loss: 2.4492 Acc: 0.5060\n",
      "val Loss: 4.4668 Acc: 0.0896\n",
      "\n",
      "Epoch 121/199\n",
      "----------\n",
      "train Loss: 2.4878 Acc: 0.5012\n",
      "val Loss: 4.5022 Acc: 0.0836\n",
      "\n",
      "Epoch 122/199\n",
      "----------\n",
      "train Loss: 2.3929 Acc: 0.5422\n",
      "val Loss: 4.5033 Acc: 0.0896\n",
      "\n",
      "Epoch 123/199\n",
      "----------\n",
      "train Loss: 2.4784 Acc: 0.4964\n",
      "val Loss: 4.4654 Acc: 0.0896\n",
      "\n",
      "Epoch 124/199\n",
      "----------\n",
      "train Loss: 2.3821 Acc: 0.5470\n",
      "val Loss: 4.4972 Acc: 0.0925\n",
      "\n",
      "Epoch 125/199\n",
      "----------\n",
      "train Loss: 2.3773 Acc: 0.5590\n",
      "val Loss: 4.4913 Acc: 0.0896\n",
      "\n",
      "Epoch 126/199\n",
      "----------\n",
      "train Loss: 2.4068 Acc: 0.5301\n",
      "val Loss: 4.5093 Acc: 0.0896\n",
      "\n",
      "Epoch 127/199\n",
      "----------\n",
      "train Loss: 2.3632 Acc: 0.5301\n",
      "val Loss: 4.4926 Acc: 0.0925\n",
      "\n",
      "Epoch 128/199\n",
      "----------\n",
      "train Loss: 2.3666 Acc: 0.5253\n",
      "val Loss: 4.5165 Acc: 0.0866\n",
      "\n",
      "Epoch 129/199\n",
      "----------\n",
      "train Loss: 2.3047 Acc: 0.5398\n",
      "val Loss: 4.5073 Acc: 0.1015\n",
      "\n",
      "Epoch 130/199\n",
      "----------\n",
      "train Loss: 2.3601 Acc: 0.5639\n",
      "val Loss: 4.5145 Acc: 0.0955\n",
      "\n",
      "Epoch 131/199\n",
      "----------\n",
      "train Loss: 2.2632 Acc: 0.5855\n",
      "val Loss: 4.5153 Acc: 0.0985\n",
      "\n",
      "Epoch 132/199\n",
      "----------\n",
      "train Loss: 2.3225 Acc: 0.5301\n",
      "val Loss: 4.5629 Acc: 0.0896\n",
      "\n",
      "Epoch 133/199\n",
      "----------\n",
      "train Loss: 2.2677 Acc: 0.5446\n",
      "val Loss: 4.5635 Acc: 0.0955\n",
      "\n",
      "Epoch 134/199\n",
      "----------\n",
      "train Loss: 2.2429 Acc: 0.5446\n",
      "val Loss: 4.5295 Acc: 0.1045\n",
      "\n",
      "Epoch 135/199\n",
      "----------\n",
      "train Loss: 2.2780 Acc: 0.5614\n",
      "val Loss: 4.5610 Acc: 0.0955\n",
      "\n",
      "Epoch 136/199\n",
      "----------\n",
      "train Loss: 2.1769 Acc: 0.5831\n",
      "val Loss: 4.5292 Acc: 0.0955\n",
      "\n",
      "Epoch 137/199\n",
      "----------\n",
      "train Loss: 2.2333 Acc: 0.5952\n",
      "val Loss: 4.5518 Acc: 0.0985\n",
      "\n",
      "Epoch 138/199\n",
      "----------\n",
      "train Loss: 2.2344 Acc: 0.5687\n",
      "val Loss: 4.5514 Acc: 0.0955\n",
      "\n",
      "Epoch 139/199\n",
      "----------\n",
      "train Loss: 2.2101 Acc: 0.5783\n",
      "val Loss: 4.5713 Acc: 0.1045\n",
      "\n",
      "Epoch 140/199\n",
      "----------\n",
      "train Loss: 2.1748 Acc: 0.6145\n",
      "val Loss: 4.5308 Acc: 0.1015\n",
      "\n",
      "Epoch 141/199\n",
      "----------\n",
      "train Loss: 2.2069 Acc: 0.5542\n",
      "val Loss: 4.5546 Acc: 0.1015\n",
      "\n",
      "Epoch 142/199\n",
      "----------\n",
      "train Loss: 2.1130 Acc: 0.6072\n",
      "val Loss: 4.5318 Acc: 0.0955\n",
      "\n",
      "Epoch 143/199\n",
      "----------\n",
