Car Racing/notebooks/basic_CNN_edges.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# imports\n",
    "import os\n",
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "from torchvision import transforms\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision.datasets import ImageFolder\n",
    "from torch.utils.data import DataLoader, random_split\n",
    "from sklearn.metrics import classification_report\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using device: cpu\n"
     ]
    }
   ],
   "source": [
    "# Imposta il device\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "print(f\"Using device: {device}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Percorso del dataset (immagini preprocessate con data augmentation)\n",
    "dataset_dir = \"./train_balanced_edges\"\n",
    "\n",
    "# Trasformazioni: Normalizzazione e Preprocessing finale\n",
    "transform = transforms.Compose([\n",
    "    transforms.Grayscale(),\n",
    "    transforms.ToTensor(),  # Conversione a tensore\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Numero di immagini nel train set: 1476\n",
      "Numero di immagini nel test set: 369\n",
      "Classi disponibili: ['0', '1', '2', '3', '4']\n"
     ]
    }
   ],
   "source": [
    "# Creazione del Dataset\n",
    "dataset = ImageFolder(root=dataset_dir, transform=transform)\n",
    "\n",
    "# Suddivisione Train/Test\n",
    "train_size = int(0.8 * len(dataset))\n",
    "test_size = len(dataset) - train_size\n",
    "train_dataset, test_dataset = random_split(dataset, [train_size, test_size])\n",
    "\n",
    "# DataLoaders\n",
    "train_loader = DataLoader(train_dataset, batch_size=32, shuffle= False)\n",
    "test_loader = DataLoader(test_dataset, batch_size=32, shuffle=False)\n",
    "\n",
    "# Controllo rapido\n",
    "print(f\"Numero di immagini nel train set: {len(train_dataset)}\")\n",
    "print(f\"Numero di immagini nel test set: {len(test_dataset)}\")\n",
    "print(f\"Classi disponibili: {dataset.classes}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "class SimpleCNN(nn.Module):\n",
    "    def __init__(self, dropout_rate=0.25):\n",
    "        super(SimpleCNN, self).__init__()\n",
    "\n",
    "        # Convolutional Section\n",
    "        self.network = nn.Sequential(\n",
    "            \n",
    "            nn.Conv2d(1, 8, kernel_size=3, padding=1),\n",
    "            nn.BatchNorm2d(8),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(2,2), # 48x48\n",
    "\n",
    "            nn.Conv2d(8, 16, kernel_size=3, padding=1),\n",
    "            nn.BatchNorm2d(16),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(2, 2),   # 24x24\n",
    "\n",
    "            nn.Conv2d(16,32, kernel_size=3, padding=1),\n",
    "            nn.BatchNorm2d(32),\n",
    "            nn.ReLU(),\n",
    "\n",
    "            nn.Dropout(dropout_rate),  # Dropout for regularization\n",
    "        )\n",
    "\n",
    "        # Fully Connected Section\n",
    "        self.fc = nn.Sequential(\n",
    "            nn.AdaptiveAvgPool2d(1),\n",
    "            nn.Flatten(),\n",
    "            nn.Linear(32, 5),  # Fully connected\n",
    "        )\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.network(x)\n",
    "        x = self.fc(x)\n",
    "        return x\n",
    "    \n",
    "model = SimpleCNN().to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#criterion = nn.CrossEntropyLoss()\n",
    "#optimizer = optim.Adam(model.parameters(), lr=0.001, weight_decay=1e-4)\n",
    "#optimizer = torch.optim.RMSprop(model.parameters(), lr=0.001)\n",
    "#scheduler = optim.lr_scheduler.StepLR(optimizer, step_size=3, gamma=0.5) \n",
    "\n",
    "# Loop di training\n",
    "def train(model, loader, criterion, optimizer):\n",
    "    model.train()\n",
    "    running_loss = 0\n",
    "    correct = 0\n",
    "    for inputs, labels in loader:\n",
    "        inputs, labels = inputs.to(device), labels.to(device)\n",
    "        \n",
    "        optimizer.zero_grad()\n",
    "        outputs = model(inputs)\n",
    "        loss = criterion(outputs, labels)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        #scheduler.step()\n",
    "\n",
    "        running_loss += loss.item()\n",
    "        _, preds = outputs.max(1)\n",
    "        correct += (preds == labels).sum().item()\n",
    "    \n",
    "    accuracy = 100 * correct / len(loader.dataset)\n",
    "    return running_loss / len(loader), accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate(model, loader):\n",
    "    model.eval()\n",
    "    y_true = []\n",
    "    y_pred = []\n",
    "\n",
    "    with torch.no_grad():\n",
    "        for inputs, labels in loader:\n",
    "            inputs, labels = inputs.to(device), labels.to(device)\n",
    "            outputs = model(inputs)\n",
    "            _, preds = outputs.max(1)\n",
    "            y_true.extend(labels.cpu().numpy())\n",
    "            y_pred.extend(preds.cpu().numpy())\n",
    "    \n",
    "    return y_true, y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\nnum_epochs = 10\\n\\nfor epoch in range(num_epochs):\\n    train_loss, train_acc = train(model, train_loader, criterion, optimizer)\\n    print(f\"Epoch {epoch+1}/{num_epochs}: Loss: {train_loss:.4f}, Accuracy: {train_acc:.2f}%\")\\n\\ntest_set = ImageFolder(root=\\'./test_edges\\', transform=transform)\\ntest_set_loader = DataLoader(test_set, batch_size=32, shuffle=False)\\n\\ny_true, y_pred = evaluate(model, test_set_loader)\\n\\nprint(\"\\nClassification Report:\")\\nprint(classification_report(y_true, y_pred, target_names=train_dataset.dataset.classes))\\n'"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''\n",
    "num_epochs = 10\n",
    "\n",
    "for epoch in range(num_epochs):\n",
    "    train_loss, train_acc = train(model, train_loader, criterion, optimizer)\n",
    "    print(f\"Epoch {epoch+1}/{num_epochs}: Loss: {train_loss:.4f}, Accuracy: {train_acc:.2f}%\")\n",
    "\n",
    "test_set = ImageFolder(root='./test_edges', transform=transform)\n",
    "test_set_loader = DataLoader(test_set, batch_size=32, shuffle=False)\n",
    "\n",
    "y_true, y_pred = evaluate(model, test_set_loader)\n",
    "\n",
    "print(\"\\nClassification Report:\")\n",
    "print(classification_report(y_true, y_pred, target_names=train_dataset.dataset.classes))\n",
    "'''"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Parameter Search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import precision_score, recall_score, f1_score, confusion_matrix, classification_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_set = ImageFolder(root='./test_edges', transform=transform)\n",
    "test_set_loader = DataLoader(test_set, batch_size=32, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Parametri di ricerca\n",
    "epoch_list = [5, 10, 15, 20]\n",
    "learning_rate_list = [0.01, 0.001, 0.0001]\n",
    "dropout_list = [0, 0.25, 0.5]\n",
    "weight_decay_list = [0.01, 0.001, 0.0001]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training with dropout 0, lr 0.01, wd 0.01, epochs 5...\n",
      "Epoch 1/5, Loss: 71.4279, Accuracy: 32.93%\n",
      "Epoch 2/5, Loss: 68.9869, Accuracy: 36.65%\n",
      "Epoch 3/5, Loss: 68.5426, Accuracy: 38.21%\n",
      "Epoch 4/5, Loss: 68.1609, Accuracy: 38.14%\n",
      "Epoch 5/5, Loss: 68.1197, Accuracy: 37.74%\n",
      "Training with dropout 0, lr 0.01, wd 0.01, epochs 10...\n",
      "Epoch 1/10, Loss: 71.4934, Accuracy: 33.13%\n",
      "Epoch 2/10, Loss: 68.6847, Accuracy: 37.74%\n",
      "Epoch 3/10, Loss: 68.2462, Accuracy: 39.02%\n",
      "Epoch 4/10, Loss: 68.2442, Accuracy: 39.09%\n",
      "Epoch 5/10, Loss: 68.0418, Accuracy: 38.14%\n",
      "Epoch 6/10, Loss: 67.9452, Accuracy: 37.74%\n",
      "Epoch 7/10, Loss: 67.9288, Accuracy: 38.62%\n",
      "Epoch 8/10, Loss: 67.7803, Accuracy: 38.75%\n",
      "Epoch 9/10, Loss: 67.7865, Accuracy: 38.48%\n",
      "Epoch 10/10, Loss: 67.5856, Accuracy: 38.82%\n",
      "Training with dropout 0, lr 0.01, wd 0.01, epochs 15...\n",
      "Epoch 1/15, Loss: 71.6296, Accuracy: 33.74%\n",
      "Epoch 2/15, Loss: 68.6094, Accuracy: 37.47%\n",
      "Epoch 3/15, Loss: 68.4313, Accuracy: 37.47%\n",
      "Epoch 4/15, Loss: 68.3015, Accuracy: 38.14%\n",
      "Epoch 5/15, Loss: 68.1790, Accuracy: 38.08%\n",
      "Epoch 6/15, Loss: 68.0517, Accuracy: 38.14%\n",
      "Epoch 7/15, Loss: 67.8562, Accuracy: 38.28%\n",
      "Epoch 8/15, Loss: 67.8490, Accuracy: 38.14%\n",
      "Epoch 9/15, Loss: 67.9036, Accuracy: 37.67%\n",
      "Epoch 10/15, Loss: 67.8780, Accuracy: 38.01%\n",
      "Epoch 11/15, Loss: 67.7133, Accuracy: 38.14%\n",
      "Epoch 12/15, Loss: 67.5582, Accuracy: 38.41%\n",
      "Epoch 13/15, Loss: 67.5110, Accuracy: 38.75%\n",
      "Epoch 14/15, Loss: 67.3871, Accuracy: 38.62%\n",
      "Epoch 15/15, Loss: 67.3313, Accuracy: 38.82%\n",
      "Training with dropout 0, lr 0.01, wd 0.01, epochs 20...\n",
      "Epoch 1/20, Loss: 71.4407, Accuracy: 34.69%\n",
      "Epoch 2/20, Loss: 68.8843, Accuracy: 37.13%\n",
      "Epoch 3/20, Loss: 68.5440, Accuracy: 37.06%\n",
      "Epoch 4/20, Loss: 68.0713, Accuracy: 38.62%\n",
      "Epoch 5/20, Loss: 67.8657, Accuracy: 38.48%\n",
      "Epoch 6/20, Loss: 67.8506, Accuracy: 38.28%\n",
      "Epoch 7/20, Loss: 67.6471, Accuracy: 38.41%\n",
      "Epoch 8/20, Loss: 67.7531, Accuracy: 38.89%\n",
      "Epoch 9/20, Loss: 67.8441, Accuracy: 38.48%\n",
      "Epoch 10/20, Loss: 67.7596, Accuracy: 37.80%\n",
      "Epoch 11/20, Loss: 67.5856, Accuracy: 38.14%\n",
      "Epoch 12/20, Loss: 67.5572, Accuracy: 37.94%\n",
      "Epoch 13/20, Loss: 67.6002, Accuracy: 38.89%\n",
      "Epoch 14/20, Loss: 67.5243, Accuracy: 37.33%\n",
      "Epoch 15/20, Loss: 67.4775, Accuracy: 39.09%\n",
      "Epoch 16/20, Loss: 67.5886, Accuracy: 37.47%\n",
      "Epoch 17/20, Loss: 67.3713, Accuracy: 38.62%\n",
      "Epoch 18/20, Loss: 67.8125, Accuracy: 37.67%\n",
      "Epoch 19/20, Loss: 67.7078, Accuracy: 38.08%\n",
      "Epoch 20/20, Loss: 67.5380, Accuracy: 38.55%\n",
      "Training with dropout 0, lr 0.01, wd 0.001, epochs 5...\n",
      "Epoch 1/5, Loss: 71.1000, Accuracy: 33.20%\n",
      "Epoch 2/5, Loss: 67.9595, Accuracy: 36.38%\n",
      "Epoch 3/5, Loss: 67.3215, Accuracy: 37.06%\n",
      "Epoch 4/5, Loss: 66.9709, Accuracy: 37.80%\n",
      "Epoch 5/5, Loss: 66.6353, Accuracy: 38.35%\n",
      "Training with dropout 0, lr 0.01, wd 0.001, epochs 10...\n",
      "Epoch 1/10, Loss: 70.6030, Accuracy: 33.60%\n",
      "Epoch 2/10, Loss: 67.2518, Accuracy: 38.14%\n",
      "Epoch 3/10, Loss: 66.3547, Accuracy: 38.69%\n",
      "Epoch 4/10, Loss: 66.0468, Accuracy: 39.30%\n",
      "Epoch 5/10, Loss: 65.6320, Accuracy: 39.57%\n",
      "Epoch 6/10, Loss: 65.4333, Accuracy: 39.91%\n",
      "Epoch 7/10, Loss: 65.2680, Accuracy: 39.23%\n",
      "Epoch 8/10, Loss: 65.0368, Accuracy: 40.11%\n",
      "Epoch 9/10, Loss: 64.8445, Accuracy: 39.70%\n",
      "Epoch 10/10, Loss: 64.6329, Accuracy: 39.30%\n",
      "Training with dropout 0, lr 0.01, wd 0.001, epochs 15...\n",
      "Epoch 1/15, Loss: 71.3437, Accuracy: 31.37%\n",
      "Epoch 2/15, Loss: 67.6823, Accuracy: 37.26%\n",
      "Epoch 3/15, Loss: 67.0010, Accuracy: 38.21%\n",
      "Epoch 4/15, Loss: 66.5534, Accuracy: 38.28%\n",
      "Epoch 5/15, Loss: 66.2945, Accuracy: 38.62%\n",
      "Epoch 6/15, Loss: 66.0627, Accuracy: 39.30%\n",
      "Epoch 7/15, Loss: 65.6200, Accuracy: 40.45%\n",
      "Epoch 8/15, Loss: 65.3675, Accuracy: 39.84%\n",
      "Epoch 9/15, Loss: 65.1421, Accuracy: 40.18%\n",
      "Epoch 10/15, Loss: 64.8226, Accuracy: 40.92%\n",
      "Epoch 11/15, Loss: 64.7191, Accuracy: 41.46%\n",
      "Epoch 12/15, Loss: 64.7208, Accuracy: 39.77%\n",
      "Epoch 13/15, Loss: 64.3590, Accuracy: 41.12%\n",
      "Epoch 14/15, Loss: 64.0679, Accuracy: 41.46%\n",
      "Epoch 15/15, Loss: 63.8335, Accuracy: 42.01%\n",
      "Training with dropout 0, lr 0.01, wd 0.001, epochs 20...\n",
      "Epoch 1/20, Loss: 71.0835, Accuracy: 32.52%\n",
      "Epoch 2/20, Loss: 67.8761, Accuracy: 37.06%\n",
      "Epoch 3/20, Loss: 67.2736, Accuracy: 37.20%\n",
      "Epoch 4/20, Loss: 66.8818, Accuracy: 37.53%\n",
      "Epoch 5/20, Loss: 66.6546, Accuracy: 37.94%\n",
      "Epoch 6/20, Loss: 66.1925, Accuracy: 38.08%\n",
      "Epoch 7/20, Loss: 65.9344, Accuracy: 38.62%\n",
      "Epoch 8/20, Loss: 65.5844, Accuracy: 37.94%\n",
      "Epoch 9/20, Loss: 65.3501, Accuracy: 38.41%\n",
      "Epoch 10/20, Loss: 65.1133, Accuracy: 38.89%\n",
      "Epoch 11/20, Loss: 65.0345, Accuracy: 38.48%\n",
      "Epoch 12/20, Loss: 64.9988, Accuracy: 39.36%\n",
      "Epoch 13/20, Loss: 64.6033, Accuracy: 40.45%\n",
      "Epoch 14/20, Loss: 64.4606, Accuracy: 39.09%\n",
      "Epoch 15/20, Loss: 64.2335, Accuracy: 40.24%\n",
      "Epoch 16/20, Loss: 64.3879, Accuracy: 39.63%\n",
      "Epoch 17/20, Loss: 64.2054, Accuracy: 40.04%\n",
      "Epoch 18/20, Loss: 63.9773, Accuracy: 40.65%\n",
      "Epoch 19/20, Loss: 63.8473, Accuracy: 41.19%\n",
      "Epoch 20/20, Loss: 63.7198, Accuracy: 41.06%\n",
      "Training with dropout 0, lr 0.01, wd 0.0001, epochs 5...\n",
      "Epoch 1/5, Loss: 70.3658, Accuracy: 34.01%\n",
      "Epoch 2/5, Loss: 66.9607, Accuracy: 38.14%\n",
      "Epoch 3/5, Loss: 66.3413, Accuracy: 38.82%\n",
      "Epoch 4/5, Loss: 65.8775, Accuracy: 38.62%\n",
      "Epoch 5/5, Loss: 65.6614, Accuracy: 38.55%\n",
      "Training with dropout 0, lr 0.01, wd 0.0001, epochs 10...\n",
      "Epoch 1/10, Loss: 70.9266, Accuracy: 33.27%\n",
      "Epoch 2/10, Loss: 67.2036, Accuracy: 38.21%\n",
      "Epoch 3/10, Loss: 66.3149, Accuracy: 37.60%\n",
      "Epoch 4/10, Loss: 65.7570, Accuracy: 38.75%\n",
      "Epoch 5/10, Loss: 65.3456, Accuracy: 38.41%\n",
      "Epoch 6/10, Loss: 64.8693, Accuracy: 40.04%\n",
      "Epoch 7/10, Loss: 64.3519, Accuracy: 40.85%\n",
      "Epoch 8/10, Loss: 63.7289, Accuracy: 42.14%\n",
      "Epoch 9/10, Loss: 63.2257, Accuracy: 42.75%\n",
      "Epoch 10/10, Loss: 62.8532, Accuracy: 42.95%\n",
      "Training with dropout 0, lr 0.01, wd 0.0001, epochs 15...\n",
      "Epoch 1/15, Loss: 71.0370, Accuracy: 32.52%\n",
      "Epoch 2/15, Loss: 67.6794, Accuracy: 36.65%\n",
      "Epoch 3/15, Loss: 66.7616, Accuracy: 38.14%\n",
      "Epoch 4/15, Loss: 66.1349, Accuracy: 38.28%\n",
      "Epoch 5/15, Loss: 65.6859, Accuracy: 38.48%\n",
      "Epoch 6/15, Loss: 65.4542, Accuracy: 38.82%\n",
      "Epoch 7/15, Loss: 64.7534, Accuracy: 39.70%\n",
      "Epoch 8/15, Loss: 64.0642, Accuracy: 40.85%\n",
      "Epoch 9/15, Loss: 63.6767, Accuracy: 40.99%\n",
      "Epoch 10/15, Loss: 63.6418, Accuracy: 41.80%\n",
      "Epoch 11/15, Loss: 62.7443, Accuracy: 42.62%\n",
      "Epoch 12/15, Loss: 62.4856, Accuracy: 43.36%\n",
      "Epoch 13/15, Loss: 62.0398, Accuracy: 43.83%\n",
      "Epoch 14/15, Loss: 61.3396, Accuracy: 45.33%\n",
      "Epoch 15/15, Loss: 61.1014, Accuracy: 46.07%\n",
      "Training with dropout 0, lr 0.01, wd 0.0001, epochs 20...\n",
      "Epoch 1/20, Loss: 70.5764, Accuracy: 34.21%\n",
      "Epoch 2/20, Loss: 67.3869, Accuracy: 38.55%\n",
      "Epoch 3/20, Loss: 66.4496, Accuracy: 38.14%\n",
      "Epoch 4/20, Loss: 66.0120, Accuracy: 38.55%\n",
      "Epoch 5/20, Loss: 65.6469, Accuracy: 38.69%\n",
      "Epoch 6/20, Loss: 65.1423, Accuracy: 38.62%\n",
      "Epoch 7/20, Loss: 64.6605, Accuracy: 39.16%\n",
      "Epoch 8/20, Loss: 64.2632, Accuracy: 40.04%\n",
      "Epoch 9/20, Loss: 63.8642, Accuracy: 40.65%\n",
      "Epoch 10/20, Loss: 63.6356, Accuracy: 42.07%\n",
      "Epoch 11/20, Loss: 63.3006, Accuracy: 42.07%\n",
      "Epoch 12/20, Loss: 62.5473, Accuracy: 43.02%\n",
      "Epoch 13/20, Loss: 62.2368, Accuracy: 43.63%\n",
      "Epoch 14/20, Loss: 61.9032, Accuracy: 43.77%\n",
      "Epoch 15/20, Loss: 61.4665, Accuracy: 45.46%\n",
      "Epoch 16/20, Loss: 61.0407, Accuracy: 46.00%\n",
      "Epoch 17/20, Loss: 60.7216, Accuracy: 46.54%\n",
      "Epoch 18/20, Loss: 60.2557, Accuracy: 46.21%\n",
      "Epoch 19/20, Loss: 60.2081, Accuracy: 46.68%\n",
      "Epoch 20/20, Loss: 59.7426, Accuracy: 46.48%\n",
      "Training with dropout 0, lr 0.001, wd 0.01, epochs 5...\n",
      "Epoch 1/5, Loss: 73.6400, Accuracy: 30.69%\n",
      "Epoch 2/5, Loss: 70.3032, Accuracy: 39.09%\n",
      "Epoch 3/5, Loss: 68.7768, Accuracy: 39.30%\n",
      "Epoch 4/5, Loss: 67.9490, Accuracy: 39.77%\n",
      "Epoch 5/5, Loss: 67.3681, Accuracy: 39.91%\n",
      "Training with dropout 0, lr 0.001, wd 0.01, epochs 10...\n",
      "Epoch 1/10, Loss: 73.8766, Accuracy: 27.10%\n",
      "Epoch 2/10, Loss: 70.1411, Accuracy: 38.35%\n",
      "Epoch 3/10, Loss: 68.5195, Accuracy: 38.69%\n",
      "Epoch 4/10, Loss: 67.7689, Accuracy: 38.48%\n",
      "Epoch 5/10, Loss: 67.3242, Accuracy: 39.09%\n",
      "Epoch 6/10, Loss: 67.0536, Accuracy: 39.30%\n",
      "Epoch 7/10, Loss: 66.8786, Accuracy: 39.36%\n",
      "Epoch 8/10, Loss: 66.7004, Accuracy: 39.16%\n",
      "Epoch 9/10, Loss: 66.5986, Accuracy: 39.63%\n",
      "Epoch 10/10, Loss: 66.5321, Accuracy: 38.75%\n",
      "Training with dropout 0, lr 0.001, wd 0.01, epochs 15...\n",
      "Epoch 1/15, Loss: 73.4080, Accuracy: 29.74%\n",
      "Epoch 2/15, Loss: 69.9263, Accuracy: 38.62%\n",
      "Epoch 3/15, Loss: 68.3290, Accuracy: 39.77%\n",
      "Epoch 4/15, Loss: 67.6094, Accuracy: 40.31%\n",
      "Epoch 5/15, Loss: 67.1275, Accuracy: 40.38%\n",
      "Epoch 6/15, Loss: 66.7646, Accuracy: 40.18%\n",
      "Epoch 7/15, Loss: 66.4960, Accuracy: 40.65%\n",
      "Epoch 8/15, Loss: 66.2604, Accuracy: 40.45%\n",
      "Epoch 9/15, Loss: 66.0804, Accuracy: 40.11%\n",
      "Epoch 10/15, Loss: 65.9146, Accuracy: 40.51%\n",
      "Epoch 11/15, Loss: 65.7783, Accuracy: 40.65%\n",
      "Epoch 12/15, Loss: 65.6559, Accuracy: 40.51%\n",
      "Epoch 13/15, Loss: 65.5292, Accuracy: 40.72%\n",
      "Epoch 14/15, Loss: 65.4334, Accuracy: 40.65%\n",
      "Epoch 15/15, Loss: 65.3337, Accuracy: 41.33%\n",
      "Training with dropout 0, lr 0.001, wd 0.01, epochs 20...\n",
      "Epoch 1/20, Loss: 73.0060, Accuracy: 32.59%\n",
      "Epoch 2/20, Loss: 69.4727, Accuracy: 38.62%\n",
      "Epoch 3/20, Loss: 68.2769, Accuracy: 39.02%\n",
      "Epoch 4/20, Loss: 67.6245, Accuracy: 39.36%\n",
      "Epoch 5/20, Loss: 67.1608, Accuracy: 39.36%\n",
      "Epoch 6/20, Loss: 66.8212, Accuracy: 39.97%\n",
      "Epoch 7/20, Loss: 66.5726, Accuracy: 39.91%\n",
      "Epoch 8/20, Loss: 66.3751, Accuracy: 40.04%\n",
      "Epoch 9/20, Loss: 66.2608, Accuracy: 40.31%\n",
      "Epoch 10/20, Loss: 66.1167, Accuracy: 40.11%\n",
      "Epoch 11/20, Loss: 65.9252, Accuracy: 39.91%\n",
      "Epoch 12/20, Loss: 65.8280, Accuracy: 40.72%\n",
      "Epoch 13/20, Loss: 65.7626, Accuracy: 40.45%\n",
      "Epoch 14/20, Loss: 65.6294, Accuracy: 40.38%\n",
      "Epoch 15/20, Loss: 65.5379, Accuracy: 40.65%\n",
      "Epoch 16/20, Loss: 65.6235, Accuracy: 40.51%\n",
      "Epoch 17/20, Loss: 65.5168, Accuracy: 40.38%\n",
      "Epoch 18/20, Loss: 65.4492, Accuracy: 40.38%\n",
      "Epoch 19/20, Loss: 65.1829, Accuracy: 41.12%\n",
      "Epoch 20/20, Loss: 65.1548, Accuracy: 41.06%\n",
      "Training with dropout 0, lr 0.001, wd 0.001, epochs 5...\n",
      "Epoch 1/5, Loss: 73.0329, Accuracy: 29.00%\n",
      "Epoch 2/5, Loss: 69.4026, Accuracy: 39.30%\n",
      "Epoch 3/5, Loss: 67.9022, Accuracy: 40.24%\n",
      "Epoch 4/5, Loss: 67.0625, Accuracy: 40.24%\n",
      "Epoch 5/5, Loss: 66.4942, Accuracy: 39.91%\n",
      "Training with dropout 0, lr 0.001, wd 0.001, epochs 10...\n",
      "Epoch 1/10, Loss: 72.6729, Accuracy: 32.25%\n",
      "Epoch 2/10, Loss: 69.6168, Accuracy: 37.80%\n",
      "Epoch 3/10, Loss: 68.0961, Accuracy: 38.55%\n",
      "Epoch 4/10, Loss: 67.2055, Accuracy: 38.89%\n",
      "Epoch 5/10, Loss: 66.6159, Accuracy: 38.48%\n",
      "Epoch 6/10, Loss: 66.1762, Accuracy: 38.96%\n",
      "Epoch 7/10, Loss: 65.8330, Accuracy: 39.02%\n",
      "Epoch 8/10, Loss: 65.5085, Accuracy: 39.63%\n",
      "Epoch 9/10, Loss: 65.2991, Accuracy: 39.50%\n",
      "Epoch 10/10, Loss: 65.0019, Accuracy: 39.70%\n",
      "Training with dropout 0, lr 0.001, wd 0.001, epochs 15...\n",
      "Epoch 1/15, Loss: 73.8261, Accuracy: 31.10%\n",
      "Epoch 2/15, Loss: 70.7524, Accuracy: 38.35%\n",
      "Epoch 3/15, Loss: 68.6512, Accuracy: 38.75%\n",
      "Epoch 4/15, Loss: 67.4400, Accuracy: 38.89%\n",
      "Epoch 5/15, Loss: 66.7115, Accuracy: 38.55%\n",
      "Epoch 6/15, Loss: 66.2230, Accuracy: 39.70%\n",
      "Epoch 7/15, Loss: 65.8437, Accuracy: 39.50%\n",
      "Epoch 8/15, Loss: 65.5131, Accuracy: 39.50%\n",
      "Epoch 9/15, Loss: 65.2094, Accuracy: 39.97%\n",
      "Epoch 10/15, Loss: 64.9249, Accuracy: 39.77%\n",
      "Epoch 11/15, Loss: 64.6585, Accuracy: 40.85%\n",
      "Epoch 12/15, Loss: 64.4019, Accuracy: 41.19%\n",
      "Epoch 13/15, Loss: 64.1406, Accuracy: 41.53%\n",
      "Epoch 14/15, Loss: 63.8702, Accuracy: 42.28%\n",
      "Epoch 15/15, Loss: 63.6083, Accuracy: 42.82%\n",
      "Training with dropout 0, lr 0.001, wd 0.001, epochs 20...\n",
      "Epoch 1/20, Loss: 73.5811, Accuracy: 27.85%\n",
      "Epoch 2/20, Loss: 70.4196, Accuracy: 36.72%\n",
      "Epoch 3/20, Loss: 68.4893, Accuracy: 38.41%\n",
      "Epoch 4/20, Loss: 67.4280, Accuracy: 39.30%\n",
      "Epoch 5/20, Loss: 66.8715, Accuracy: 39.70%\n",
      "Epoch 6/20, Loss: 66.5039, Accuracy: 39.50%\n",
      "Epoch 7/20, Loss: 66.1798, Accuracy: 39.30%\n",
      "Epoch 8/20, Loss: 65.9155, Accuracy: 39.36%\n",
      "Epoch 9/20, Loss: 65.6837, Accuracy: 39.57%\n",
      "Epoch 10/20, Loss: 65.4628, Accuracy: 39.97%\n",
      "Epoch 11/20, Loss: 65.2632, Accuracy: 39.77%\n",
      "Epoch 12/20, Loss: 65.0672, Accuracy: 39.91%\n",
      "Epoch 13/20, Loss: 64.8973, Accuracy: 40.04%\n",
      "Epoch 14/20, Loss: 64.7265, Accuracy: 40.18%\n",
      "Epoch 15/20, Loss: 64.5570, Accuracy: 40.45%\n",
      "Epoch 16/20, Loss: 64.3995, Accuracy: 41.06%\n",
      "Epoch 17/20, Loss: 64.2418, Accuracy: 41.06%\n",
      "Epoch 18/20, Loss: 64.1149, Accuracy: 41.33%\n",
      "Epoch 19/20, Loss: 63.9657, Accuracy: 42.07%\n",
      "Epoch 20/20, Loss: 63.8165, Accuracy: 42.34%\n",
      "Training with dropout 0, lr 0.001, wd 0.0001, epochs 5...\n",
      "Epoch 1/5, Loss: 73.4474, Accuracy: 31.71%\n",
      "Epoch 2/5, Loss: 69.4529, Accuracy: 38.69%\n",
      "Epoch 3/5, Loss: 67.8529, Accuracy: 38.08%\n",
      "Epoch 4/5, Loss: 67.0409, Accuracy: 37.94%\n",
      "Epoch 5/5, Loss: 66.5355, Accuracy: 38.41%\n",
      "Training with dropout 0, lr 0.001, wd 0.0001, epochs 10...\n",
      "Epoch 1/10, Loss: 72.8907, Accuracy: 31.71%\n",
      "Epoch 2/10, Loss: 69.1736, Accuracy: 39.09%\n",
      "Epoch 3/10, Loss: 67.7521, Accuracy: 39.30%\n",
      "Epoch 4/10, Loss: 67.0197, Accuracy: 39.43%\n",
      "Epoch 5/10, Loss: 66.5414, Accuracy: 39.57%\n",
      "Epoch 6/10, Loss: 66.1836, Accuracy: 39.02%\n",
      "Epoch 7/10, Loss: 65.8903, Accuracy: 39.50%\n",
      "Epoch 8/10, Loss: 65.6294, Accuracy: 39.77%\n",
      "Epoch 9/10, Loss: 65.3751, Accuracy: 39.84%\n",
      "Epoch 10/10, Loss: 65.1360, Accuracy: 39.91%\n",
      "Training with dropout 0, lr 0.001, wd 0.0001, epochs 15...\n",
      "Epoch 1/15, Loss: 74.5030, Accuracy: 27.37%\n",
      "Epoch 2/15, Loss: 70.3686, Accuracy: 37.94%\n",
      "Epoch 3/15, Loss: 68.5788, Accuracy: 38.75%\n",
      "Epoch 4/15, Loss: 67.5945, Accuracy: 39.36%\n",
      "Epoch 5/15, Loss: 66.9317, Accuracy: 39.30%\n",
      "Epoch 6/15, Loss: 66.4256, Accuracy: 39.30%\n",
      "Epoch 7/15, Loss: 65.9856, Accuracy: 39.57%\n",
      "Epoch 8/15, Loss: 65.6264, Accuracy: 39.84%\n",
      "Epoch 9/15, Loss: 65.3152, Accuracy: 39.63%\n",
      "Epoch 10/15, Loss: 65.0082, Accuracy: 39.84%\n",
      "Epoch 11/15, Loss: 64.7542, Accuracy: 40.18%\n",
      "Epoch 12/15, Loss: 64.5106, Accuracy: 40.51%\n",
      "Epoch 13/15, Loss: 64.2482, Accuracy: 41.06%\n",
      "Epoch 14/15, Loss: 64.0028, Accuracy: 41.73%\n",
      "Epoch 15/15, Loss: 63.7845, Accuracy: 41.53%\n",
      "Training with dropout 0, lr 0.001, wd 0.0001, epochs 20...\n",
      "Epoch 1/20, Loss: 73.2398, Accuracy: 30.42%\n",
      "Epoch 2/20, Loss: 69.9920, Accuracy: 38.62%\n",
      "Epoch 3/20, Loss: 68.2869, Accuracy: 39.23%\n",
      "Epoch 4/20, Loss: 67.3505, Accuracy: 39.36%\n",
      "Epoch 5/20, Loss: 66.7615, Accuracy: 39.50%\n",
      "Epoch 6/20, Loss: 66.2996, Accuracy: 39.43%\n",
      "Epoch 7/20, Loss: 65.9134, Accuracy: 39.91%\n",
      "Epoch 8/20, Loss: 65.5741, Accuracy: 40.45%\n",
      "Epoch 9/20, Loss: 65.2666, Accuracy: 40.72%\n",
      "Epoch 10/20, Loss: 64.9610, Accuracy: 40.58%\n",
      "Epoch 11/20, Loss: 64.6899, Accuracy: 40.92%\n",
      "Epoch 12/20, Loss: 64.4431, Accuracy: 41.60%\n",
      "Epoch 13/20, Loss: 64.2741, Accuracy: 42.01%\n",
      "Epoch 14/20, Loss: 64.0478, Accuracy: 42.01%\n",
      "Epoch 15/20, Loss: 63.7783, Accuracy: 42.55%\n",
      "Epoch 16/20, Loss: 63.4830, Accuracy: 42.48%\n",
      "Epoch 17/20, Loss: 63.2465, Accuracy: 43.02%\n",
      "Epoch 18/20, Loss: 63.0122, Accuracy: 43.22%\n",
      "Epoch 19/20, Loss: 62.7601, Accuracy: 43.29%\n",
      "Epoch 20/20, Loss: 62.5134, Accuracy: 43.36%\n",
      "Training with dropout 0, lr 0.0001, wd 0.01, epochs 5...\n",
      "Epoch 1/5, Loss: 76.0191, Accuracy: 20.46%\n",
      "Epoch 2/5, Loss: 75.0086, Accuracy: 21.61%\n",
      "Epoch 3/5, Loss: 74.1725, Accuracy: 23.78%\n",
      "Epoch 4/5, Loss: 73.4405, Accuracy: 27.30%\n",
      "Epoch 5/5, Loss: 72.8198, Accuracy: 30.49%\n",
      "Training with dropout 0, lr 0.0001, wd 0.01, epochs 10...\n",
      "Epoch 1/10, Loss: 75.7095, Accuracy: 20.80%\n",
      "Epoch 2/10, Loss: 74.6443, Accuracy: 29.67%\n",
      "Epoch 3/10, Loss: 73.6924, Accuracy: 33.88%\n",
      "Epoch 4/10, Loss: 72.9110, Accuracy: 36.86%\n",
      "Epoch 5/10, Loss: 72.2036, Accuracy: 37.74%\n",
      "Epoch 6/10, Loss: 71.5514, Accuracy: 39.02%\n",
      "Epoch 7/10, Loss: 70.9743, Accuracy: 39.36%\n",
      "Epoch 8/10, Loss: 70.4688, Accuracy: 39.63%\n",
      "Epoch 9/10, Loss: 70.0340, Accuracy: 39.77%\n",
      "Epoch 10/10, Loss: 69.6663, Accuracy: 39.97%\n",
      "Training with dropout 0, lr 0.0001, wd 0.01, epochs 15...\n",
      "Epoch 1/15, Loss: 75.1050, Accuracy: 26.76%\n",
      "Epoch 2/15, Loss: 74.1764, Accuracy: 31.17%\n",
      "Epoch 3/15, Loss: 73.5209, Accuracy: 32.99%\n",
      "Epoch 4/15, Loss: 72.8861, Accuracy: 34.49%\n",
      "Epoch 5/15, Loss: 72.2936, Accuracy: 35.91%\n",
      "Epoch 6/15, Loss: 71.7268, Accuracy: 36.65%\n",
      "Epoch 7/15, Loss: 71.1838, Accuracy: 36.99%\n",
      "Epoch 8/15, Loss: 70.6747, Accuracy: 37.13%\n",
      "Epoch 9/15, Loss: 70.2083, Accuracy: 37.94%\n",
      "Epoch 10/15, Loss: 69.7928, Accuracy: 38.14%\n",
      "Epoch 11/15, Loss: 69.4347, Accuracy: 38.69%\n",
      "Epoch 12/15, Loss: 69.1381, Accuracy: 38.41%\n",
      "Epoch 13/15, Loss: 68.8918, Accuracy: 38.62%\n",
      "Epoch 14/15, Loss: 68.6812, Accuracy: 38.55%\n",
      "Epoch 15/15, Loss: 68.4984, Accuracy: 38.82%\n",
      "Training with dropout 0, lr 0.0001, wd 0.01, epochs 20...\n",
      "Epoch 1/20, Loss: 75.6018, Accuracy: 20.33%\n",
      "Epoch 2/20, Loss: 74.6843, Accuracy: 20.05%\n",
      "Epoch 3/20, Loss: 73.9302, Accuracy: 23.58%\n",
      "Epoch 4/20, Loss: 73.1781, Accuracy: 31.17%\n",
      "Epoch 5/20, Loss: 72.5323, Accuracy: 34.49%\n",
      "Epoch 6/20, Loss: 71.9652, Accuracy: 36.72%\n",
      "Epoch 7/20, Loss: 71.4572, Accuracy: 36.92%\n",
      "Epoch 8/20, Loss: 71.0379, Accuracy: 37.20%\n",
      "Epoch 9/20, Loss: 70.6765, Accuracy: 37.13%\n",
      "Epoch 10/20, Loss: 70.3407, Accuracy: 37.20%\n",
      "Epoch 11/20, Loss: 70.0293, Accuracy: 37.74%\n",
      "Epoch 12/20, Loss: 69.7542, Accuracy: 37.94%\n",
      "Epoch 13/20, Loss: 69.5150, Accuracy: 38.28%\n",
      "Epoch 14/20, Loss: 69.3048, Accuracy: 38.69%\n",
      "Epoch 15/20, Loss: 69.1173, Accuracy: 38.69%\n",
      "Epoch 16/20, Loss: 68.9447, Accuracy: 38.75%\n",
      "Epoch 17/20, Loss: 68.7872, Accuracy: 38.96%\n",
      "Epoch 18/20, Loss: 68.6422, Accuracy: 39.09%\n",
      "Epoch 19/20, Loss: 68.5080, Accuracy: 39.43%\n",
      "Epoch 20/20, Loss: 68.3817, Accuracy: 39.50%\n",
      "Training with dropout 0, lr 0.0001, wd 0.001, epochs 5...\n",
      "Epoch 1/5, Loss: 75.6603, Accuracy: 21.75%\n",
      "Epoch 2/5, Loss: 74.8931, Accuracy: 21.14%\n",
      "Epoch 3/5, Loss: 74.2820, Accuracy: 25.61%\n",
      "Epoch 4/5, Loss: 73.7698, Accuracy: 26.96%\n",
      "Epoch 5/5, Loss: 73.3284, Accuracy: 27.98%\n",
      "Training with dropout 0, lr 0.0001, wd 0.001, epochs 10...\n",
      "Epoch 1/10, Loss: 76.3673, Accuracy: 20.46%\n",
      "Epoch 2/10, Loss: 75.3393, Accuracy: 20.46%\n",
      "Epoch 3/10, Loss: 74.5269, Accuracy: 21.21%\n",
      "Epoch 4/10, Loss: 73.7903, Accuracy: 28.93%\n",
      "Epoch 5/10, Loss: 73.0581, Accuracy: 35.64%\n",
      "Epoch 6/10, Loss: 72.3641, Accuracy: 36.99%\n",
      "Epoch 7/10, Loss: 71.6962, Accuracy: 37.80%\n",
      "Epoch 8/10, Loss: 71.1016, Accuracy: 38.62%\n",
      "Epoch 9/10, Loss: 70.5948, Accuracy: 38.89%\n",
      "Epoch 10/10, Loss: 70.1717, Accuracy: 39.09%\n",
      "Training with dropout 0, lr 0.0001, wd 0.001, epochs 15...\n",
      "Epoch 1/15, Loss: 75.4121, Accuracy: 23.85%\n",
      "Epoch 2/15, Loss: 74.6549, Accuracy: 24.19%\n",
      "Epoch 3/15, Loss: 73.9800, Accuracy: 25.54%\n",
      "Epoch 4/15, Loss: 73.3860, Accuracy: 29.40%\n",
      "Epoch 5/15, Loss: 72.8197, Accuracy: 34.21%\n",
      "Epoch 6/15, Loss: 72.2813, Accuracy: 36.11%\n",
      "Epoch 7/15, Loss: 71.8102, Accuracy: 37.53%\n",
      "Epoch 8/15, Loss: 71.3971, Accuracy: 37.74%\n",
      "Epoch 9/15, Loss: 71.0151, Accuracy: 37.74%\n",
      "Epoch 10/15, Loss: 70.6205, Accuracy: 38.01%\n",
      "Epoch 11/15, Loss: 70.2413, Accuracy: 38.62%\n",
      "Epoch 12/15, Loss: 69.9380, Accuracy: 38.96%\n",
      "Epoch 13/15, Loss: 69.6743, Accuracy: 39.02%\n",
      "Epoch 14/15, Loss: 69.4362, Accuracy: 39.09%\n",
      "Epoch 15/15, Loss: 69.2190, Accuracy: 39.16%\n",
      "Training with dropout 0, lr 0.0001, wd 0.001, epochs 20...\n",
      "Epoch 1/20, Loss: 75.7360, Accuracy: 22.22%\n",
      "Epoch 2/20, Loss: 75.0572, Accuracy: 23.17%\n",
      "Epoch 3/20, Loss: 74.4973, Accuracy: 28.18%\n",
      "Epoch 4/20, Loss: 73.9086, Accuracy: 34.62%\n",
      "Epoch 5/20, Loss: 73.2613, Accuracy: 36.25%\n",
      "Epoch 6/20, Loss: 72.6474, Accuracy: 37.26%\n",
      "Epoch 7/20, Loss: 72.1089, Accuracy: 37.13%\n",
      "Epoch 8/20, Loss: 71.6375, Accuracy: 38.28%\n",
      "Epoch 9/20, Loss: 71.2110, Accuracy: 39.30%\n",
      "Epoch 10/20, Loss: 70.8203, Accuracy: 39.43%\n",
      "Epoch 11/20, Loss: 70.4658, Accuracy: 39.57%\n",
      "Epoch 12/20, Loss: 70.1405, Accuracy: 39.77%\n",
      "Epoch 13/20, Loss: 69.8387, Accuracy: 39.70%\n",
      "Epoch 14/20, Loss: 69.5606, Accuracy: 39.70%\n",
      "Epoch 15/20, Loss: 69.3067, Accuracy: 39.91%\n",