      "train Loss: 2.0676 Acc: 0.6169\n",
      "val Loss: 4.5541 Acc: 0.0985\n",
      "\n",
      "Epoch 144/199\n",
      "----------\n",
      "train Loss: 2.0268 Acc: 0.6265\n",
      "val Loss: 4.5681 Acc: 0.1045\n",
      "\n",
      "Epoch 145/199\n",
      "----------\n",
      "train Loss: 2.1262 Acc: 0.5590\n",
      "val Loss: 4.5908 Acc: 0.1075\n",
      "\n",
      "Epoch 146/199\n",
      "----------\n",
      "train Loss: 2.0343 Acc: 0.6000\n",
      "val Loss: 4.5780 Acc: 0.1075\n",
      "\n",
      "Epoch 147/199\n",
      "----------\n",
      "train Loss: 2.1364 Acc: 0.5518\n",
      "val Loss: 4.5830 Acc: 0.0896\n",
      "\n",
      "Epoch 148/199\n",
      "----------\n",
      "train Loss: 1.9627 Acc: 0.6554\n",
      "val Loss: 4.5923 Acc: 0.1015\n",
      "\n",
      "Epoch 149/199\n",
      "----------\n",
      "train Loss: 2.0254 Acc: 0.5976\n",
      "val Loss: 4.5819 Acc: 0.0896\n",
      "\n",
      "Epoch 150/199\n",
      "----------\n",
      "train Loss: 1.9755 Acc: 0.6530\n",
      "val Loss: 4.5970 Acc: 0.1075\n",
      "\n",
      "Epoch 151/199\n",
      "----------\n",
      "train Loss: 2.1002 Acc: 0.5952\n",
      "val Loss: 4.5397 Acc: 0.0925\n",
      "\n",
      "Epoch 152/199\n",
      "----------\n",
      "train Loss: 2.0705 Acc: 0.6241\n",
      "val Loss: 4.5573 Acc: 0.0925\n",
      "\n",
      "Epoch 153/199\n",
      "----------\n",
      "train Loss: 1.9223 Acc: 0.6313\n",
      "val Loss: 4.6544 Acc: 0.1015\n",
      "\n",
      "Epoch 154/199\n",
      "----------\n",
      "train Loss: 2.0225 Acc: 0.6120\n",
      "val Loss: 4.5768 Acc: 0.0955\n",
      "\n",
      "Epoch 155/199\n",
      "----------\n",
      "train Loss: 1.9666 Acc: 0.6554\n",
      "val Loss: 4.5781 Acc: 0.1104\n",
      "\n",
      "Epoch 156/199\n",
      "----------\n",
      "train Loss: 1.9442 Acc: 0.6386\n",
      "val Loss: 4.5921 Acc: 0.1164\n",
      "\n",
      "Epoch 157/199\n",
      "----------\n",
      "train Loss: 1.9414 Acc: 0.6410\n",
      "val Loss: 4.6029 Acc: 0.1104\n",
      "\n",
      "Epoch 158/199\n",
      "----------\n",
      "train Loss: 1.9592 Acc: 0.6313\n",
      "val Loss: 4.5962 Acc: 0.1104\n",
      "\n",
      "Epoch 159/199\n",
      "----------\n",
      "train Loss: 1.8338 Acc: 0.6627\n",
      "val Loss: 4.6393 Acc: 0.1015\n",
      "\n",
      "Epoch 160/199\n",
      "----------\n",
      "train Loss: 1.7846 Acc: 0.7205\n",
      "val Loss: 4.6413 Acc: 0.1104\n",
      "\n",
      "Epoch 161/199\n",
      "----------\n",
      "train Loss: 1.8542 Acc: 0.6747\n",
      "val Loss: 4.6304 Acc: 0.1104\n",
      "\n",
      "Epoch 162/199\n",
      "----------\n",
      "train Loss: 1.8055 Acc: 0.6675\n",
      "val Loss: 4.6108 Acc: 0.1134\n",
      "\n",
      "Epoch 163/199\n",
      "----------\n",
      "train Loss: 1.7784 Acc: 0.6964\n",
      "val Loss: 4.6231 Acc: 0.1134\n",
      "\n",
      "Epoch 164/199\n",
      "----------\n",
      "train Loss: 1.7630 Acc: 0.7181\n",
      "val Loss: 4.6375 Acc: 0.0925\n",
      "\n",
      "Epoch 165/199\n",
      "----------\n",
      "train Loss: 1.8025 Acc: 0.6988\n",
      "val Loss: 4.6229 Acc: 0.1075\n",
      "\n",
      "Epoch 166/199\n",
      "----------\n",