      "Epoch 16/20, Loss: 69.0763, Accuracy: 40.04%\n",
      "Epoch 17/20, Loss: 68.8653, Accuracy: 40.11%\n",
      "Epoch 18/20, Loss: 68.6710, Accuracy: 40.24%\n",
      "Epoch 19/20, Loss: 68.4900, Accuracy: 40.18%\n",
      "Epoch 20/20, Loss: 68.3219, Accuracy: 40.24%\n",
      "Training with dropout 0, lr 0.0001, wd 0.0001, epochs 5...\n",
      "Epoch 1/5, Loss: 75.4412, Accuracy: 23.17%\n",
      "Epoch 2/5, Loss: 74.6917, Accuracy: 23.71%\n",
      "Epoch 3/5, Loss: 74.0382, Accuracy: 26.90%\n",
      "Epoch 4/5, Loss: 73.4077, Accuracy: 30.49%\n",
      "Epoch 5/5, Loss: 72.7713, Accuracy: 34.55%\n",
      "Training with dropout 0, lr 0.0001, wd 0.0001, epochs 10...\n",
      "Epoch 1/10, Loss: 75.5761, Accuracy: 21.82%\n",
      "Epoch 2/10, Loss: 74.8006, Accuracy: 28.66%\n",
      "Epoch 3/10, Loss: 74.2319, Accuracy: 30.22%\n",
      "Epoch 4/10, Loss: 73.7080, Accuracy: 31.71%\n",
      "Epoch 5/10, Loss: 73.1902, Accuracy: 34.21%\n",
      "Epoch 6/10, Loss: 72.6658, Accuracy: 35.91%\n",
      "Epoch 7/10, Loss: 72.1644, Accuracy: 36.18%\n",
      "Epoch 8/10, Loss: 71.6999, Accuracy: 37.47%\n",
      "Epoch 9/10, Loss: 71.2815, Accuracy: 38.48%\n",
      "Epoch 10/10, Loss: 70.9028, Accuracy: 38.69%\n",
      "Training with dropout 0, lr 0.0001, wd 0.0001, epochs 15...\n",
      "Epoch 1/15, Loss: 75.9706, Accuracy: 17.48%\n",
      "Epoch 2/15, Loss: 75.0057, Accuracy: 23.44%\n",
      "Epoch 3/15, Loss: 74.1993, Accuracy: 27.78%\n",
      "Epoch 4/15, Loss: 73.3856, Accuracy: 33.60%\n",
      "Epoch 5/15, Loss: 72.5634, Accuracy: 37.74%\n",
      "Epoch 6/15, Loss: 71.8041, Accuracy: 38.35%\n",
      "Epoch 7/15, Loss: 71.1370, Accuracy: 38.96%\n",
      "Epoch 8/15, Loss: 70.5509, Accuracy: 39.16%\n",
      "Epoch 9/15, Loss: 70.0625, Accuracy: 39.43%\n",
      "Epoch 10/15, Loss: 69.6716, Accuracy: 39.30%\n",
      "Epoch 11/15, Loss: 69.3451, Accuracy: 39.30%\n",
      "Epoch 12/15, Loss: 69.0640, Accuracy: 39.50%\n",
      "Epoch 13/15, Loss: 68.8187, Accuracy: 39.57%\n",
      "Epoch 14/15, Loss: 68.6017, Accuracy: 39.77%\n",
      "Epoch 15/15, Loss: 68.4061, Accuracy: 39.70%\n",
      "Training with dropout 0, lr 0.0001, wd 0.0001, epochs 20...\n",
      "Epoch 1/20, Loss: 75.2983, Accuracy: 21.14%\n",
      "Epoch 2/20, Loss: 74.5287, Accuracy: 25.27%\n",
      "Epoch 3/20, Loss: 73.8069, Accuracy: 33.13%\n",
      "Epoch 4/20, Loss: 73.1544, Accuracy: 36.99%\n",
      "Epoch 5/20, Loss: 72.5817, Accuracy: 37.33%\n",
      "Epoch 6/20, Loss: 72.0477, Accuracy: 37.80%\n",
      "Epoch 7/20, Loss: 71.5478, Accuracy: 38.21%\n",
      "Epoch 8/20, Loss: 71.0790, Accuracy: 38.01%\n",
      "Epoch 9/20, Loss: 70.6471, Accuracy: 38.21%\n",
      "Epoch 10/20, Loss: 70.2462, Accuracy: 38.55%\n",
      "Epoch 11/20, Loss: 69.8843, Accuracy: 39.09%\n",
      "Epoch 12/20, Loss: 69.5585, Accuracy: 39.09%\n",
      "Epoch 13/20, Loss: 69.2664, Accuracy: 39.09%\n",
      "Epoch 14/20, Loss: 69.0039, Accuracy: 39.23%\n",
      "Epoch 15/20, Loss: 68.7676, Accuracy: 39.23%\n",
      "Epoch 16/20, Loss: 68.5568, Accuracy: 39.36%\n",
      "Epoch 17/20, Loss: 68.3681, Accuracy: 39.16%\n",
      "Epoch 18/20, Loss: 68.1973, Accuracy: 39.50%\n",
      "Epoch 19/20, Loss: 68.0419, Accuracy: 39.50%\n",
      "Epoch 20/20, Loss: 67.9000, Accuracy: 39.57%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.01, epochs 5...\n",
      "Epoch 1/5, Loss: 70.8462, Accuracy: 34.01%\n",
      "Epoch 2/5, Loss: 68.8274, Accuracy: 36.59%\n",
      "Epoch 3/5, Loss: 68.5201, Accuracy: 37.47%\n",
      "Epoch 4/5, Loss: 68.2391, Accuracy: 38.08%\n",
      "Epoch 5/5, Loss: 68.2153, Accuracy: 38.21%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.01, epochs 10...\n",
      "Epoch 1/10, Loss: 71.8002, Accuracy: 32.66%\n",
      "Epoch 2/10, Loss: 69.0685, Accuracy: 36.59%\n",
      "Epoch 3/10, Loss: 68.4732, Accuracy: 38.01%\n",
      "Epoch 4/10, Loss: 68.2751, Accuracy: 37.60%\n",
      "Epoch 5/10, Loss: 68.0816, Accuracy: 37.87%\n",
      "Epoch 6/10, Loss: 67.8389, Accuracy: 38.01%\n",
      "Epoch 7/10, Loss: 68.2437, Accuracy: 37.80%\n",
      "Epoch 8/10, Loss: 68.0902, Accuracy: 37.20%\n",
      "Epoch 9/10, Loss: 68.0857, Accuracy: 37.53%\n",
      "Epoch 10/10, Loss: 67.7872, Accuracy: 38.89%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.01, epochs 15...\n",
      "Epoch 1/15, Loss: 71.2489, Accuracy: 33.06%\n",
      "Epoch 2/15, Loss: 68.6285, Accuracy: 37.20%\n",
      "Epoch 3/15, Loss: 68.3449, Accuracy: 37.87%\n",
      "Epoch 4/15, Loss: 68.1119, Accuracy: 39.02%\n",
      "Epoch 5/15, Loss: 67.9920, Accuracy: 38.55%\n",
      "Epoch 6/15, Loss: 67.7762, Accuracy: 39.23%\n",
      "Epoch 7/15, Loss: 67.9407, Accuracy: 38.21%\n",
      "Epoch 8/15, Loss: 67.8127, Accuracy: 38.21%\n",
      "Epoch 9/15, Loss: 67.8837, Accuracy: 39.02%\n",
      "Epoch 10/15, Loss: 67.6192, Accuracy: 38.75%\n",
      "Epoch 11/15, Loss: 67.7433, Accuracy: 38.96%\n",
      "Epoch 12/15, Loss: 67.5121, Accuracy: 38.89%\n",
      "Epoch 13/15, Loss: 67.5602, Accuracy: 38.08%\n",
      "Epoch 14/15, Loss: 67.5021, Accuracy: 38.89%\n",
      "Epoch 15/15, Loss: 67.3981, Accuracy: 38.35%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.01, epochs 20...\n",
      "Epoch 1/20, Loss: 70.6596, Accuracy: 34.49%\n",
      "Epoch 2/20, Loss: 68.4066, Accuracy: 37.53%\n",
      "Epoch 3/20, Loss: 68.2524, Accuracy: 37.13%\n",
      "Epoch 4/20, Loss: 68.1679, Accuracy: 38.41%\n",
      "Epoch 5/20, Loss: 68.0465, Accuracy: 38.21%\n",
      "Epoch 6/20, Loss: 67.8776, Accuracy: 38.69%\n",
      "Epoch 7/20, Loss: 67.9754, Accuracy: 37.87%\n",
      "Epoch 8/20, Loss: 67.7864, Accuracy: 38.69%\n",
      "Epoch 9/20, Loss: 67.6969, Accuracy: 39.16%\n",
      "Epoch 10/20, Loss: 67.5527, Accuracy: 38.75%\n",
      "Epoch 11/20, Loss: 67.8696, Accuracy: 38.69%\n",
      "Epoch 12/20, Loss: 67.6135, Accuracy: 38.82%\n",
      "Epoch 13/20, Loss: 67.7296, Accuracy: 38.14%\n",
      "Epoch 14/20, Loss: 67.3781, Accuracy: 38.55%\n",
      "Epoch 15/20, Loss: 67.7113, Accuracy: 38.55%\n",
      "Epoch 16/20, Loss: 67.3816, Accuracy: 38.69%\n",
      "Epoch 17/20, Loss: 67.4188, Accuracy: 38.41%\n",
      "Epoch 18/20, Loss: 67.3249, Accuracy: 38.55%\n",
      "Epoch 19/20, Loss: 67.4895, Accuracy: 38.62%\n",
      "Epoch 20/20, Loss: 67.5285, Accuracy: 38.48%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.001, epochs 5...\n",
      "Epoch 1/5, Loss: 70.9317, Accuracy: 33.60%\n",
      "Epoch 2/5, Loss: 67.4710, Accuracy: 37.80%\n",
      "Epoch 3/5, Loss: 66.8187, Accuracy: 38.82%\n",
      "Epoch 4/5, Loss: 66.1864, Accuracy: 39.43%\n",
      "Epoch 5/5, Loss: 65.9114, Accuracy: 39.77%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.001, epochs 10...\n",
      "Epoch 1/10, Loss: 71.0088, Accuracy: 32.66%\n",
      "Epoch 2/10, Loss: 67.8304, Accuracy: 36.52%\n",
      "Epoch 3/10, Loss: 67.0551, Accuracy: 37.33%\n",
      "Epoch 4/10, Loss: 66.5815, Accuracy: 37.80%\n",
      "Epoch 5/10, Loss: 66.2576, Accuracy: 38.55%\n",
      "Epoch 6/10, Loss: 65.7558, Accuracy: 39.02%\n",
      "Epoch 7/10, Loss: 65.5417, Accuracy: 39.36%\n",
      "Epoch 8/10, Loss: 65.3374, Accuracy: 39.63%\n",
      "Epoch 9/10, Loss: 65.2379, Accuracy: 38.89%\n",
      "Epoch 10/10, Loss: 65.1305, Accuracy: 39.57%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.001, epochs 15...\n",
      "Epoch 1/15, Loss: 70.5638, Accuracy: 34.08%\n",
      "Epoch 2/15, Loss: 67.3851, Accuracy: 38.14%\n",
      "Epoch 3/15, Loss: 66.9060, Accuracy: 38.35%\n",
      "Epoch 4/15, Loss: 66.6034, Accuracy: 37.74%\n",
      "Epoch 5/15, Loss: 66.4298, Accuracy: 37.74%\n",
      "Epoch 6/15, Loss: 65.9165, Accuracy: 38.41%\n",
      "Epoch 7/15, Loss: 65.7228, Accuracy: 38.69%\n",
      "Epoch 8/15, Loss: 65.5249, Accuracy: 39.57%\n",
      "Epoch 9/15, Loss: 65.3972, Accuracy: 39.36%\n",
      "Epoch 10/15, Loss: 65.0326, Accuracy: 39.70%\n",
      "Epoch 11/15, Loss: 65.0052, Accuracy: 40.24%\n",
      "Epoch 12/15, Loss: 64.5271, Accuracy: 39.97%\n",
      "Epoch 13/15, Loss: 64.3853, Accuracy: 40.65%\n",
      "Epoch 14/15, Loss: 64.2121, Accuracy: 41.40%\n",
      "Epoch 15/15, Loss: 64.2269, Accuracy: 40.85%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.001, epochs 20...\n",
      "Epoch 1/20, Loss: 70.2848, Accuracy: 35.43%\n",
      "Epoch 2/20, Loss: 67.4303, Accuracy: 37.20%\n",
      "Epoch 3/20, Loss: 66.9290, Accuracy: 37.47%\n",
      "Epoch 4/20, Loss: 66.4001, Accuracy: 38.89%\n",
      "Epoch 5/20, Loss: 66.0051, Accuracy: 39.30%\n",
      "Epoch 6/20, Loss: 65.5624, Accuracy: 39.57%\n",
      "Epoch 7/20, Loss: 65.5730, Accuracy: 39.16%\n",
      "Epoch 8/20, Loss: 65.0764, Accuracy: 39.77%\n",
      "Epoch 9/20, Loss: 64.7795, Accuracy: 40.72%\n",
      "Epoch 10/20, Loss: 64.7191, Accuracy: 40.24%\n",
      "Epoch 11/20, Loss: 64.3675, Accuracy: 41.06%\n",
      "Epoch 12/20, Loss: 64.0869, Accuracy: 40.85%\n",
      "Epoch 13/20, Loss: 63.8697, Accuracy: 40.99%\n",
      "Epoch 14/20, Loss: 63.6791, Accuracy: 42.01%\n",
      "Epoch 15/20, Loss: 63.5386, Accuracy: 42.01%\n",
      "Epoch 16/20, Loss: 63.2141, Accuracy: 42.89%\n",
      "Epoch 17/20, Loss: 63.1003, Accuracy: 41.67%\n",
      "Epoch 18/20, Loss: 63.0309, Accuracy: 43.56%\n",
      "Epoch 19/20, Loss: 62.9098, Accuracy: 43.09%\n",
      "Epoch 20/20, Loss: 62.2438, Accuracy: 43.70%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.0001, epochs 5...\n",
      "Epoch 1/5, Loss: 70.4100, Accuracy: 32.86%\n",
      "Epoch 2/5, Loss: 67.0697, Accuracy: 38.01%\n",
      "Epoch 3/5, Loss: 66.3752, Accuracy: 38.41%\n",
      "Epoch 4/5, Loss: 65.8489, Accuracy: 38.41%\n",
      "Epoch 5/5, Loss: 65.4631, Accuracy: 38.89%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.0001, epochs 10...\n",
      "Epoch 1/10, Loss: 71.0566, Accuracy: 33.20%\n",
      "Epoch 2/10, Loss: 67.6242, Accuracy: 37.13%\n",
      "Epoch 3/10, Loss: 66.9627, Accuracy: 38.01%\n",
      "Epoch 4/10, Loss: 66.3790, Accuracy: 38.21%\n",
      "Epoch 5/10, Loss: 65.8704, Accuracy: 37.87%\n",
      "Epoch 6/10, Loss: 65.3900, Accuracy: 39.36%\n",
      "Epoch 7/10, Loss: 65.0139, Accuracy: 39.36%\n",
      "Epoch 8/10, Loss: 64.4765, Accuracy: 39.30%\n",
      "Epoch 9/10, Loss: 64.2895, Accuracy: 39.43%\n",
      "Epoch 10/10, Loss: 63.9842, Accuracy: 40.38%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.0001, epochs 15...\n",
      "Epoch 1/15, Loss: 70.5787, Accuracy: 34.69%\n",
      "Epoch 2/15, Loss: 67.4181, Accuracy: 37.80%\n",
      "Epoch 3/15, Loss: 66.6394, Accuracy: 37.87%\n",
      "Epoch 4/15, Loss: 66.1320, Accuracy: 38.35%\n",
      "Epoch 5/15, Loss: 65.7055, Accuracy: 39.23%\n",
      "Epoch 6/15, Loss: 65.2259, Accuracy: 39.43%\n",
      "Epoch 7/15, Loss: 64.7858, Accuracy: 39.97%\n",
      "Epoch 8/15, Loss: 64.7354, Accuracy: 39.09%\n",
      "Epoch 9/15, Loss: 64.2647, Accuracy: 40.04%\n",
      "Epoch 10/15, Loss: 63.4873, Accuracy: 42.07%\n",
      "Epoch 11/15, Loss: 63.0682, Accuracy: 42.55%\n",
      "Epoch 12/15, Loss: 62.5007, Accuracy: 42.68%\n",
      "Epoch 13/15, Loss: 62.0075, Accuracy: 44.44%\n",
      "Epoch 14/15, Loss: 61.7760, Accuracy: 45.19%\n",
      "Epoch 15/15, Loss: 61.2570, Accuracy: 45.93%\n",
      "Training with dropout 0.25, lr 0.01, wd 0.0001, epochs 20...\n",
      "Epoch 1/20, Loss: 70.5816, Accuracy: 33.81%\n",
      "Epoch 2/20, Loss: 67.3471, Accuracy: 37.53%\n",
      "Epoch 3/20, Loss: 66.5179, Accuracy: 38.69%\n",
      "Epoch 4/20, Loss: 66.0959, Accuracy: 39.50%\n",
      "Epoch 5/20, Loss: 65.5703, Accuracy: 39.50%\n",
      "Epoch 6/20, Loss: 65.4635, Accuracy: 39.16%\n",
      "Epoch 7/20, Loss: 64.6080, Accuracy: 40.18%\n",
      "Epoch 8/20, Loss: 64.3185, Accuracy: 40.65%\n",
      "Epoch 9/20, Loss: 63.8873, Accuracy: 41.67%\n",
      "Epoch 10/20, Loss: 63.2257, Accuracy: 41.67%\n",
      "Epoch 11/20, Loss: 62.8569, Accuracy: 42.21%\n",
      "Epoch 12/20, Loss: 62.4997, Accuracy: 43.29%\n",
      "Epoch 13/20, Loss: 61.9898, Accuracy: 43.97%\n",
      "Epoch 14/20, Loss: 61.2958, Accuracy: 46.07%\n",
      "Epoch 15/20, Loss: 60.4749, Accuracy: 46.48%\n",
      "Epoch 16/20, Loss: 60.1008, Accuracy: 47.43%\n",
      "Epoch 17/20, Loss: 59.4539, Accuracy: 47.49%\n",
      "Epoch 18/20, Loss: 59.0882, Accuracy: 48.58%\n",
      "Epoch 19/20, Loss: 58.7824, Accuracy: 49.80%\n",
      "Epoch 20/20, Loss: 57.9969, Accuracy: 49.12%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.01, epochs 5...\n",
      "Epoch 1/5, Loss: 73.8865, Accuracy: 28.39%\n",
      "Epoch 2/5, Loss: 71.0732, Accuracy: 35.37%\n",
      "Epoch 3/5, Loss: 69.5263, Accuracy: 37.80%\n",
      "Epoch 4/5, Loss: 68.4571, Accuracy: 38.89%\n",
      "Epoch 5/5, Loss: 67.7434, Accuracy: 39.57%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.01, epochs 10...\n",
      "Epoch 1/10, Loss: 73.2793, Accuracy: 30.15%\n",
      "Epoch 2/10, Loss: 70.0641, Accuracy: 37.74%\n",
      "Epoch 3/10, Loss: 68.5390, Accuracy: 39.16%\n",
      "Epoch 4/10, Loss: 67.6552, Accuracy: 39.91%\n",
      "Epoch 5/10, Loss: 67.1224, Accuracy: 39.57%\n",
      "Epoch 6/10, Loss: 66.7475, Accuracy: 40.24%\n",
      "Epoch 7/10, Loss: 66.4738, Accuracy: 40.45%\n",
      "Epoch 8/10, Loss: 66.2584, Accuracy: 40.24%\n",
      "Epoch 9/10, Loss: 66.0752, Accuracy: 40.51%\n",
      "Epoch 10/10, Loss: 65.9054, Accuracy: 40.51%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.01, epochs 15...\n",
      "Epoch 1/15, Loss: 73.5094, Accuracy: 31.10%\n",
      "Epoch 2/15, Loss: 70.1830, Accuracy: 38.48%\n",
      "Epoch 3/15, Loss: 68.6994, Accuracy: 39.02%\n",
      "Epoch 4/15, Loss: 67.8848, Accuracy: 39.16%\n",
      "Epoch 5/15, Loss: 67.4382, Accuracy: 39.02%\n",
      "Epoch 6/15, Loss: 67.1251, Accuracy: 39.63%\n",
      "Epoch 7/15, Loss: 66.8648, Accuracy: 40.04%\n",
      "Epoch 8/15, Loss: 66.6305, Accuracy: 40.72%\n",
      "Epoch 9/15, Loss: 66.4502, Accuracy: 40.85%\n",
      "Epoch 10/15, Loss: 66.3384, Accuracy: 41.06%\n",
      "Epoch 11/15, Loss: 66.1892, Accuracy: 40.45%\n",
      "Epoch 12/15, Loss: 66.0255, Accuracy: 40.65%\n",
      "Epoch 13/15, Loss: 65.9084, Accuracy: 40.72%\n",
      "Epoch 14/15, Loss: 65.8003, Accuracy: 40.72%\n",
      "Epoch 15/15, Loss: 65.7201, Accuracy: 40.79%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.01, epochs 20...\n",
      "Epoch 1/20, Loss: 73.4762, Accuracy: 29.47%\n",
      "Epoch 2/20, Loss: 70.0217, Accuracy: 38.28%\n",
      "Epoch 3/20, Loss: 68.5035, Accuracy: 38.82%\n",
      "Epoch 4/20, Loss: 67.7796, Accuracy: 39.02%\n",
      "Epoch 5/20, Loss: 67.4033, Accuracy: 38.62%\n",
      "Epoch 6/20, Loss: 67.1510, Accuracy: 38.35%\n",
      "Epoch 7/20, Loss: 66.9566, Accuracy: 38.14%\n",
      "Epoch 8/20, Loss: 66.7972, Accuracy: 38.62%\n",
      "Epoch 9/20, Loss: 66.6475, Accuracy: 39.02%\n",
      "Epoch 10/20, Loss: 66.5400, Accuracy: 38.96%\n",
      "Epoch 11/20, Loss: 66.4577, Accuracy: 39.16%\n",
      "Epoch 12/20, Loss: 66.3689, Accuracy: 38.96%\n",
      "Epoch 13/20, Loss: 66.2754, Accuracy: 39.30%\n",
      "Epoch 14/20, Loss: 66.2411, Accuracy: 38.96%\n",
      "Epoch 15/20, Loss: 66.1749, Accuracy: 39.50%\n",
      "Epoch 16/20, Loss: 66.1382, Accuracy: 39.02%\n",
      "Epoch 17/20, Loss: 66.0727, Accuracy: 39.70%\n",
      "Epoch 18/20, Loss: 66.1216, Accuracy: 39.70%\n",
      "Epoch 19/20, Loss: 66.0728, Accuracy: 39.91%\n",
      "Epoch 20/20, Loss: 65.8882, Accuracy: 39.70%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.001, epochs 5...\n",
      "Epoch 1/5, Loss: 73.0617, Accuracy: 31.03%\n",
      "Epoch 2/5, Loss: 69.5055, Accuracy: 38.28%\n",
      "Epoch 3/5, Loss: 68.1503, Accuracy: 38.89%\n",
      "Epoch 4/5, Loss: 67.3869, Accuracy: 39.57%\n",
      "Epoch 5/5, Loss: 66.8365, Accuracy: 39.91%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.001, epochs 10...\n",
      "Epoch 1/10, Loss: 73.2422, Accuracy: 31.98%\n",
      "Epoch 2/10, Loss: 70.1711, Accuracy: 38.28%\n",
      "Epoch 3/10, Loss: 68.5482, Accuracy: 39.70%\n",
      "Epoch 4/10, Loss: 67.7176, Accuracy: 39.43%\n",
      "Epoch 5/10, Loss: 67.1927, Accuracy: 39.16%\n",
      "Epoch 6/10, Loss: 66.7778, Accuracy: 39.57%\n",
      "Epoch 7/10, Loss: 66.4446, Accuracy: 39.50%\n",
      "Epoch 8/10, Loss: 66.1212, Accuracy: 40.04%\n",
      "Epoch 9/10, Loss: 65.8556, Accuracy: 40.51%\n",
      "Epoch 10/10, Loss: 65.6124, Accuracy: 40.51%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.001, epochs 15...\n",
      "Epoch 1/15, Loss: 73.1914, Accuracy: 30.83%\n",
      "Epoch 2/15, Loss: 70.1690, Accuracy: 37.53%\n",
      "Epoch 3/15, Loss: 68.3726, Accuracy: 38.62%\n",
      "Epoch 4/15, Loss: 67.5818, Accuracy: 38.55%\n",
      "Epoch 5/15, Loss: 67.0439, Accuracy: 39.43%\n",
      "Epoch 6/15, Loss: 66.6763, Accuracy: 39.23%\n",
      "Epoch 7/15, Loss: 66.3506, Accuracy: 39.57%\n",
      "Epoch 8/15, Loss: 66.0440, Accuracy: 39.91%\n",
      "Epoch 9/15, Loss: 65.8341, Accuracy: 40.45%\n",
      "Epoch 10/15, Loss: 65.5932, Accuracy: 39.63%\n",
      "Epoch 11/15, Loss: 65.3481, Accuracy: 39.97%\n",
      "Epoch 12/15, Loss: 65.1367, Accuracy: 40.11%\n",
      "Epoch 13/15, Loss: 64.8937, Accuracy: 40.18%\n",
      "Epoch 14/15, Loss: 64.6665, Accuracy: 40.72%\n",
      "Epoch 15/15, Loss: 64.4309, Accuracy: 40.92%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.001, epochs 20...\n",
      "Epoch 1/20, Loss: 73.7688, Accuracy: 26.90%\n",
      "Epoch 2/20, Loss: 70.1120, Accuracy: 37.53%\n",
      "Epoch 3/20, Loss: 68.3130, Accuracy: 38.14%\n",
      "Epoch 4/20, Loss: 67.4790, Accuracy: 37.74%\n",
      "Epoch 5/20, Loss: 66.8859, Accuracy: 38.14%\n",
      "Epoch 6/20, Loss: 66.4565, Accuracy: 38.89%\n",
      "Epoch 7/20, Loss: 66.0705, Accuracy: 39.23%\n",
      "Epoch 8/20, Loss: 65.7887, Accuracy: 39.02%\n",
      "Epoch 9/20, Loss: 65.4999, Accuracy: 39.70%\n",
      "Epoch 10/20, Loss: 65.2198, Accuracy: 39.91%\n",
      "Epoch 11/20, Loss: 64.9717, Accuracy: 40.72%\n",
      "Epoch 12/20, Loss: 64.6612, Accuracy: 40.65%\n",
      "Epoch 13/20, Loss: 64.3829, Accuracy: 41.06%\n",
      "Epoch 14/20, Loss: 64.1089, Accuracy: 41.53%\n",
      "Epoch 15/20, Loss: 63.9201, Accuracy: 42.28%\n",
      "Epoch 16/20, Loss: 63.6817, Accuracy: 42.34%\n",
      "Epoch 17/20, Loss: 63.4115, Accuracy: 43.22%\n",
      "Epoch 18/20, Loss: 63.2108, Accuracy: 43.63%\n",
      "Epoch 19/20, Loss: 62.9858, Accuracy: 44.17%\n",
      "Epoch 20/20, Loss: 62.7591, Accuracy: 44.51%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.0001, epochs 5...\n",
      "Epoch 1/5, Loss: 72.5758, Accuracy: 32.18%\n",
      "Epoch 2/5, Loss: 69.1118, Accuracy: 38.48%\n",
      "Epoch 3/5, Loss: 67.8978, Accuracy: 38.35%\n",
      "Epoch 4/5, Loss: 67.1523, Accuracy: 38.08%\n",
      "Epoch 5/5, Loss: 66.6331, Accuracy: 38.21%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.0001, epochs 10...\n",
      "Epoch 1/10, Loss: 73.5018, Accuracy: 30.49%\n",
      "Epoch 2/10, Loss: 69.9824, Accuracy: 38.35%\n",
      "Epoch 3/10, Loss: 68.3425, Accuracy: 38.28%\n",
      "Epoch 4/10, Loss: 67.4904, Accuracy: 38.21%\n",
      "Epoch 5/10, Loss: 66.8873, Accuracy: 38.41%\n",
      "Epoch 6/10, Loss: 66.4363, Accuracy: 38.96%\n",
      "Epoch 7/10, Loss: 66.0150, Accuracy: 39.43%\n",
      "Epoch 8/10, Loss: 65.6646, Accuracy: 39.36%\n",
      "Epoch 9/10, Loss: 65.3633, Accuracy: 39.77%\n",
      "Epoch 10/10, Loss: 65.0248, Accuracy: 40.31%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.0001, epochs 15...\n",
      "Epoch 1/15, Loss: 73.3669, Accuracy: 32.86%\n",
      "Epoch 2/15, Loss: 69.7752, Accuracy: 37.13%\n",
      "Epoch 3/15, Loss: 68.0842, Accuracy: 38.28%\n",
      "Epoch 4/15, Loss: 67.1901, Accuracy: 38.35%\n",
      "Epoch 5/15, Loss: 66.6367, Accuracy: 38.14%\n",
      "Epoch 6/15, Loss: 66.1891, Accuracy: 38.35%\n",
      "Epoch 7/15, Loss: 65.8461, Accuracy: 38.96%\n",
      "Epoch 8/15, Loss: 65.4981, Accuracy: 39.43%\n",
      "Epoch 9/15, Loss: 65.1722, Accuracy: 39.70%\n",
      "Epoch 10/15, Loss: 64.8784, Accuracy: 40.18%\n",
      "Epoch 11/15, Loss: 64.6481, Accuracy: 40.38%\n",
      "Epoch 12/15, Loss: 64.3819, Accuracy: 41.06%\n",
      "Epoch 13/15, Loss: 64.0328, Accuracy: 40.99%\n",
      "Epoch 14/15, Loss: 63.8229, Accuracy: 42.07%\n",
      "Epoch 15/15, Loss: 63.5824, Accuracy: 42.01%\n",
      "Training with dropout 0.25, lr 0.001, wd 0.0001, epochs 20...\n",
      "Epoch 1/20, Loss: 72.7239, Accuracy: 30.42%\n",
      "Epoch 2/20, Loss: 69.3494, Accuracy: 38.28%\n",
      "Epoch 3/20, Loss: 67.9054, Accuracy: 38.69%\n",
      "Epoch 4/20, Loss: 67.1594, Accuracy: 38.82%\n",
      "Epoch 5/20, Loss: 66.7313, Accuracy: 38.96%\n",
      "Epoch 6/20, Loss: 66.3538, Accuracy: 39.36%\n",
      "Epoch 7/20, Loss: 66.0490, Accuracy: 39.16%\n",
      "Epoch 8/20, Loss: 65.7650, Accuracy: 39.50%\n",
      "Epoch 9/20, Loss: 65.4908, Accuracy: 39.77%\n",
      "Epoch 10/20, Loss: 65.2104, Accuracy: 39.97%\n",
      "Epoch 11/20, Loss: 64.9672, Accuracy: 40.51%\n",
      "Epoch 12/20, Loss: 64.7387, Accuracy: 40.45%\n",
      "Epoch 13/20, Loss: 64.5035, Accuracy: 40.65%\n",
      "Epoch 14/20, Loss: 64.2642, Accuracy: 41.12%\n",
      "Epoch 15/20, Loss: 64.1077, Accuracy: 40.65%\n",
      "Epoch 16/20, Loss: 63.7974, Accuracy: 41.26%\n",
      "Epoch 17/20, Loss: 63.6420, Accuracy: 40.99%\n",
      "Epoch 18/20, Loss: 63.4308, Accuracy: 41.06%\n",
      "Epoch 19/20, Loss: 63.1554, Accuracy: 42.21%\n",
      "Epoch 20/20, Loss: 62.9082, Accuracy: 42.75%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.01, epochs 5...\n",
      "Epoch 1/5, Loss: 76.0889, Accuracy: 19.44%\n",
      "Epoch 2/5, Loss: 75.0144, Accuracy: 21.75%\n",
      "Epoch 3/5, Loss: 74.2711, Accuracy: 27.30%\n",
      "Epoch 4/5, Loss: 73.4868, Accuracy: 29.81%\n",
      "Epoch 5/5, Loss: 72.8062, Accuracy: 32.59%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.01, epochs 10...\n",
      "Epoch 1/10, Loss: 75.5749, Accuracy: 20.73%\n",
      "Epoch 2/10, Loss: 74.6450, Accuracy: 27.71%\n",
      "Epoch 3/10, Loss: 73.8844, Accuracy: 29.61%\n",
      "Epoch 4/10, Loss: 73.1415, Accuracy: 31.37%\n",
      "Epoch 5/10, Loss: 72.4434, Accuracy: 32.32%\n",
      "Epoch 6/10, Loss: 71.8031, Accuracy: 32.32%\n",
      "Epoch 7/10, Loss: 71.2637, Accuracy: 33.20%\n",
      "Epoch 8/10, Loss: 70.8413, Accuracy: 34.89%\n",
      "Epoch 9/10, Loss: 70.4738, Accuracy: 35.03%\n",
      "Epoch 10/10, Loss: 70.1104, Accuracy: 35.98%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.01, epochs 15...\n",
      "Epoch 1/15, Loss: 76.4172, Accuracy: 20.66%\n",
      "Epoch 2/15, Loss: 75.2680, Accuracy: 21.75%\n",
      "Epoch 3/15, Loss: 74.3615, Accuracy: 27.85%\n",
      "Epoch 4/15, Loss: 73.5776, Accuracy: 33.74%\n",
      "Epoch 5/15, Loss: 72.8233, Accuracy: 35.09%\n",
      "Epoch 6/15, Loss: 72.1318, Accuracy: 36.04%\n",
      "Epoch 7/15, Loss: 71.5283, Accuracy: 36.59%\n",
      "Epoch 8/15, Loss: 71.0196, Accuracy: 38.21%\n",
      "Epoch 9/15, Loss: 70.5488, Accuracy: 39.09%\n",
      "Epoch 10/15, Loss: 70.1635, Accuracy: 39.30%\n",
      "Epoch 11/15, Loss: 69.7929, Accuracy: 39.09%\n",
      "Epoch 12/15, Loss: 69.4955, Accuracy: 39.23%\n",
      "Epoch 13/15, Loss: 69.2074, Accuracy: 39.09%\n",
      "Epoch 14/15, Loss: 68.9603, Accuracy: 38.96%\n",
      "Epoch 15/15, Loss: 68.7634, Accuracy: 39.09%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.01, epochs 20...\n",
      "Epoch 1/20, Loss: 76.6332, Accuracy: 20.33%\n",
      "Epoch 2/20, Loss: 75.2714, Accuracy: 20.39%\n",
      "Epoch 3/20, Loss: 74.3528, Accuracy: 21.00%\n",
      "Epoch 4/20, Loss: 73.5661, Accuracy: 23.10%\n",
      "Epoch 5/20, Loss: 72.8742, Accuracy: 25.41%\n",
      "Epoch 6/20, Loss: 72.3440, Accuracy: 26.83%\n",
      "Epoch 7/20, Loss: 71.8920, Accuracy: 28.05%\n",
      "Epoch 8/20, Loss: 71.5003, Accuracy: 29.88%\n",
      "Epoch 9/20, Loss: 71.1235, Accuracy: 32.45%\n",
      "Epoch 10/20, Loss: 70.8280, Accuracy: 34.15%\n",
      "Epoch 11/20, Loss: 70.5605, Accuracy: 34.62%\n",
      "Epoch 12/20, Loss: 70.3434, Accuracy: 35.98%\n",
      "Epoch 13/20, Loss: 70.0698, Accuracy: 36.65%\n",
      "Epoch 14/20, Loss: 69.8736, Accuracy: 37.13%\n",
      "Epoch 15/20, Loss: 69.6984, Accuracy: 37.47%\n",
      "Epoch 16/20, Loss: 69.4865, Accuracy: 37.53%\n",
      "Epoch 17/20, Loss: 69.3093, Accuracy: 37.80%\n",
      "Epoch 18/20, Loss: 69.1471, Accuracy: 37.74%\n",
      "Epoch 19/20, Loss: 68.9867, Accuracy: 37.74%\n",
      "Epoch 20/20, Loss: 68.8297, Accuracy: 38.14%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.001, epochs 5...\n",
      "Epoch 1/5, Loss: 76.2212, Accuracy: 22.15%\n",
      "Epoch 2/5, Loss: 75.3045, Accuracy: 22.97%\n",
      "Epoch 3/5, Loss: 74.4983, Accuracy: 23.31%\n",
      "Epoch 4/5, Loss: 73.7956, Accuracy: 24.32%\n",
      "Epoch 5/5, Loss: 73.2058, Accuracy: 25.88%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.001, epochs 10...\n",
      "Epoch 1/10, Loss: 75.4375, Accuracy: 20.05%\n",
      "Epoch 2/10, Loss: 74.4926, Accuracy: 26.15%\n",
      "Epoch 3/10, Loss: 73.8369, Accuracy: 33.81%\n",
      "Epoch 4/10, Loss: 73.2347, Accuracy: 34.69%\n",
      "Epoch 5/10, Loss: 72.7057, Accuracy: 35.57%\n",
      "Epoch 6/10, Loss: 72.2485, Accuracy: 35.03%\n",
      "Epoch 7/10, Loss: 71.7731, Accuracy: 36.18%\n",
      "Epoch 8/10, Loss: 71.3381, Accuracy: 36.18%\n",
      "Epoch 9/10, Loss: 70.9526, Accuracy: 36.45%\n",
      "Epoch 10/10, Loss: 70.6315, Accuracy: 37.13%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.001, epochs 15...\n",
      "Epoch 1/15, Loss: 75.5067, Accuracy: 18.90%\n",
      "Epoch 2/15, Loss: 74.3765, Accuracy: 23.24%\n",
      "Epoch 3/15, Loss: 73.3657, Accuracy: 31.23%\n",
      "Epoch 4/15, Loss: 72.4519, Accuracy: 36.18%\n",
      "Epoch 5/15, Loss: 71.6320, Accuracy: 36.52%\n",
      "Epoch 6/15, Loss: 70.9041, Accuracy: 37.20%\n",
      "Epoch 7/15, Loss: 70.2751, Accuracy: 37.33%\n",
      "Epoch 8/15, Loss: 69.7737, Accuracy: 37.80%\n",
      "Epoch 9/15, Loss: 69.3307, Accuracy: 38.21%\n",
      "Epoch 10/15, Loss: 69.0067, Accuracy: 38.28%\n",
      "Epoch 11/15, Loss: 68.7231, Accuracy: 38.21%\n",
      "Epoch 12/15, Loss: 68.4808, Accuracy: 38.48%\n",
      "Epoch 13/15, Loss: 68.2829, Accuracy: 38.75%\n",
      "Epoch 14/15, Loss: 68.1351, Accuracy: 38.14%\n",
      "Epoch 15/15, Loss: 67.9269, Accuracy: 38.96%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.001, epochs 20...\n",
      "Epoch 1/20, Loss: 75.3717, Accuracy: 24.80%\n",
      "Epoch 2/20, Loss: 74.6413, Accuracy: 33.60%\n",
      "Epoch 3/20, Loss: 74.0183, Accuracy: 37.53%\n",
      "Epoch 4/20, Loss: 73.4361, Accuracy: 38.96%\n",
      "Epoch 5/20, Loss: 72.8553, Accuracy: 38.55%\n",
      "Epoch 6/20, Loss: 72.2932, Accuracy: 39.02%\n",
      "Epoch 7/20, Loss: 71.7886, Accuracy: 39.09%\n",
      "Epoch 8/20, Loss: 71.3445, Accuracy: 38.96%\n",
      "Epoch 9/20, Loss: 70.9405, Accuracy: 39.02%\n",
      "Epoch 10/20, Loss: 70.5903, Accuracy: 39.02%\n",
      "Epoch 11/20, Loss: 70.2647, Accuracy: 39.30%\n",
      "Epoch 12/20, Loss: 70.0057, Accuracy: 39.57%\n",
      "Epoch 13/20, Loss: 69.7249, Accuracy: 39.57%\n",
      "Epoch 14/20, Loss: 69.4852, Accuracy: 39.43%\n",
      "Epoch 15/20, Loss: 69.2922, Accuracy: 39.84%\n",
      "Epoch 16/20, Loss: 69.0543, Accuracy: 39.77%\n",
      "Epoch 17/20, Loss: 68.8392, Accuracy: 40.24%\n",
      "Epoch 18/20, Loss: 68.6915, Accuracy: 40.11%\n",
      "Epoch 19/20, Loss: 68.4978, Accuracy: 40.11%\n",
      "Epoch 20/20, Loss: 68.3524, Accuracy: 40.38%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.0001, epochs 5...\n",
      "Epoch 1/5, Loss: 75.6450, Accuracy: 19.44%\n",
      "Epoch 2/5, Loss: 74.9095, Accuracy: 26.83%\n",
      "Epoch 3/5, Loss: 74.3393, Accuracy: 31.30%\n",
      "Epoch 4/5, Loss: 73.8443, Accuracy: 34.82%\n",
      "Epoch 5/5, Loss: 73.3810, Accuracy: 36.86%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.0001, epochs 10...\n",
      "Epoch 1/10, Loss: 76.0824, Accuracy: 28.79%\n",
      "Epoch 2/10, Loss: 75.1802, Accuracy: 33.54%\n",
      "Epoch 3/10, Loss: 74.4323, Accuracy: 35.03%\n",
      "Epoch 4/10, Loss: 73.7770, Accuracy: 38.01%\n",
      "Epoch 5/10, Loss: 73.1634, Accuracy: 38.69%\n",
      "Epoch 6/10, Loss: 72.5641, Accuracy: 39.84%\n",
      "Epoch 7/10, Loss: 71.9858, Accuracy: 39.91%\n",
      "Epoch 8/10, Loss: 71.4692, Accuracy: 39.77%\n",
      "Epoch 9/10, Loss: 70.9422, Accuracy: 39.63%\n",
      "Epoch 10/10, Loss: 70.5197, Accuracy: 39.57%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.0001, epochs 15...\n",
      "Epoch 1/15, Loss: 75.5691, Accuracy: 19.51%\n",
      "Epoch 2/15, Loss: 74.8124, Accuracy: 19.78%\n",
      "Epoch 3/15, Loss: 74.2177, Accuracy: 22.43%\n",
      "Epoch 4/15, Loss: 73.6516, Accuracy: 25.75%\n",
      "Epoch 5/15, Loss: 73.1280, Accuracy: 27.91%\n",
      "Epoch 6/15, Loss: 72.5919, Accuracy: 31.37%\n",
      "Epoch 7/15, Loss: 72.0744, Accuracy: 34.15%\n",
      "Epoch 8/15, Loss: 71.5815, Accuracy: 35.23%\n",
      "Epoch 9/15, Loss: 71.1565, Accuracy: 36.79%\n",
      "Epoch 10/15, Loss: 70.7245, Accuracy: 37.74%\n",
      "Epoch 11/15, Loss: 70.3427, Accuracy: 37.94%\n",
      "Epoch 12/15, Loss: 70.0029, Accuracy: 38.55%\n",
      "Epoch 13/15, Loss: 69.7092, Accuracy: 38.82%\n",
      "Epoch 14/15, Loss: 69.4555, Accuracy: 39.30%\n",
      "Epoch 15/15, Loss: 69.1673, Accuracy: 39.16%\n",
      "Training with dropout 0.25, lr 0.0001, wd 0.0001, epochs 20...\n",
      "Epoch 1/20, Loss: 75.1446, Accuracy: 21.41%\n",
      "Epoch 2/20, Loss: 74.3095, Accuracy: 27.37%\n",
      "Epoch 3/20, Loss: 73.7126, Accuracy: 31.17%\n",
      "Epoch 4/20, Loss: 73.1397, Accuracy: 34.28%\n",
      "Epoch 5/20, Loss: 72.5617, Accuracy: 35.23%\n",
      "Epoch 6/20, Loss: 71.9740, Accuracy: 36.25%\n",
      "Epoch 7/20, Loss: 71.4467, Accuracy: 37.20%\n",
      "Epoch 8/20, Loss: 70.9737, Accuracy: 37.33%\n",
      "Epoch 9/20, Loss: 70.5991, Accuracy: 37.60%\n",
      "Epoch 10/20, Loss: 70.1945, Accuracy: 38.28%\n",
      "Epoch 11/20, Loss: 69.8026, Accuracy: 38.35%\n",
      "Epoch 12/20, Loss: 69.4818, Accuracy: 38.14%\n",
      "Epoch 13/20, Loss: 69.1550, Accuracy: 38.96%\n",
      "Epoch 14/20, Loss: 68.8777, Accuracy: 38.82%\n",
      "Epoch 15/20, Loss: 68.6714, Accuracy: 39.09%\n",
      "Epoch 16/20, Loss: 68.4076, Accuracy: 38.82%\n",
      "Epoch 17/20, Loss: 68.2292, Accuracy: 38.89%\n",
      "Epoch 18/20, Loss: 68.0820, Accuracy: 39.23%\n",
      "Epoch 19/20, Loss: 67.9260, Accuracy: 39.36%\n",
      "Epoch 20/20, Loss: 67.7603, Accuracy: 38.96%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.01, epochs 5...\n",
      "Epoch 1/5, Loss: 71.4459, Accuracy: 34.08%\n",
      "Epoch 2/5, Loss: 68.8472, Accuracy: 36.99%\n",
      "Epoch 3/5, Loss: 68.3768, Accuracy: 36.72%\n",
      "Epoch 4/5, Loss: 68.2750, Accuracy: 37.60%\n",