      "train Loss: 1.7410 Acc: 0.6916\n",
      "val Loss: 4.6179 Acc: 0.1015\n",
      "\n",
      "Epoch 167/199\n",
      "----------\n",
      "train Loss: 1.6906 Acc: 0.7229\n",
      "val Loss: 4.6429 Acc: 0.1134\n",
      "\n",
      "Epoch 168/199\n",
      "----------\n",
      "train Loss: 1.7111 Acc: 0.6867\n",
      "val Loss: 4.6617 Acc: 0.1045\n",
      "\n",
      "Epoch 169/199\n",
      "----------\n",
      "train Loss: 1.6818 Acc: 0.7181\n",
      "val Loss: 4.6239 Acc: 0.1104\n",
      "\n",
      "Epoch 170/199\n",
      "----------\n",
      "train Loss: 1.6041 Acc: 0.7398\n",
      "val Loss: 4.6250 Acc: 0.1104\n",
      "\n",
      "Epoch 171/199\n",
      "----------\n",
      "train Loss: 1.7550 Acc: 0.6771\n",
      "val Loss: 4.6670 Acc: 0.1075\n",
      "\n",
      "Epoch 172/199\n",
      "----------\n",
      "train Loss: 1.7093 Acc: 0.6747\n",
      "val Loss: 4.6518 Acc: 0.1045\n",
      "\n",
      "Epoch 173/199\n",
      "----------\n",
      "train Loss: 1.6392 Acc: 0.7205\n",
      "val Loss: 4.6822 Acc: 0.1134\n",
      "\n",
      "Epoch 174/199\n",
      "----------\n",
      "train Loss: 1.7355 Acc: 0.6747\n",
      "val Loss: 4.6983 Acc: 0.1134\n",
      "\n",
      "Epoch 175/199\n",
      "----------\n",
      "train Loss: 1.6559 Acc: 0.7060\n",
      "val Loss: 4.7315 Acc: 0.0955\n",
      "\n",
      "Epoch 176/199\n",
      "----------\n",
      "train Loss: 1.6671 Acc: 0.6892\n",
      "val Loss: 4.6766 Acc: 0.1194\n",
      "\n",
      "Epoch 177/199\n",
      "----------\n",
      "train Loss: 1.6168 Acc: 0.7133\n",
      "val Loss: 4.6980 Acc: 0.1194\n",
      "\n",
      "Epoch 178/199\n",
      "----------\n",
      "train Loss: 1.7252 Acc: 0.6602\n",
      "val Loss: 4.7026 Acc: 0.1134\n",
      "\n",
      "Epoch 179/199\n",
      "----------\n",
      "train Loss: 1.6043 Acc: 0.7373\n",
      "val Loss: 4.6995 Acc: 0.1104\n",
      "\n",
      "Epoch 180/199\n",
      "----------\n",
      "train Loss: 1.5568 Acc: 0.7277\n",
      "val Loss: 4.7128 Acc: 0.1134\n",
      "\n",
      "Epoch 181/199\n",
      "----------\n",
      "train Loss: 1.5153 Acc: 0.7711\n",
      "val Loss: 4.7293 Acc: 0.1075\n",
      "\n",
      "Epoch 182/199\n",
      "----------\n",
      "train Loss: 1.6542 Acc: 0.7108\n",
      "val Loss: 4.6966 Acc: 0.1134\n",
      "\n",
      "Epoch 183/199\n",
      "----------\n",
      "train Loss: 1.5287 Acc: 0.7542\n",
      "val Loss: 4.7204 Acc: 0.1104\n",
      "\n",
      "Epoch 184/199\n",
      "----------\n",
      "train Loss: 1.4987 Acc: 0.7470\n",
      "val Loss: 4.7037 Acc: 0.1104\n",
      "\n",
      "Epoch 185/199\n",
      "----------\n",
      "train Loss: 1.6178 Acc: 0.6843\n",
      "val Loss: 4.7026 Acc: 0.1164\n",
      "\n",
      "Epoch 186/199\n",
      "----------\n",
      "train Loss: 1.5321 Acc: 0.7470\n",
      "val Loss: 4.6953 Acc: 0.1164\n",
      "\n",
      "Epoch 187/199\n",
      "----------\n",
      "train Loss: 1.4265 Acc: 0.7663\n",
      "val Loss: 4.7298 Acc: 0.1045\n",
      "\n",
      "Epoch 188/199\n",
      "----------\n",
      "train Loss: 1.5868 Acc: 0.7157\n",
      "val Loss: 4.6959 Acc: 0.1224\n",
      "\n",
      "Epoch 189/199\n",