      "Epoch 5/5, Loss: 68.1825, Accuracy: 36.79%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.01, epochs 10...\n",
      "Epoch 1/10, Loss: 72.1676, Accuracy: 31.71%\n",
      "Epoch 2/10, Loss: 69.3058, Accuracy: 37.33%\n",
      "Epoch 3/10, Loss: 68.8435, Accuracy: 36.92%\n",
      "Epoch 4/10, Loss: 68.5980, Accuracy: 37.67%\n",
      "Epoch 5/10, Loss: 68.5283, Accuracy: 38.35%\n",
      "Epoch 6/10, Loss: 68.2950, Accuracy: 37.47%\n",
      "Epoch 7/10, Loss: 68.1545, Accuracy: 37.26%\n",
      "Epoch 8/10, Loss: 68.2908, Accuracy: 37.06%\n",
      "Epoch 9/10, Loss: 68.1297, Accuracy: 37.87%\n",
      "Epoch 10/10, Loss: 68.0917, Accuracy: 37.20%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.01, epochs 15...\n",
      "Epoch 1/15, Loss: 71.5240, Accuracy: 33.60%\n",
      "Epoch 2/15, Loss: 68.8164, Accuracy: 38.35%\n",
      "Epoch 3/15, Loss: 68.5136, Accuracy: 37.47%\n",
      "Epoch 4/15, Loss: 68.3991, Accuracy: 38.21%\n",
      "Epoch 5/15, Loss: 68.3949, Accuracy: 38.35%\n",
      "Epoch 6/15, Loss: 68.0725, Accuracy: 37.87%\n",
      "Epoch 7/15, Loss: 68.0755, Accuracy: 38.14%\n",
      "Epoch 8/15, Loss: 67.9266, Accuracy: 38.14%\n",
      "Epoch 9/15, Loss: 67.8535, Accuracy: 38.08%\n",
      "Epoch 10/15, Loss: 67.9760, Accuracy: 37.80%\n",
      "Epoch 11/15, Loss: 67.9633, Accuracy: 38.01%\n",
      "Epoch 12/15, Loss: 67.8306, Accuracy: 38.21%\n",
      "Epoch 13/15, Loss: 67.7772, Accuracy: 38.08%\n",
      "Epoch 14/15, Loss: 67.6677, Accuracy: 37.87%\n",
      "Epoch 15/15, Loss: 67.7587, Accuracy: 37.94%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.01, epochs 20...\n",
      "Epoch 1/20, Loss: 71.7847, Accuracy: 31.64%\n",
      "Epoch 2/20, Loss: 68.6475, Accuracy: 37.80%\n",
      "Epoch 3/20, Loss: 68.3824, Accuracy: 37.33%\n",
      "Epoch 4/20, Loss: 67.9975, Accuracy: 38.96%\n",
      "Epoch 5/20, Loss: 68.1454, Accuracy: 37.87%\n",
      "Epoch 6/20, Loss: 67.9077, Accuracy: 38.21%\n",
      "Epoch 7/20, Loss: 67.9006, Accuracy: 38.35%\n",
      "Epoch 8/20, Loss: 67.7156, Accuracy: 38.96%\n",
      "Epoch 9/20, Loss: 67.5417, Accuracy: 38.55%\n",
      "Epoch 10/20, Loss: 67.6520, Accuracy: 38.41%\n",
      "Epoch 11/20, Loss: 67.5657, Accuracy: 37.87%\n",
      "Epoch 12/20, Loss: 67.5195, Accuracy: 38.41%\n",
      "Epoch 13/20, Loss: 67.5769, Accuracy: 37.80%\n",
      "Epoch 14/20, Loss: 67.5039, Accuracy: 38.62%\n",
      "Epoch 15/20, Loss: 67.8078, Accuracy: 37.94%\n",
      "Epoch 16/20, Loss: 67.7147, Accuracy: 39.09%\n",
      "Epoch 17/20, Loss: 67.4490, Accuracy: 39.02%\n",
      "Epoch 18/20, Loss: 67.6737, Accuracy: 38.41%\n",
      "Epoch 19/20, Loss: 67.6540, Accuracy: 38.41%\n",
      "Epoch 20/20, Loss: 67.4819, Accuracy: 39.23%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.001, epochs 5...\n",
      "Epoch 1/5, Loss: 70.6047, Accuracy: 33.60%\n",
      "Epoch 2/5, Loss: 67.4336, Accuracy: 37.80%\n",
      "Epoch 3/5, Loss: 66.9237, Accuracy: 38.96%\n",
      "Epoch 4/5, Loss: 66.1890, Accuracy: 39.30%\n",
      "Epoch 5/5, Loss: 65.8492, Accuracy: 39.70%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.001, epochs 10...\n",
      "Epoch 1/10, Loss: 71.2861, Accuracy: 32.66%\n",
      "Epoch 2/10, Loss: 67.8406, Accuracy: 36.79%\n",
      "Epoch 3/10, Loss: 67.2196, Accuracy: 37.67%\n",
      "Epoch 4/10, Loss: 66.7989, Accuracy: 38.21%\n",
      "Epoch 5/10, Loss: 66.6248, Accuracy: 37.26%\n",
      "Epoch 6/10, Loss: 66.1908, Accuracy: 37.94%\n",
      "Epoch 7/10, Loss: 65.7588, Accuracy: 38.01%\n",
      "Epoch 8/10, Loss: 65.7435, Accuracy: 38.75%\n",
      "Epoch 9/10, Loss: 65.4124, Accuracy: 39.02%\n",
      "Epoch 10/10, Loss: 65.0916, Accuracy: 38.75%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.001, epochs 15...\n",
      "Epoch 1/15, Loss: 71.0121, Accuracy: 33.13%\n",
      "Epoch 2/15, Loss: 67.5933, Accuracy: 37.87%\n",
      "Epoch 3/15, Loss: 66.9816, Accuracy: 38.48%\n",
      "Epoch 4/15, Loss: 66.5894, Accuracy: 37.94%\n",
      "Epoch 5/15, Loss: 66.0535, Accuracy: 38.41%\n",
      "Epoch 6/15, Loss: 65.7728, Accuracy: 39.16%\n",
      "Epoch 7/15, Loss: 65.3770, Accuracy: 39.43%\n",
      "Epoch 8/15, Loss: 65.1631, Accuracy: 39.57%\n",
      "Epoch 9/15, Loss: 65.2802, Accuracy: 38.75%\n",
      "Epoch 10/15, Loss: 65.2930, Accuracy: 39.02%\n",
      "Epoch 11/15, Loss: 64.9186, Accuracy: 40.11%\n",
      "Epoch 12/15, Loss: 64.9597, Accuracy: 39.91%\n",
      "Epoch 13/15, Loss: 64.8331, Accuracy: 40.85%\n",
      "Epoch 14/15, Loss: 64.5654, Accuracy: 40.24%\n",
      "Epoch 15/15, Loss: 64.2773, Accuracy: 41.46%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.001, epochs 20...\n",
      "Epoch 1/20, Loss: 71.3048, Accuracy: 33.54%\n",
      "Epoch 2/20, Loss: 67.8065, Accuracy: 37.53%\n",
      "Epoch 3/20, Loss: 67.0337, Accuracy: 37.94%\n",
      "Epoch 4/20, Loss: 66.8069, Accuracy: 38.62%\n",
      "Epoch 5/20, Loss: 66.5208, Accuracy: 39.63%\n",
      "Epoch 6/20, Loss: 66.1729, Accuracy: 38.75%\n",
      "Epoch 7/20, Loss: 65.8938, Accuracy: 38.89%\n",
      "Epoch 8/20, Loss: 65.7710, Accuracy: 39.57%\n",
      "Epoch 9/20, Loss: 65.5584, Accuracy: 38.62%\n",
      "Epoch 10/20, Loss: 65.2302, Accuracy: 38.55%\n",
      "Epoch 11/20, Loss: 65.1337, Accuracy: 38.96%\n",
      "Epoch 12/20, Loss: 65.2304, Accuracy: 38.62%\n",
      "Epoch 13/20, Loss: 64.8942, Accuracy: 39.30%\n",
      "Epoch 14/20, Loss: 64.4952, Accuracy: 40.11%\n",
      "Epoch 15/20, Loss: 64.3224, Accuracy: 40.38%\n",
      "Epoch 16/20, Loss: 64.2440, Accuracy: 40.04%\n",
      "Epoch 17/20, Loss: 64.1383, Accuracy: 39.70%\n",
      "Epoch 18/20, Loss: 63.7965, Accuracy: 40.45%\n",
      "Epoch 19/20, Loss: 63.7354, Accuracy: 41.46%\n",
      "Epoch 20/20, Loss: 63.2859, Accuracy: 40.92%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.0001, epochs 5...\n",
      "Epoch 1/5, Loss: 70.1240, Accuracy: 33.13%\n",
      "Epoch 2/5, Loss: 67.2855, Accuracy: 37.06%\n",
      "Epoch 3/5, Loss: 66.7852, Accuracy: 37.94%\n",
      "Epoch 4/5, Loss: 66.2622, Accuracy: 37.94%\n",
      "Epoch 5/5, Loss: 65.6459, Accuracy: 38.82%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.0001, epochs 10...\n",
      "Epoch 1/10, Loss: 70.4191, Accuracy: 34.01%\n",
      "Epoch 2/10, Loss: 66.9614, Accuracy: 38.14%\n",
      "Epoch 3/10, Loss: 66.2432, Accuracy: 39.09%\n",
      "Epoch 4/10, Loss: 65.7131, Accuracy: 38.21%\n",
      "Epoch 5/10, Loss: 65.3891, Accuracy: 39.43%\n",
      "Epoch 6/10, Loss: 65.0510, Accuracy: 40.04%\n",
      "Epoch 7/10, Loss: 64.5558, Accuracy: 39.63%\n",
      "Epoch 8/10, Loss: 64.1848, Accuracy: 39.84%\n",
      "Epoch 9/10, Loss: 63.7382, Accuracy: 40.92%\n",
      "Epoch 10/10, Loss: 63.4119, Accuracy: 41.67%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.0001, epochs 15...\n",
      "Epoch 1/15, Loss: 71.1757, Accuracy: 32.25%\n",
      "Epoch 2/15, Loss: 67.7997, Accuracy: 37.26%\n",
      "Epoch 3/15, Loss: 67.0544, Accuracy: 37.33%\n",
      "Epoch 4/15, Loss: 66.4460, Accuracy: 37.74%\n",
      "Epoch 5/15, Loss: 66.1272, Accuracy: 37.74%\n",
      "Epoch 6/15, Loss: 65.7082, Accuracy: 38.75%\n",
      "Epoch 7/15, Loss: 65.4065, Accuracy: 39.63%\n",
      "Epoch 8/15, Loss: 64.9456, Accuracy: 39.63%\n",
      "Epoch 9/15, Loss: 64.5682, Accuracy: 39.84%\n",
      "Epoch 10/15, Loss: 64.1825, Accuracy: 40.51%\n",
      "Epoch 11/15, Loss: 63.8595, Accuracy: 40.65%\n",
      "Epoch 12/15, Loss: 63.3283, Accuracy: 41.87%\n",
      "Epoch 13/15, Loss: 63.4792, Accuracy: 41.26%\n",
      "Epoch 14/15, Loss: 62.7064, Accuracy: 42.82%\n",
      "Epoch 15/15, Loss: 62.2974, Accuracy: 42.48%\n",
      "Training with dropout 0.5, lr 0.01, wd 0.0001, epochs 20...\n",
      "Epoch 1/20, Loss: 71.1253, Accuracy: 32.11%\n",
      "Epoch 2/20, Loss: 67.7103, Accuracy: 38.08%\n",
      "Epoch 3/20, Loss: 66.7851, Accuracy: 37.60%\n",
      "Epoch 4/20, Loss: 66.3606, Accuracy: 38.82%\n",
      "Epoch 5/20, Loss: 65.8923, Accuracy: 38.01%\n",
      "Epoch 6/20, Loss: 65.6466, Accuracy: 38.55%\n",
      "Epoch 7/20, Loss: 65.0978, Accuracy: 39.50%\n",
      "Epoch 8/20, Loss: 64.8293, Accuracy: 39.43%\n",
      "Epoch 9/20, Loss: 64.7546, Accuracy: 38.48%\n",
      "Epoch 10/20, Loss: 64.5423, Accuracy: 40.31%\n",
      "Epoch 11/20, Loss: 64.0393, Accuracy: 40.79%\n",
      "Epoch 12/20, Loss: 63.7738, Accuracy: 41.80%\n",
      "Epoch 13/20, Loss: 63.6593, Accuracy: 40.85%\n",
      "Epoch 14/20, Loss: 63.0298, Accuracy: 42.14%\n",
      "Epoch 15/20, Loss: 62.7635, Accuracy: 42.01%\n",
      "Epoch 16/20, Loss: 62.3305, Accuracy: 43.70%\n",
      "Epoch 17/20, Loss: 62.2081, Accuracy: 43.83%\n",
      "Epoch 18/20, Loss: 61.4581, Accuracy: 43.97%\n",
      "Epoch 19/20, Loss: 60.9728, Accuracy: 46.00%\n",
      "Epoch 20/20, Loss: 60.9573, Accuracy: 44.72%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.01, epochs 5...\n",
      "Epoch 1/5, Loss: 73.8376, Accuracy: 27.30%\n",
      "Epoch 2/5, Loss: 70.0505, Accuracy: 39.02%\n",
      "Epoch 3/5, Loss: 68.3019, Accuracy: 39.70%\n",
      "Epoch 4/5, Loss: 67.4741, Accuracy: 40.18%\n",
      "Epoch 5/5, Loss: 67.0207, Accuracy: 39.84%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.01, epochs 10...\n",
      "Epoch 1/10, Loss: 73.4265, Accuracy: 29.20%\n",
      "Epoch 2/10, Loss: 70.1671, Accuracy: 37.94%\n",
      "Epoch 3/10, Loss: 68.7361, Accuracy: 39.43%\n",
      "Epoch 4/10, Loss: 68.0548, Accuracy: 39.43%\n",
      "Epoch 5/10, Loss: 67.6097, Accuracy: 39.16%\n",
      "Epoch 6/10, Loss: 67.2621, Accuracy: 39.16%\n",
      "Epoch 7/10, Loss: 66.9874, Accuracy: 39.36%\n",
      "Epoch 8/10, Loss: 66.7827, Accuracy: 39.77%\n",
      "Epoch 9/10, Loss: 66.5928, Accuracy: 40.24%\n",
      "Epoch 10/10, Loss: 66.4059, Accuracy: 39.77%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.01, epochs 15...\n",
      "Epoch 1/15, Loss: 73.9974, Accuracy: 28.39%\n",
      "Epoch 2/15, Loss: 70.1492, Accuracy: 36.25%\n",
      "Epoch 3/15, Loss: 68.5465, Accuracy: 38.14%\n",
      "Epoch 4/15, Loss: 67.6550, Accuracy: 39.09%\n",
      "Epoch 5/15, Loss: 67.2098, Accuracy: 38.69%\n",
      "Epoch 6/15, Loss: 66.8442, Accuracy: 38.89%\n",
      "Epoch 7/15, Loss: 66.5753, Accuracy: 38.69%\n",
      "Epoch 8/15, Loss: 66.4312, Accuracy: 38.21%\n",
      "Epoch 9/15, Loss: 66.2641, Accuracy: 39.23%\n",
      "Epoch 10/15, Loss: 66.1376, Accuracy: 38.62%\n",
      "Epoch 11/15, Loss: 65.9873, Accuracy: 39.77%\n",
      "Epoch 12/15, Loss: 65.8460, Accuracy: 39.84%\n",
      "Epoch 13/15, Loss: 65.7647, Accuracy: 40.11%\n",
      "Epoch 14/15, Loss: 65.5973, Accuracy: 39.57%\n",
      "Epoch 15/15, Loss: 65.5082, Accuracy: 40.38%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.01, epochs 20...\n",
      "Epoch 1/20, Loss: 73.8587, Accuracy: 30.69%\n",
      "Epoch 2/20, Loss: 69.8915, Accuracy: 38.75%\n",
      "Epoch 3/20, Loss: 68.4072, Accuracy: 39.36%\n",
      "Epoch 4/20, Loss: 67.6480, Accuracy: 39.50%\n",
      "Epoch 5/20, Loss: 67.1491, Accuracy: 39.16%\n",
      "Epoch 6/20, Loss: 66.7942, Accuracy: 39.16%\n",
      "Epoch 7/20, Loss: 66.5695, Accuracy: 39.02%\n",
      "Epoch 8/20, Loss: 66.3736, Accuracy: 39.63%\n",
      "Epoch 9/20, Loss: 66.1777, Accuracy: 39.50%\n",
      "Epoch 10/20, Loss: 66.0714, Accuracy: 39.70%\n",
      "Epoch 11/20, Loss: 65.9555, Accuracy: 39.84%\n",
      "Epoch 12/20, Loss: 65.8429, Accuracy: 40.79%\n",
      "Epoch 13/20, Loss: 65.7434, Accuracy: 40.51%\n",
      "Epoch 14/20, Loss: 65.6589, Accuracy: 40.58%\n",
      "Epoch 15/20, Loss: 65.5804, Accuracy: 40.79%\n",
      "Epoch 16/20, Loss: 65.5751, Accuracy: 41.06%\n",
      "Epoch 17/20, Loss: 65.3797, Accuracy: 41.26%\n",
      "Epoch 18/20, Loss: 65.3501, Accuracy: 41.26%\n",
      "Epoch 19/20, Loss: 65.2732, Accuracy: 41.19%\n",
      "Epoch 20/20, Loss: 65.1136, Accuracy: 41.40%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.001, epochs 5...\n",
      "Epoch 1/5, Loss: 73.6859, Accuracy: 29.74%\n",
      "Epoch 2/5, Loss: 69.9395, Accuracy: 38.01%\n",
      "Epoch 3/5, Loss: 68.1883, Accuracy: 38.21%\n",
      "Epoch 4/5, Loss: 67.2882, Accuracy: 38.69%\n",
      "Epoch 5/5, Loss: 66.7667, Accuracy: 39.02%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.001, epochs 10...\n",
      "Epoch 1/10, Loss: 72.9085, Accuracy: 30.35%\n",
      "Epoch 2/10, Loss: 69.3481, Accuracy: 39.09%\n",
      "Epoch 3/10, Loss: 67.9298, Accuracy: 39.23%\n",
      "Epoch 4/10, Loss: 67.1759, Accuracy: 39.43%\n",
      "Epoch 5/10, Loss: 66.7294, Accuracy: 39.63%\n",
      "Epoch 6/10, Loss: 66.3849, Accuracy: 39.77%\n",
      "Epoch 7/10, Loss: 66.0404, Accuracy: 39.50%\n",
      "Epoch 8/10, Loss: 65.8300, Accuracy: 39.16%\n",
      "Epoch 9/10, Loss: 65.6460, Accuracy: 39.57%\n",
      "Epoch 10/10, Loss: 65.3633, Accuracy: 39.57%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.001, epochs 15...\n",
      "Epoch 1/15, Loss: 73.1388, Accuracy: 32.93%\n",
      "Epoch 2/15, Loss: 69.6917, Accuracy: 39.50%\n",
      "Epoch 3/15, Loss: 68.2229, Accuracy: 38.82%\n",
      "Epoch 4/15, Loss: 67.3926, Accuracy: 39.23%\n",
      "Epoch 5/15, Loss: 66.8581, Accuracy: 39.50%\n",
      "Epoch 6/15, Loss: 66.4249, Accuracy: 38.96%\n",
      "Epoch 7/15, Loss: 66.0980, Accuracy: 39.57%\n",
      "Epoch 8/15, Loss: 65.8357, Accuracy: 39.50%\n",
      "Epoch 9/15, Loss: 65.5253, Accuracy: 39.63%\n",
      "Epoch 10/15, Loss: 65.3662, Accuracy: 39.91%\n",
      "Epoch 11/15, Loss: 65.0990, Accuracy: 39.84%\n",
      "Epoch 12/15, Loss: 64.9655, Accuracy: 40.11%\n",
      "Epoch 13/15, Loss: 64.6863, Accuracy: 40.45%\n",
      "Epoch 14/15, Loss: 64.4588, Accuracy: 40.51%\n",
      "Epoch 15/15, Loss: 64.2460, Accuracy: 41.12%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.001, epochs 20...\n",
      "Epoch 1/20, Loss: 73.7109, Accuracy: 29.47%\n",
      "Epoch 2/20, Loss: 70.3196, Accuracy: 37.53%\n",
      "Epoch 3/20, Loss: 68.5651, Accuracy: 38.89%\n",
      "Epoch 4/20, Loss: 67.4256, Accuracy: 39.36%\n",
      "Epoch 5/20, Loss: 66.7400, Accuracy: 39.91%\n",
      "Epoch 6/20, Loss: 66.2190, Accuracy: 39.30%\n",
      "Epoch 7/20, Loss: 65.7867, Accuracy: 40.31%\n",
      "Epoch 8/20, Loss: 65.3991, Accuracy: 40.31%\n",
      "Epoch 9/20, Loss: 65.1606, Accuracy: 39.91%\n",
      "Epoch 10/20, Loss: 64.8619, Accuracy: 40.92%\n",
      "Epoch 11/20, Loss: 64.5689, Accuracy: 40.85%\n",
      "Epoch 12/20, Loss: 64.3594, Accuracy: 41.73%\n",
      "Epoch 13/20, Loss: 64.1274, Accuracy: 42.07%\n",
      "Epoch 14/20, Loss: 63.8894, Accuracy: 42.28%\n",
      "Epoch 15/20, Loss: 63.5697, Accuracy: 42.68%\n",
      "Epoch 16/20, Loss: 63.4219, Accuracy: 42.21%\n",
      "Epoch 17/20, Loss: 63.1184, Accuracy: 43.77%\n",
      "Epoch 18/20, Loss: 62.8671, Accuracy: 43.16%\n",
      "Epoch 19/20, Loss: 62.5265, Accuracy: 43.63%\n",
      "Epoch 20/20, Loss: 62.4168, Accuracy: 44.04%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.0001, epochs 5...\n",
      "Epoch 1/5, Loss: 73.3552, Accuracy: 27.10%\n",
      "Epoch 2/5, Loss: 69.6868, Accuracy: 38.14%\n",
      "Epoch 3/5, Loss: 67.9823, Accuracy: 38.89%\n",
      "Epoch 4/5, Loss: 67.0254, Accuracy: 38.41%\n",
      "Epoch 5/5, Loss: 66.4141, Accuracy: 38.55%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.0001, epochs 10...\n",
      "Epoch 1/10, Loss: 73.1764, Accuracy: 35.03%\n",
      "Epoch 2/10, Loss: 69.4944, Accuracy: 39.02%\n",
      "Epoch 3/10, Loss: 67.9175, Accuracy: 38.21%\n",
      "Epoch 4/10, Loss: 67.1526, Accuracy: 38.35%\n",
      "Epoch 5/10, Loss: 66.6517, Accuracy: 38.28%\n",
      "Epoch 6/10, Loss: 66.3638, Accuracy: 39.23%\n",
      "Epoch 7/10, Loss: 66.0304, Accuracy: 38.14%\n",
      "Epoch 8/10, Loss: 65.6245, Accuracy: 39.30%\n",
      "Epoch 9/10, Loss: 65.3808, Accuracy: 39.43%\n",
      "Epoch 10/10, Loss: 65.0210, Accuracy: 39.57%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.0001, epochs 15...\n",
      "Epoch 1/15, Loss: 73.4907, Accuracy: 30.49%\n",
      "Epoch 2/15, Loss: 69.9455, Accuracy: 38.28%\n",
      "Epoch 3/15, Loss: 68.1271, Accuracy: 38.35%\n",
      "Epoch 4/15, Loss: 67.2015, Accuracy: 38.82%\n",
      "Epoch 5/15, Loss: 66.6026, Accuracy: 38.75%\n",
      "Epoch 6/15, Loss: 66.2584, Accuracy: 39.09%\n",
      "Epoch 7/15, Loss: 65.9304, Accuracy: 39.36%\n",
      "Epoch 8/15, Loss: 65.6630, Accuracy: 39.23%\n",
      "Epoch 9/15, Loss: 65.3963, Accuracy: 39.70%\n",
      "Epoch 10/15, Loss: 65.1707, Accuracy: 39.50%\n",
      "Epoch 11/15, Loss: 64.9693, Accuracy: 39.84%\n",
      "Epoch 12/15, Loss: 64.7636, Accuracy: 39.70%\n",
      "Epoch 13/15, Loss: 64.6510, Accuracy: 39.97%\n",
      "Epoch 14/15, Loss: 64.4370, Accuracy: 40.24%\n",
      "Epoch 15/15, Loss: 64.1495, Accuracy: 40.45%\n",
      "Training with dropout 0.5, lr 0.001, wd 0.0001, epochs 20...\n",
      "Epoch 1/20, Loss: 73.6943, Accuracy: 34.08%\n",
      "Epoch 2/20, Loss: 70.5238, Accuracy: 38.62%\n",
      "Epoch 3/20, Loss: 68.6652, Accuracy: 39.43%\n",
      "Epoch 4/20, Loss: 67.4117, Accuracy: 40.11%\n",
      "Epoch 5/20, Loss: 66.6864, Accuracy: 40.11%\n",
      "Epoch 6/20, Loss: 66.2095, Accuracy: 40.18%\n",
      "Epoch 7/20, Loss: 65.8987, Accuracy: 40.18%\n",
      "Epoch 8/20, Loss: 65.5680, Accuracy: 40.04%\n",
      "Epoch 9/20, Loss: 65.2899, Accuracy: 40.24%\n",
      "Epoch 10/20, Loss: 65.1100, Accuracy: 40.65%\n",
      "Epoch 11/20, Loss: 64.8877, Accuracy: 40.38%\n",
      "Epoch 12/20, Loss: 64.8036, Accuracy: 40.85%\n",
      "Epoch 13/20, Loss: 64.5019, Accuracy: 41.19%\n",
      "Epoch 14/20, Loss: 64.3047, Accuracy: 41.19%\n",
      "Epoch 15/20, Loss: 64.1878, Accuracy: 41.53%\n",
      "Epoch 16/20, Loss: 64.0612, Accuracy: 41.67%\n",
      "Epoch 17/20, Loss: 63.7956, Accuracy: 42.21%\n",
      "Epoch 18/20, Loss: 63.7223, Accuracy: 41.87%\n",
      "Epoch 19/20, Loss: 63.4059, Accuracy: 42.28%\n",
      "Epoch 20/20, Loss: 63.2274, Accuracy: 42.55%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.01, epochs 5...\n",
      "Epoch 1/5, Loss: 77.6471, Accuracy: 19.24%\n",
      "Epoch 2/5, Loss: 76.2998, Accuracy: 19.24%\n",
      "Epoch 3/5, Loss: 75.3500, Accuracy: 19.44%\n",
      "Epoch 4/5, Loss: 74.4295, Accuracy: 21.41%\n",
      "Epoch 5/5, Loss: 73.6367, Accuracy: 23.64%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.01, epochs 10...\n",
      "Epoch 1/10, Loss: 75.4959, Accuracy: 23.44%\n",
      "Epoch 2/10, Loss: 74.6864, Accuracy: 31.84%\n",
      "Epoch 3/10, Loss: 74.0120, Accuracy: 35.91%\n",
      "Epoch 4/10, Loss: 73.3220, Accuracy: 37.33%\n",
      "Epoch 5/10, Loss: 72.8068, Accuracy: 37.06%\n",
      "Epoch 6/10, Loss: 72.2591, Accuracy: 37.26%\n",
      "Epoch 7/10, Loss: 71.7919, Accuracy: 37.94%\n",
      "Epoch 8/10, Loss: 71.3201, Accuracy: 38.21%\n",
      "Epoch 9/10, Loss: 70.9090, Accuracy: 38.28%\n",
      "Epoch 10/10, Loss: 70.5387, Accuracy: 38.48%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.01, epochs 15...\n",
      "Epoch 1/15, Loss: 75.9625, Accuracy: 20.73%\n",
      "Epoch 2/15, Loss: 74.8656, Accuracy: 21.27%\n",
      "Epoch 3/15, Loss: 74.0428, Accuracy: 21.95%\n",
      "Epoch 4/15, Loss: 73.2939, Accuracy: 28.18%\n",
      "Epoch 5/15, Loss: 72.5971, Accuracy: 33.20%\n",
      "Epoch 6/15, Loss: 71.9260, Accuracy: 35.91%\n",
      "Epoch 7/15, Loss: 71.2981, Accuracy: 37.20%\n",
      "Epoch 8/15, Loss: 70.7207, Accuracy: 36.79%\n",
      "Epoch 9/15, Loss: 70.2031, Accuracy: 37.87%\n",
      "Epoch 10/15, Loss: 69.7803, Accuracy: 38.08%\n",
      "Epoch 11/15, Loss: 69.4345, Accuracy: 38.75%\n",
      "Epoch 12/15, Loss: 69.1910, Accuracy: 38.75%\n",
      "Epoch 13/15, Loss: 68.9706, Accuracy: 38.35%\n",
      "Epoch 14/15, Loss: 68.7393, Accuracy: 38.75%\n",
      "Epoch 15/15, Loss: 68.5693, Accuracy: 39.02%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.01, epochs 20...\n",
      "Epoch 1/20, Loss: 76.1709, Accuracy: 23.64%\n",
      "Epoch 2/20, Loss: 75.2198, Accuracy: 27.10%\n",
      "Epoch 3/20, Loss: 74.3695, Accuracy: 26.96%\n",
      "Epoch 4/20, Loss: 73.6705, Accuracy: 28.46%\n",
      "Epoch 5/20, Loss: 72.9966, Accuracy: 30.56%\n",
      "Epoch 6/20, Loss: 72.4122, Accuracy: 33.20%\n",
      "Epoch 7/20, Loss: 71.8847, Accuracy: 34.55%\n",
      "Epoch 8/20, Loss: 71.4320, Accuracy: 36.52%\n",
      "Epoch 9/20, Loss: 70.9746, Accuracy: 37.13%\n",
      "Epoch 10/20, Loss: 70.6243, Accuracy: 38.48%\n",
      "Epoch 11/20, Loss: 70.2424, Accuracy: 38.69%\n",
      "Epoch 12/20, Loss: 69.9037, Accuracy: 39.57%\n",
      "Epoch 13/20, Loss: 69.6372, Accuracy: 39.50%\n",
      "Epoch 14/20, Loss: 69.3843, Accuracy: 39.36%\n",
      "Epoch 15/20, Loss: 69.1203, Accuracy: 39.36%\n",
      "Epoch 16/20, Loss: 68.9147, Accuracy: 39.70%\n",
      "Epoch 17/20, Loss: 68.7254, Accuracy: 39.77%\n",
      "Epoch 18/20, Loss: 68.5118, Accuracy: 40.04%\n",
      "Epoch 19/20, Loss: 68.3685, Accuracy: 40.11%\n",
      "Epoch 20/20, Loss: 68.2254, Accuracy: 39.57%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.001, epochs 5...\n",
      "Epoch 1/5, Loss: 76.2173, Accuracy: 19.58%\n",
      "Epoch 2/5, Loss: 75.3762, Accuracy: 21.21%\n",
      "Epoch 3/5, Loss: 74.6384, Accuracy: 23.85%\n",
      "Epoch 4/5, Loss: 73.9252, Accuracy: 27.64%\n",
      "Epoch 5/5, Loss: 73.3140, Accuracy: 29.07%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.001, epochs 10...\n",
      "Epoch 1/10, Loss: 76.3184, Accuracy: 26.56%\n",
      "Epoch 2/10, Loss: 75.0947, Accuracy: 30.08%\n",
      "Epoch 3/10, Loss: 74.2609, Accuracy: 33.06%\n",
      "Epoch 4/10, Loss: 73.5131, Accuracy: 34.69%\n",
      "Epoch 5/10, Loss: 72.8175, Accuracy: 35.91%\n",
      "Epoch 6/10, Loss: 72.1776, Accuracy: 36.72%\n",
      "Epoch 7/10, Loss: 71.6249, Accuracy: 37.20%\n",
      "Epoch 8/10, Loss: 71.1329, Accuracy: 38.28%\n",
      "Epoch 9/10, Loss: 70.7127, Accuracy: 38.21%\n",
      "Epoch 10/10, Loss: 70.3392, Accuracy: 38.41%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.001, epochs 15...\n",
      "Epoch 1/15, Loss: 75.1695, Accuracy: 24.32%\n",
      "Epoch 2/15, Loss: 74.0975, Accuracy: 34.62%\n",
      "Epoch 3/15, Loss: 73.2591, Accuracy: 37.60%\n",
      "Epoch 4/15, Loss: 72.4995, Accuracy: 38.62%\n",
      "Epoch 5/15, Loss: 71.8328, Accuracy: 39.50%\n",
      "Epoch 6/15, Loss: 71.2177, Accuracy: 39.57%\n",
      "Epoch 7/15, Loss: 70.6692, Accuracy: 39.50%\n",
      "Epoch 8/15, Loss: 70.2588, Accuracy: 39.36%\n",
      "Epoch 9/15, Loss: 69.8818, Accuracy: 39.09%\n",
      "Epoch 10/15, Loss: 69.5379, Accuracy: 39.43%\n",
      "Epoch 11/15, Loss: 69.2654, Accuracy: 39.09%\n",
      "Epoch 12/15, Loss: 69.0528, Accuracy: 39.09%\n",
      "Epoch 13/15, Loss: 68.8346, Accuracy: 39.30%\n",
      "Epoch 14/15, Loss: 68.6667, Accuracy: 39.43%\n",
      "Epoch 15/15, Loss: 68.5011, Accuracy: 39.36%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.001, epochs 20...\n",
      "Epoch 1/20, Loss: 75.3440, Accuracy: 22.09%\n",
      "Epoch 2/20, Loss: 74.4647, Accuracy: 27.37%\n",
      "Epoch 3/20, Loss: 73.8095, Accuracy: 32.99%\n",
      "Epoch 4/20, Loss: 73.2641, Accuracy: 34.62%\n",
      "Epoch 5/20, Loss: 72.7553, Accuracy: 35.77%\n",
      "Epoch 6/20, Loss: 72.2166, Accuracy: 36.04%\n",
      "Epoch 7/20, Loss: 71.7717, Accuracy: 36.79%\n",
      "Epoch 8/20, Loss: 71.3199, Accuracy: 37.06%\n",
      "Epoch 9/20, Loss: 70.8855, Accuracy: 37.80%\n",
      "Epoch 10/20, Loss: 70.3895, Accuracy: 37.94%\n",
      "Epoch 11/20, Loss: 69.9712, Accuracy: 37.74%\n",
      "Epoch 12/20, Loss: 69.6015, Accuracy: 38.01%\n",
      "Epoch 13/20, Loss: 69.2228, Accuracy: 38.35%\n",
      "Epoch 14/20, Loss: 68.9284, Accuracy: 38.89%\n",
      "Epoch 15/20, Loss: 68.6400, Accuracy: 39.16%\n",
      "Epoch 16/20, Loss: 68.4351, Accuracy: 39.16%\n",
      "Epoch 17/20, Loss: 68.2747, Accuracy: 39.09%\n",
      "Epoch 18/20, Loss: 68.0224, Accuracy: 38.96%\n",
      "Epoch 19/20, Loss: 67.9034, Accuracy: 39.36%\n",
      "Epoch 20/20, Loss: 67.7243, Accuracy: 39.30%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.0001, epochs 5...\n",
      "Epoch 1/5, Loss: 75.6448, Accuracy: 21.21%\n",
      "Epoch 2/5, Loss: 74.8423, Accuracy: 23.92%\n",
      "Epoch 3/5, Loss: 74.1755, Accuracy: 29.81%\n",
      "Epoch 4/5, Loss: 73.4372, Accuracy: 34.89%\n",
      "Epoch 5/5, Loss: 72.8251, Accuracy: 36.04%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.0001, epochs 10...\n",
      "Epoch 1/10, Loss: 76.3435, Accuracy: 21.41%\n",
      "Epoch 2/10, Loss: 74.9939, Accuracy: 23.17%\n",
      "Epoch 3/10, Loss: 74.0162, Accuracy: 26.96%\n",
      "Epoch 4/10, Loss: 73.2727, Accuracy: 29.00%\n",
      "Epoch 5/10, Loss: 72.6229, Accuracy: 31.71%\n",
      "Epoch 6/10, Loss: 72.0680, Accuracy: 34.82%\n",
      "Epoch 7/10, Loss: 71.6392, Accuracy: 35.77%\n",
      "Epoch 8/10, Loss: 71.1429, Accuracy: 36.79%\n",
      "Epoch 9/10, Loss: 70.7136, Accuracy: 37.26%\n",
      "Epoch 10/10, Loss: 70.3282, Accuracy: 37.87%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.0001, epochs 15...\n",
      "Epoch 1/15, Loss: 76.3619, Accuracy: 19.31%\n",
      "Epoch 2/15, Loss: 75.3146, Accuracy: 20.73%\n",
      "Epoch 3/15, Loss: 74.5681, Accuracy: 22.36%\n",
      "Epoch 4/15, Loss: 73.8861, Accuracy: 24.12%\n",
      "Epoch 5/15, Loss: 73.2221, Accuracy: 27.17%\n",
      "Epoch 6/15, Loss: 72.5364, Accuracy: 30.49%\n",
      "Epoch 7/15, Loss: 71.9063, Accuracy: 33.13%\n",
      "Epoch 8/15, Loss: 71.2191, Accuracy: 34.28%\n",
      "Epoch 9/15, Loss: 70.6582, Accuracy: 35.70%\n",
      "Epoch 10/15, Loss: 70.2265, Accuracy: 36.04%\n",
      "Epoch 11/15, Loss: 69.7988, Accuracy: 36.99%\n",
      "Epoch 12/15, Loss: 69.4947, Accuracy: 37.87%\n",
      "Epoch 13/15, Loss: 69.2309, Accuracy: 38.08%\n",
      "Epoch 14/15, Loss: 68.9049, Accuracy: 38.14%\n",
      "Epoch 15/15, Loss: 68.7730, Accuracy: 38.28%\n",
      "Training with dropout 0.5, lr 0.0001, wd 0.0001, epochs 20...\n",
      "Epoch 1/20, Loss: 75.7175, Accuracy: 21.34%\n",
      "Epoch 2/20, Loss: 74.9130, Accuracy: 25.95%\n",
      "Epoch 3/20, Loss: 74.1141, Accuracy: 27.71%\n",
      "Epoch 4/20, Loss: 73.3629, Accuracy: 33.40%\n",
      "Epoch 5/20, Loss: 72.6313, Accuracy: 36.25%\n",
      "Epoch 6/20, Loss: 71.9870, Accuracy: 37.20%\n",
      "Epoch 7/20, Loss: 71.3913, Accuracy: 37.60%\n",
      "Epoch 8/20, Loss: 70.9083, Accuracy: 37.80%\n",
      "Epoch 9/20, Loss: 70.5202, Accuracy: 38.28%\n",
      "Epoch 10/20, Loss: 70.0909, Accuracy: 38.41%\n",
      "Epoch 11/20, Loss: 69.7564, Accuracy: 38.82%\n",
      "Epoch 12/20, Loss: 69.5295, Accuracy: 38.35%\n",
      "Epoch 13/20, Loss: 69.1830, Accuracy: 39.16%\n",
      "Epoch 14/20, Loss: 68.9549, Accuracy: 39.09%\n",
      "Epoch 15/20, Loss: 68.7175, Accuracy: 39.02%\n",
      "Epoch 16/20, Loss: 68.5340, Accuracy: 38.82%\n",
      "Epoch 17/20, Loss: 68.3318, Accuracy: 38.96%\n",
      "Epoch 18/20, Loss: 68.1981, Accuracy: 39.23%\n",
      "Epoch 19/20, Loss: 68.0664, Accuracy: 39.16%\n",
      "Epoch 20/20, Loss: 67.9122, Accuracy: 38.69%\n",
      "Results:\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 5, 'final_accuracy': 37.737127371273715, 'final_loss': 68.11973071098328, 'precision': 0.3335146470512324, 'recall': 0.21951219512195122, 'f1_score': 0.1412749495513697}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 10, 'final_accuracy': 38.82113821138211, 'final_loss': 67.58561503887177, 'precision': 0.4103667912554671, 'recall': 0.3604336043360434, 'f1_score': 0.2917831782375345}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 15, 'final_accuracy': 38.82113821138211, 'final_loss': 67.33127641677856, 'precision': 0.05306218373836855, 'recall': 0.23035230352303523, 'f1_score': 0.086255267839022}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 20, 'final_accuracy': 38.550135501355015, 'final_loss': 67.53796458244324, 'precision': 0.0446065278641455, 'recall': 0.17615176151761516, 'f1_score': 0.05973598111331254}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 5, 'final_accuracy': 38.34688346883469, 'final_loss': 66.6352869272232, 'precision': 0.3174521941529918, 'recall': 0.3170731707317073, 'f1_score': 0.25943287917336144}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 10, 'final_accuracy': 39.295392953929536, 'final_loss': 64.63287556171417, 'precision': 0.2663855852196895, 'recall': 0.2899728997289973, 'f1_score': 0.23568082227307333}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 15, 'final_accuracy': 42.00542005420054, 'final_loss': 63.83353769779205, 'precision': 0.09450743945890766, 'recall': 0.23035230352303523, 'f1_score': 0.12798664937334692}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 20, 'final_accuracy': 41.05691056910569, 'final_loss': 63.719835102558136, 'precision': 0.29954882235974833, 'recall': 0.31978319783197834, 'f1_score': 0.27301697456696417}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 5, 'final_accuracy': 38.550135501355015, 'final_loss': 65.66139090061188, 'precision': 0.232565625209483, 'recall': 0.25474254742547425, 'f1_score': 0.17817826438001377}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 10, 'final_accuracy': 42.953929539295395, 'final_loss': 62.85319638252258, 'precision': 0.0352525656404105, 'recall': 0.18699186991869918, 'f1_score': 0.05932155873282872}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 15, 'final_accuracy': 46.070460704607044, 'final_loss': 61.10139238834381, 'precision': 0.03544626105601715, 'recall': 0.18699186991869918, 'f1_score': 0.059595561313580805}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 20, 'final_accuracy': 46.476964769647694, 'final_loss': 59.742588222026825, 'precision': 0.15430124955157049, 'recall': 0.28184281842818426, 'f1_score': 0.19910391263843102}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 5, 'final_accuracy': 39.90514905149052, 'final_loss': 67.36811196804047, 'precision': 0.4521742344613196, 'recall': 0.45528455284552843, 'f1_score': 0.4323591775006956}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 10, 'final_accuracy': 38.75338753387534, 'final_loss': 66.53211855888367, 'precision': 0.2733953651006881, 'recall': 0.3712737127371274, 'f1_score': 0.3015480715800711}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 15, 'final_accuracy': 41.327913279132794, 'final_loss': 65.33368480205536, 'precision': 0.26893773132381965, 'recall': 0.36314363143631434, 'f1_score': 0.29246107045813696}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 20, 'final_accuracy': 41.05691056910569, 'final_loss': 65.15482878684998, 'precision': 0.42604226697512565, 'recall': 0.4092140921409214, 'f1_score': 0.3513750620213155}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 5, 'final_accuracy': 39.90514905149052, 'final_loss': 66.49415743350983, 'precision': 0.2708609584976279, 'recall': 0.34688346883468835, 'f1_score': 0.30111201445347785}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 10, 'final_accuracy': 39.70189701897019, 'final_loss': 