      "----------\n",
      "train Loss: 1.4622 Acc: 0.7542\n",
      "val Loss: 4.7248 Acc: 0.1164\n",
      "\n",
      "Epoch 190/199\n",
      "----------\n",
      "train Loss: 1.4407 Acc: 0.7711\n",
      "val Loss: 4.6806 Acc: 0.1254\n",
      "\n",
      "Epoch 191/199\n",
      "----------\n",
      "train Loss: 1.4903 Acc: 0.7518\n",
      "val Loss: 4.7079 Acc: 0.1224\n",
      "\n",
      "Epoch 192/199\n",
      "----------\n",
      "train Loss: 1.4915 Acc: 0.7373\n",
      "val Loss: 4.7478 Acc: 0.1224\n",
      "\n",
      "Epoch 193/199\n",
      "----------\n",
      "train Loss: 1.4295 Acc: 0.7759\n",
      "val Loss: 4.7073 Acc: 0.1254\n",
      "\n",
      "Epoch 194/199\n",
      "----------\n",
      "train Loss: 1.3997 Acc: 0.7831\n",
      "val Loss: 4.7063 Acc: 0.1104\n",
      "\n",
      "Epoch 195/199\n",
      "----------\n",
      "train Loss: 1.3244 Acc: 0.7831\n",
      "val Loss: 4.6849 Acc: 0.1194\n",
      "\n",
      "Epoch 196/199\n",
      "----------\n",
      "train Loss: 1.4473 Acc: 0.7590\n",
      "val Loss: 4.7589 Acc: 0.1224\n",
      "\n",
      "Epoch 197/199\n",
      "----------\n",
      "train Loss: 1.4001 Acc: 0.7398\n",
      "val Loss: 4.7701 Acc: 0.1194\n",
      "\n",
      "Epoch 198/199\n",
      "----------\n",
      "train Loss: 1.3842 Acc: 0.7831\n",
      "val Loss: 4.7345 Acc: 0.1224\n",
      "\n",
      "Epoch 199/199\n",
      "----------\n",
      "train Loss: 1.3721 Acc: 0.7446\n",
      "val Loss: 4.7187 Acc: 0.1164\n",
      "\n",
      "Training complete in 48m 54s\n",
      "Best val Acc: 0.125373\n"
     ]
    }
   ],
   "source": [
    "model = CNNEarDetector(model=\"resnet\", num_classes=train_data.get_num_classes(), extract_features=False)\n",
    "\n",
    "# Setup the loss fxn\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "\n",
    "# Train and evaluate\n",
    "num_epochs = 200\n",
    "hist = model.train(dataloaders_dict, criterion, num_epochs=num_epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "fuWQeEEgo03v",
   "metadata": {
    "id": "fuWQeEEgo03v"
   },
   "outputs": [],
   "source": [
    "model.save(os.getcwd(), \"trained_resnet_sharpening\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "F6aXcwbmo3Ps",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "executionInfo": {
     "elapsed": 451,
     "status": "ok",
     "timestamp": 1642408088539,
     "user": {
      "displayName": "Domen VreĹˇ",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "08720259634886868903"
     },
     "user_tz": -60
    },
    "id": "F6aXcwbmo3Ps",
    "outputId": "0c9d68b9-c003-4183-9508-1150b75eb14a"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "val_hist = [h.cpu().numpy() for h in hist]\n",
    "\n",
    "plt.title(\"Validation Accuracy vs. Number of Training Epochs\")\n",
    "plt.xlabel(\"Training Epochs\")\n",
    "plt.ylabel(\"Validation Accuracy\")\n",
    "plt.plot(range(1,num_epochs+1), val_hist)\n",
    "plt.savefig(os.getcwd()+\"/resnet_val_sharpening.png\")\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "resnet.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Assigment2",
   "language": "python",
   "name": "assigment2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}