65.00187873840332, 'precision': 0.3317261376715776, 'recall': 0.24390243902439024, 'f1_score': 0.1775907085825172}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 15, 'final_accuracy': 42.81842818428184, 'final_loss': 63.6082900762558, 'precision': 0.3552053473427719, 'recall': 0.4173441734417344, 'f1_score': 0.37606866394582594}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 20, 'final_accuracy': 42.34417344173442, 'final_loss': 63.81653547286987, 'precision': 0.33087530008079774, 'recall': 0.4037940379403794, 'f1_score': 0.3291731869961581}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 5, 'final_accuracy': 38.41463414634146, 'final_loss': 66.53548586368561, 'precision': 0.5092435874207376, 'recall': 0.39295392953929537, 'f1_score': 0.35516763486350017}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 10, 'final_accuracy': 39.90514905149052, 'final_loss': 65.13604837656021, 'precision': 0.3774249733412121, 'recall': 0.3604336043360434, 'f1_score': 0.30392879963811276}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 15, 'final_accuracy': 41.53116531165312, 'final_loss': 63.78447985649109, 'precision': 0.1714101576948917, 'recall': 0.24390243902439024, 'f1_score': 0.11248261248261247}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 20, 'final_accuracy': 43.360433604336045, 'final_loss': 62.51343637704849, 'precision': 0.45966505234862404, 'recall': 0.3821138211382114, 'f1_score': 0.34006120526159156}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 5, 'final_accuracy': 30.48780487804878, 'final_loss': 72.81979286670685, 'precision': 0.3817959243317923, 'recall': 0.32791327913279134, 'f1_score': 0.2622556511948207}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 10, 'final_accuracy': 39.97289972899729, 'final_loss': 69.66633903980255, 'precision': 0.2691820807552084, 'recall': 0.37669376693766937, 'f1_score': 0.29751833415712337}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 15, 'final_accuracy': 38.82113821138211, 'final_loss': 68.49839317798615, 'precision': 0.3719983766284276, 'recall': 0.37669376693766937, 'f1_score': 0.2906849639998918}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 20, 'final_accuracy': 39.49864498644986, 'final_loss': 68.38168036937714, 'precision': 0.2483761791416186, 'recall': 0.35501355013550134, 'f1_score': 0.28621796996652193}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 5, 'final_accuracy': 27.981029810298104, 'final_loss': 73.32843840122223, 'precision': 0.16137559957556102, 'recall': 0.25474254742547425, 'f1_score': 0.18848819423520716}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 10, 'final_accuracy': 39.09214092140921, 'final_loss': 70.1717300415039, 'precision': 0.21935631776813097, 'recall': 0.34417344173441733, 'f1_score': 0.26425079861168416}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 15, 'final_accuracy': 39.15989159891599, 'final_loss': 69.21896922588348, 'precision': 0.26030250980583236, 'recall': 0.36585365853658536, 'f1_score': 0.29054687614662916}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 20, 'final_accuracy': 40.24390243902439, 'final_loss': 68.32186543941498, 'precision': 0.2821440819867256, 'recall': 0.3794037940379404, 'f1_score': 0.3126068820343552}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 5, 'final_accuracy': 34.552845528455286, 'final_loss': 72.77134096622467, 'precision': 0.31677929686262735, 'recall': 0.3983739837398374, 'f1_score': 0.32341505730820697}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 10, 'final_accuracy': 38.68563685636856, 'final_loss': 70.90280139446259, 'precision': 0.2745943457325571, 'recall': 0.3604336043360434, 'f1_score': 0.29925087479289914}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 15, 'final_accuracy': 39.70189701897019, 'final_loss': 68.40607106685638, 'precision': 0.27274270304714604, 'recall': 0.3794037940379404, 'f1_score': 0.29335638557665755}\n",
      "{'dropout_rate': 0, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 20, 'final_accuracy': 39.56639566395664, 'final_loss': 67.90002977848053, 'precision': 0.2605887483411694, 'recall': 0.36314363143631434, 'f1_score': 0.2919310813046729}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 5, 'final_accuracy': 38.21138211382114, 'final_loss': 68.21526515483856, 'precision': 0.3573326731993691, 'recall': 0.3224932249322493, 'f1_score': 0.23887838603777337}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 10, 'final_accuracy': 38.888888888888886, 'final_loss': 67.78718447685242, 'precision': 0.10531165311653116, 'recall': 0.2222222222222222, 'f1_score': 0.11886222276551311}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 15, 'final_accuracy': 38.34688346883469, 'final_loss': 67.3980621099472, 'precision': 0.16965946321607042, 'recall': 0.25474254742547425, 'f1_score': 0.13466750052115906}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 20, 'final_accuracy': 38.482384823848236, 'final_loss': 67.52854001522064, 'precision': 0.33369208208737106, 'recall': 0.34688346883468835, 'f1_score': 0.26582044699727275}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 5, 'final_accuracy': 39.769647696476966, 'final_loss': 65.91139578819275, 'precision': 0.21263186394702416, 'recall': 0.21680216802168023, 'f1_score': 0.13325001375953102}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 10, 'final_accuracy': 39.56639566395664, 'final_loss': 65.13047647476196, 'precision': 0.3018726552355159, 'recall': 0.3685636856368564, 'f1_score': 0.32955930942473816}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 15, 'final_accuracy': 40.853658536585364, 'final_loss': 64.22692632675171, 'precision': 0.034965959415691715, 'recall': 0.18699186991869918, 'f1_score': 0.05891524668671344}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 20, 'final_accuracy': 43.699186991869915, 'final_loss': 62.24375033378601, 'precision': 0.03274245857212671, 'recall': 0.17886178861788618, 'f1_score': 0.05535214705495784}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 5, 'final_accuracy': 38.888888888888886, 'final_loss': 65.46312189102173, 'precision': 0.3546304840048212, 'recall': 0.23848238482384823, 'f1_score': 0.1566021042865895}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 10, 'final_accuracy': 40.37940379403794, 'final_loss': 63.984230756759644, 'precision': 0.04588626036636306, 'recall': 0.18699186991869918, 'f1_score': 0.06491685885943614}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 15, 'final_accuracy': 45.9349593495935, 'final_loss': 61.257021844387054, 'precision': 0.034965959415691715, 'recall': 0.18699186991869918, 'f1_score': 0.05891524668671344}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 20, 'final_accuracy': 49.11924119241193, 'final_loss': 57.996921479701996, 'precision': 0.09322965863216723, 'recall': 0.24119241192411925, 'f1_score': 0.13155056463215434}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 5, 'final_accuracy': 39.56639566395664, 'final_loss': 67.74339354038239, 'precision': 0.35672438965460335, 'recall': 0.3523035230352303, 'f1_score': 0.2899011215913247}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 10, 'final_accuracy': 40.51490514905149, 'final_loss': 65.90541219711304, 'precision': 0.3995325242436924, 'recall': 0.4363143631436314, 'f1_score': 0.3989151024018561}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 15, 'final_accuracy': 40.78590785907859, 'final_loss': 65.72005701065063, 'precision': 0.44531169193232084, 'recall': 0.3794037940379404, 'f1_score': 0.31644330200880383}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 20, 'final_accuracy': 39.70189701897019, 'final_loss': 65.888188123703, 'precision': 0.3934898248845428, 'recall': 0.37398373983739835, 'f1_score': 0.3262425192484462}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 5, 'final_accuracy': 39.90514905149052, 'final_loss': 66.83646380901337, 'precision': 0.2645772726226999, 'recall': 0.3252032520325203, 'f1_score': 0.27856479256146655}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 10, 'final_accuracy': 40.51490514905149, 'final_loss': 65.61244130134583, 'precision': 0.3189516355958118, 'recall': 0.35501355013550134, 'f1_score': 0.2908704890384418}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 15, 'final_accuracy': 40.921409214092144, 'final_loss': 64.43090462684631, 'precision': 0.46319419801396366, 'recall': 0.4363143631436314, 'f1_score': 0.3975944304306732}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 20, 'final_accuracy': 44.51219512195122, 'final_loss': 62.75911355018616, 'precision': 0.3569664426363299, 'recall': 0.1951219512195122, 'f1_score': 0.081830592177126}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 5, 'final_accuracy': 38.21138211382114, 'final_loss': 66.63312304019928, 'precision': 0.45532964753604543, 'recall': 0.3902439024390244, 'f1_score': 0.3390120223783933}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 10, 'final_accuracy': 40.31165311653117, 'final_loss': 65.024817943573, 'precision': 0.4866899863280155, 'recall': 0.39295392953929537, 'f1_score': 0.3415454989737762}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 15, 'final_accuracy': 42.00542005420054, 'final_loss': 63.582428216934204, 'precision': 0.6385682971048824, 'recall': 0.33604336043360433, 'f1_score': 0.28586708839557323}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 20, 'final_accuracy': 42.75067750677507, 'final_loss': 62.90822273492813, 'precision': 0.36703870197515, 'recall': 0.4010840108401084, 'f1_score': 0.35839662110484544}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 5, 'final_accuracy': 32.58807588075881, 'final_loss': 72.80622637271881, 'precision': 0.41963982251572196, 'recall': 0.33875338753387535, 'f1_score': 0.24008548374053384}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 10, 'final_accuracy': 35.97560975609756, 'final_loss': 70.11041605472565, 'precision': 0.4313886644867009, 'recall': 0.3794037940379404, 'f1_score': 0.31679721763003676}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 15, 'final_accuracy': 39.09214092140921, 'final_loss': 68.76338517665863, 'precision': 0.2810584505424937, 'recall': 0.3794037940379404, 'f1_score': 0.3129285655760644}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 20, 'final_accuracy': 38.143631436314365, 'final_loss': 68.82970607280731, 'precision': 0.33171289636662205, 'recall': 0.3685636856368564, 'f1_score': 0.31645813636232656}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 5, 'final_accuracy': 25.880758807588077, 'final_loss': 73.2057636976242, 'precision': 0.43568072408516423, 'recall': 0.21138211382113822, 'f1_score': 0.15830567549977473}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 10, 'final_accuracy': 37.12737127371274, 'final_loss': 70.63146495819092, 'precision': 0.2335224466038611, 'recall': 0.3794037940379404, 'f1_score': 0.2884959520279636}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 15, 'final_accuracy': 38.956639566395665, 'final_loss': 67.92685103416443, 'precision': 0.2982306032248758, 'recall': 0.37669376693766937, 'f1_score': 0.29724227274059134}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 20, 'final_accuracy': 40.37940379403794, 'final_loss': 68.35239291191101, 'precision': 0.2195694117333972, 'recall': 0.36314363143631434, 'f1_score': 0.2735104151429847}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 5, 'final_accuracy': 36.856368563685635, 'final_loss': 73.38098156452179, 'precision': 0.23404310558515978, 'recall': 0.36585365853658536, 'f1_score': 0.2793991977685525}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 10, 'final_accuracy': 39.56639566395664, 'final_loss': 70.5197491645813, 'precision': 0.22657372661005762, 'recall': 0.37398373983739835, 'f1_score': 0.2816638375198209}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 15, 'final_accuracy': 39.15989159891599, 'final_loss': 69.16725277900696, 'precision': 0.37124418434198214, 'recall': 0.37398373983739835, 'f1_score': 0.30068651489033826}\n",
      "{'dropout_rate': 0.25, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 20, 'final_accuracy': 38.956639566395665, 'final_loss': 67.76030385494232, 'precision': 0.2694059923603194, 'recall': 0.37669376693766937, 'f1_score': 0.3040345528482686}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 5, 'final_accuracy': 36.78861788617886, 'final_loss': 68.18247091770172, 'precision': 0.2627228180892474, 'recall': 0.23848238482384823, 'f1_score': 0.16234890417920855}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 10, 'final_accuracy': 37.19512195121951, 'final_loss': 68.0916599035263, 'precision': 0.31207856179996596, 'recall': 0.36585365853658536, 'f1_score': 0.3117802653304363}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 15, 'final_accuracy': 37.94037940379404, 'final_loss': 67.7586681842804, 'precision': 0.2193186217576461, 'recall': 0.19241192411924118, 'f1_score': 0.08702907955464996}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.01, 'epochs': 20, 'final_accuracy': 39.22764227642276, 'final_loss': 67.48192477226257, 'precision': 0.22527768544028703, 'recall': 0.2601626016260163, 'f1_score': 0.14030830340201328}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 5, 'final_accuracy': 39.70189701897019, 'final_loss': 65.849161028862, 'precision': 0.28797037718804863, 'recall': 0.34146341463414637, 'f1_score': 0.2485617422712194}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 10, 'final_accuracy': 38.75338753387534, 'final_loss': 65.09158384799957, 'precision': 0.034965959415691715, 'recall': 0.18699186991869918, 'f1_score': 0.05891524668671344}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 15, 'final_accuracy': 41.46341463414634, 'final_loss': 64.27730858325958, 'precision': 0.07945760427973776, 'recall': 0.2140921409214092, 'f1_score': 0.11583633028893285}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.001, 'epochs': 20, 'final_accuracy': 40.921409214092144, 'final_loss': 63.28587532043457, 'precision': 0.30862365073504594, 'recall': 0.21138211382113822, 'f1_score': 0.12958699645691515}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 5, 'final_accuracy': 38.82113821138211, 'final_loss': 65.64591312408447, 'precision': 0.278206482629798, 'recall': 0.2005420054200542, 'f1_score': 0.09498096868359383}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 10, 'final_accuracy': 41.666666666666664, 'final_loss': 63.41188007593155, 'precision': 0.3171550657091583, 'recall': 0.39295392953929537, 'f1_score': 0.3194691954329887}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 15, 'final_accuracy': 42.479674796747965, 'final_loss': 62.297402799129486, 'precision': 0.03296832426318843, 'recall': 0.18157181571815717, 'f1_score': 0.05580418189502996}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.01, 'weight_decay': 0.0001, 'epochs': 20, 'final_accuracy': 44.71544715447155, 'final_loss': 60.95726412534714, 'precision': 0.17454499681608834, 'recall': 0.23577235772357724, 'f1_score': 0.09720456026226401}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 5, 'final_accuracy': 39.83739837398374, 'final_loss': 67.02074778079987, 'precision': 0.5101052863775303, 'recall': 0.3712737127371274, 'f1_score': 0.314845489629434}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 10, 'final_accuracy': 39.769647696476966, 'final_loss': 66.4059146642685, 'precision': 0.2729070732441742, 'recall': 0.37669376693766937, 'f1_score': 0.3114465469999221}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 15, 'final_accuracy': 40.37940379403794, 'final_loss': 65.50816130638123, 'precision': 0.3409070914354884, 'recall': 0.37398373983739835, 'f1_score': 0.323893853277485}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.01, 'epochs': 20, 'final_accuracy': 41.39566395663957, 'final_loss': 65.1136234998703, 'precision': 0.4181511674193232, 'recall': 0.3875338753387534, 'f1_score': 0.3126096270422642}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 5, 'final_accuracy': 39.02439024390244, 'final_loss': 66.76673889160156, 'precision': 0.31441578570295575, 'recall': 0.3875338753387534, 'f1_score': 0.32713979899315526}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 10, 'final_accuracy': 39.56639566395664, 'final_loss': 65.36333274841309, 'precision': 0.46110426796000287, 'recall': 0.3875338753387534, 'f1_score': 0.33362538909965955}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 15, 'final_accuracy': 41.12466124661247, 'final_loss': 64.24604022502899, 'precision': 0.43078868373950047, 'recall': 0.4065040650406504, 'f1_score': 0.3495527462498651}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.001, 'epochs': 20, 'final_accuracy': 44.03794037940379, 'final_loss': 62.41681361198425, 'precision': 0.4603188142140458, 'recall': 0.44173441734417346, 'f1_score': 0.42088269557667135}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 5, 'final_accuracy': 38.550135501355015, 'final_loss': 66.41407799720764, 'precision': 0.2866341992372623, 'recall': 0.36585365853658536, 'f1_score': 0.31028316755677804}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 10, 'final_accuracy': 39.56639566395664, 'final_loss': 65.0210485458374, 'precision': 0.4586636503630996, 'recall': 0.3604336043360434, 'f1_score': 0.30944772493174877}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 15, 'final_accuracy': 40.447154471544714, 'final_loss': 64.14946830272675, 'precision': 0.3295859158291164, 'recall': 0.4092140921409214, 'f1_score': 0.33889444342124114}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.0001, 'epochs': 20, 'final_accuracy': 42.547425474254744, 'final_loss': 63.22741216421127, 'precision': 0.3308820437657123, 'recall': 0.4065040650406504, 'f1_score': 0.35677697789569274}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 5, 'final_accuracy': 23.644986449864497, 'final_loss': 73.63667285442352, 'precision': 0.2142204990284081, 'recall': 0.28184281842818426, 'f1_score': 0.16732563115104837}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 10, 'final_accuracy': 38.482384823848236, 'final_loss': 70.53874981403351, 'precision': 0.21915082903517205, 'recall': 0.35501355013550134, 'f1_score': 0.269800707227904}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 15, 'final_accuracy': 39.02439024390244, 'final_loss': 68.56926822662354, 'precision': 0.2642989419733741, 'recall': 0.3685636856368564, 'f1_score': 0.30101678228803613}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.01, 'epochs': 20, 'final_accuracy': 39.56639566395664, 'final_loss': 68.22540664672852, 'precision': 0.26096313083303735, 'recall': 0.38482384823848237, 'f1_score': 0.29441595018647404}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 5, 'final_accuracy': 29.065040650406505, 'final_loss': 73.31403195858002, 'precision': 0.2553518940806187, 'recall': 0.33875338753387535, 'f1_score': 0.2703328841993376}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 10, 'final_accuracy': 38.41463414634146, 'final_loss': 70.33924114704132, 'precision': 0.22019828731613156, 'recall': 0.36585365853658536, 'f1_score': 0.27474129086193055}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 15, 'final_accuracy': 39.363143631436316, 'final_loss': 68.5010517835617, 'precision': 0.22514231358359593, 'recall': 0.3712737127371274, 'f1_score': 0.2799449849259012}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.001, 'epochs': 20, 'final_accuracy': 39.295392953929536, 'final_loss': 67.72433233261108, 'precision': 0.2652205255680593, 'recall': 0.36314363143631434, 'f1_score': 0.2973242653392982}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 5, 'final_accuracy': 36.043360433604335, 'final_loss': 72.82512044906616, 'precision': 0.3621226714394416, 'recall': 0.36585365853658536, 'f1_score': 0.31402383563693487}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 10, 'final_accuracy': 37.87262872628726, 'final_loss': 70.32824635505676, 'precision': 0.23995559029339197, 'recall': 0.35772357723577236, 'f1_score': 0.2755007048706279}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 15, 'final_accuracy': 38.27913279132791, 'final_loss': 68.77304744720459, 'precision': 0.32093291203273794, 'recall': 0.37398373983739835, 'f1_score': 0.3099544016212877}\n",
      "{'dropout_rate': 0.5, 'learning_rate': 0.0001, 'weight_decay': 0.0001, 'epochs': 20, 'final_accuracy': 38.68563685636856, 'final_loss': 67.9121984243393, 'precision': 0.2681069186760244, 'recall': 0.37398373983739835, 'f1_score': 0.3032281758831481}\n"
     ]
    }
   ],
   "source": [
    "results = []\n",
    "# Inizia la grid search\n",
    "for dropout_rate in dropout_list:\n",
    "    for lr in learning_rate_list:\n",
    "        for weight_decay in weight_decay_list:\n",
    "            for epochs in epoch_list:\n",
    "                print(f\"Training with dropout {dropout_rate}, lr {lr}, wd {weight_decay}, epochs {epochs}...\")\n",
    "                \n",
    "                # Definisci il modello\n",
    "                model = SimpleCNN(dropout_rate=dropout_rate).to(device)\n",
    "                optimizer = optim.Adam(model.parameters(), lr=lr, weight_decay=weight_decay)\n",
    "                criterion = nn.CrossEntropyLoss()\n",
    "\n",
    "                # Training loop\n",
    "                model.train()\n",
    "                all_labels = []  # Per tenere traccia di tutte le etichette\n",
    "                all_predictions = []  # Per tenere traccia di tutte le predizioni\n",
    "\n",
    "                for epoch in range(epochs):\n",
    "                    epoch_loss = 0\n",
    "                    correct = 0\n",
    "                    total = 0\n",
    "                    for images, labels in train_loader:\n",
    "                        images, labels = images.to(device), labels.to(device)\n",
    "\n",
    "                        optimizer.zero_grad()\n",
    "                        outputs = model(images)\n",
    "                        loss = criterion(outputs, labels)\n",
    "                        loss.backward()\n",
    "                        optimizer.step()\n",
    "\n",
    "                        epoch_loss += loss.item()\n",
    "                        _, predicted = outputs.max(1)\n",
    "                        total += labels.size(0)\n",
    "                        correct += predicted.eq(labels).sum().item()\n",
    "                        \n",
    "                        # Aggiungi le etichette e le predizioni alla lista\n",
    "                        all_labels.extend(labels.cpu().numpy())\n",
    "                        all_predictions.extend(predicted.cpu().numpy())\n",
    "                    \n",
    "                    print(f\"Epoch {epoch+1}/{epochs}, Loss: {epoch_loss:.4f}, Accuracy: {100. * correct / total:.2f}%\")\n",
    "\n",
    "                # Ora valutiamo il modello sul test set\n",
    "                model.eval()\n",
    "                all_labels = []  # Per tenere traccia di tutte le etichette\n",
    "                all_predictions = []  # Per tenere traccia di tutte le predizioni\n",
    "\n",
    "                with torch.no_grad():  # Disabilita il calcolo del gradiente per la valutazione\n",
    "                    for images, labels in test_loader:\n",
    "                        images, labels = images.to(device), labels.to(device)\n",
    "\n",
    "                        outputs = model(images)\n",
    "                        _, predicted = outputs.max(1)\n",
    "\n",
    "                        all_labels.extend(labels.cpu().numpy())\n",
    "                        all_predictions.extend(predicted.cpu().numpy())\n",
    "\n",
    "                # Calcola precision, recall, f1 score sul test set\n",
    "                precision = precision_score(all_labels, all_predictions, average='weighted', zero_division=0)\n",
    "                recall = recall_score(all_labels, all_predictions, average='weighted', zero_division=0)\n",
    "                f1 = f1_score(all_labels, all_predictions, average='weighted', zero_division=0)\n",
    "\n",
    "                # Registra risultati finali\n",
    "                results.append({\n",
    "                    \"dropout_rate\": dropout_rate,\n",
    "                    \"learning_rate\": lr,\n",
    "                    \"weight_decay\": weight_decay,\n",
    "                    \"epochs\": epochs,\n",
    "                    \"final_accuracy\": 100. * correct / total,\n",
    "                    \"final_loss\": epoch_loss,\n",
    "                    \"precision\": precision,\n",
    "                    \"recall\": recall,\n",
    "                    \"f1_score\": f1\n",
    "                })\n",
    "\n",
    "# Mostra i risultati\n",
    "print(\"Results:\")\n",
    "for result in results:\n",
    "    print(result)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Find best parameters (tradeoff between Accuracy and F1-score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "\n",
    "# Converte la lista dei risultati in un DataFrame pandas\n",
    "results_df = pd.DataFrame(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Combinazione migliore per Accuracy e F1 Score:\n",
      "dropout_rate       0.250000\n",
      "learning_rate      0.010000\n",
      "weight_decay       0.000100\n",
      "epochs            20.000000\n",
      "final_accuracy    49.119241\n",
      "final_loss        57.996921\n",
      "precision          0.093230\n",
      "recall             0.241192\n",
      "f1_score           0.131551\n",
      "tradeoff_score    24.625396\n",
      "Name: 47, dtype: float64\n",
      "\n",
      "Top 5 combinazioni per Accuracy e F1 Score:\n",
      "    dropout_rate  learning_rate  weight_decay  epochs  final_accuracy  \\\n",
      "47          0.25           0.01        0.0001      20       49.119241   \n",
      "11          0.00           0.01        0.0001      20       46.476965   \n",
      "10          0.00           0.01        0.0001      15       46.070461   \n",
      "46          0.25           0.01        0.0001      15       45.934959   \n",
      "83          0.50           0.01        0.0001      20       44.715447   \n",
      "\n",
      "    final_loss  precision    recall  f1_score  tradeoff_score  \n",
      "47   57.996921   0.093230  0.241192  0.131551       24.625396  \n",
      "11   59.742588   0.154301  0.281843  0.199104       23.338034  \n",
      "10   61.101392   0.035446  0.186992  0.059596       23.065028  \n",
      "46   61.257022   0.034966  0.186992  0.058915       22.996937  \n",
      "83   60.957264   0.174545  0.235772  0.097205       22.406326  \n"
     ]
    }
   ],
   "source": [
    "# Calcolo del punteggio di trade-off\n",
    "results_df[\"tradeoff_score\"] = (results_df[\"final_accuracy\"] + results_df[\"f1_score\"]) / 2\n",
    "\n",
    "# Trovare la combinazione con il punteggio di trade-off massimo\n",
    "best_combination = results_df.sort_values(by=\"tradeoff_score\", ascending=False).iloc[0]\n",
    "\n",
    "print(\"Combinazione migliore per Accuracy e F1 Score:\")\n",
    "print(best_combination)\n",
    "\n",
    "# Visualizzare le prime 5 combinazioni\n",
    "top_5_combinations = results_df.sort_values(by=\"tradeoff_score\", ascending=False).head(5)\n",
    "print(\"\\nTop 5 combinazioni per Accuracy e F1 Score:\")\n",
    "print(top_5_combinations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA1cAAAIjCAYAAADvBuGTAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjAsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvlHJYcgAAAAlwSFlzAAAPYQAAD2EBqD+naQAA4DJJREFUeJzs3Xd0VNXawOHf9JmUSa+QQBJ671ykFwmIIooFVLpiARVRUbxeREW5dvRK+VQEpIgN0WsBkSYiTZp0AiS0FEggPdPP9weXkSEVSIX3WStrZfbZ55x9Zk4m887e+90qRVEUhBBCCCGEEEJcE3VVN0AIIYQQQgghrgcSXAkhhBBCCCFEOZDgSgghhBBCCCHKgQRXQgghhBBCCFEOJLgSQgghhBBCiHIgwZUQQgghhBBClAMJroQQQgghhBCiHEhwJYQQQgghhBDlQIIrIYQQQgghhCgHElwJIUQV6dGjBz169Kiy8y9cuJBGjRqh0+nw9/d3l7/11lvExsai0Who1apVlbVPCCGEqGkkuBJC3LBUKlWZftatW1fVTS13Bw8eZOTIkcTFxfHxxx/z0UcfAfDLL78wadIkOnfuzLx583j99deLPcaSJUuYMWNGJbXYk9PpJDIyEpVKxc8//1wlbbgRrVu3rti/kyFDhrjrbd26lccee4y2bdui0+lQqVRXdB6bzcb7779P69atMZvN+Pv707RpU8aOHcvBgwfL+7KEEKLcaKu6AUIIUVUWLlzo8fizzz5j1apVhcobN25cmc2qFOvWrcPlcvH+++9Tr149d/maNWtQq9XMnTsXvV5f4jGWLFnC3r17mTBhQgW3trA1a9aQkpJC3bp1Wbx4Mf3796/0NtzInnjiCdq3b+9RVrduXffvP/30E5988gktWrQgNjaWw4cPX9HxBw8ezM8//8zQoUN56KGHsNvtHDx4kB9++IGbbrqJRo0alcdlCCFEuZPgSghxw3rggQc8Hm/evJlVq1YVKr9cfn4+Xl5eFdm0CnfmzBkAj+GAF8tNJlOpgdWVslgs6PV61OryGTCxaNEi2rRpw4gRI3jhhRfIy8vD29u7XI5dnhwOBy6Xq9yfz6rWtWtX7rrrrmK3P/roozz33HOYTCbGjx9/RcHVtm3b+OGHH3jttdd44YUXPLZ9+OGHZGZmXm2zr1h537dCiOufvFsIIUQJevToQbNmzdi+fTvdunXDy8vL/YHvu+++Y8CAAURGRmIwGIiLi+PVV1/F6XQWOs5HH31EXFwcJpOJDh06sGHDhiLPZ7Vaeemll6hXrx4Gg4GoqCgmTZqE1Wotc5tnzZpF06ZNMRgMREZGMm7cOI8PpHXr1uWll14CICQkBJVKxdSpU1GpVMybN4+8vDz3UK/58+cX+7z8+OOPHD9+3F33Ys/FxaFjS5cu5cUXX6RWrVp4eXmRnZ3NuXPneOaZZ2jevDk+Pj6YzWb69+/P7t27y3x9BQUFfPvttwwZMoR77rmHgoICvvvuuyLr/vzzz3Tv3h1fX1/MZjPt27dnyZIlHnW2bNnCLbfcQkBAAN7e3rRo0YL333/f41qLmhs3cuRIj96apKQkVCoVb7/9NjNmzCAuLg6DwcD+/fux2WxMmTKFtm3b4ufnh7e3N127dmXt2rWFjnuxR7F58+YYjUZCQkLo168ff/75JwDdu3enZcuWRV5vw4YNiY+PL+0p5Oeff6Zr1654e3vj6+vLgAED2LdvX6n7lVVYWBgmk+mq9j169CgAnTt3LrRNo9EQFBTkUXb69GnGjBnj/juMiYnh0UcfxWazuescO3aMu+++m8DAQLy8vPjHP/7Bjz/+6HGcku5buHCf9OvXDz8/P7y8vOjevTsbN268qmsUQly/pOdKCCFKkZGRQf/+/RkyZAgPPPAAYWFhAMyfPx8fHx8mTpyIj48Pa9asYcqUKWRnZ/PWW2+59587dy4PP/wwN910ExMmTODYsWMMHDiQwMBAoqKi3PVcLhcDBw7k999/Z+zYsTRu3Jg9e/bw3nvvcfjwYZYvX15qW6dOncrLL79Mnz59ePTRRzl06BCzZ89m27ZtbNy4EZ1Ox4wZM/jss8/49ttvmT17Nj4+PrRo0YJ69erx0UcfsXXrVj755BMAbrrppiLP889//pOsrCxOnTrFe++9B4CPj49HnVdffRW9Xs8zzzyD1WpFr9ezf/9+li9fzt13301MTAxpaWn83//9H927d2f//v1ERkaWeo3ff/89ubm5DBkyhPDwcHr06MHixYu57777POrNnz+f0aNH07RpUyZPnoy/vz87d+5kxYoV7rqrVq3i1ltvJSIigieffJLw8HAOHDjADz/8wJNPPllqW4oyb948LBYLY8eOxWAwEBgYSHZ2Np988ol7mFtOTg5z584lPj6erVu3eiQOGTNmDPPnz6d///48+OCDOBwONmzYwObNm2nXrh3Dhg3joYceYu/evTRr1sy937Zt2zh8+DAvvvhiie1buHAhI0aMID4+njfeeIP8/Hxmz55Nly5d2Llzp0fAWJycnBzS09M9ygIDA8ulh6dOnToALF68mM6dO6PVFv9RJTk5mQ4dOpCZmcnYsWNp1KgRp0+f5uuvvyY/Px+9Xk9aWho33XQT+fn5PPHEEwQFBbFgwQIGDhzI119/zR133OFxzKLu2zVr1tC/f3/atm3LSy+9hFqtZt68efTq1YsNGzbQoUOHa75uIcR1QhFCCKEoiqKMGzdOufxtsXv37gqgzJkzp1D9/Pz8QmUPP/yw4uXlpVgsFkVRFMVmsymhoaFKq1atFKvV6q730UcfKYDSvXt3d9nChQsVtVqtbNiwweOYc+bMUQBl48aNJbb/zJkzil6vV/r27as4nU53+YcffqgAyqeffuoue+mllxRAOXv2rMcxRowYoXh7e5d4nosGDBig1KlTp1D52rVrFUCJjY0t9BxZLBaPtimKoiQmJioGg0F55ZVXynTeW2+9VencubP78UcffaRotVrlzJkz7rLMzEzF19dX6dixo1JQUOCxv8vlUhRFURwOhxITE6PUqVNHOX/+fJF1FOXCPXDp63TRiBEjPK4/MTFRARSz2ezRlovnuvT1VxRFOX/+vBIWFqaMHj3aXbZmzRoFUJ544olC57vYpszMTMVoNCrPPfecx/YnnnhC8fb2VnJzcwvte1FOTo7i7++vPPTQQx7lqampip+fX6Hyy118bYv6SUxMLHKfov6uSuJyudx/d2FhYcrQoUOVmTNnKsePHy9Ud/jw4YparVa2bdtW5HEURVEmTJigAB5/Vzk5OUpMTIxSt25d9/1Y3H3rcrmU+vXrK/Hx8R73RX5+vhITE6PcfPPNZb42IcT1T4YFCiFEKQwGA6NGjSpUfumwp4vf5Hft2pX8/Hx3RrM///yTM2fO8Mgjj3jMuxk5ciR+fn4ex/vqq69o3LgxjRo1Ij093f3Tq1cvgCKHkF3q119/xWazMWHCBI8ehIceegiz2VxoGFRFGzFiRKGhYQaDwd02p9NJRkYGPj4+NGzYkB07dpR6zIyMDFauXMnQoUPdZYMHD0alUvHll1+6y1atWkVOTg7PP/88RqPR4xgXM9ft3LmTxMREJkyYUGju2ZVmt7vU4MGDCQkJ8SjTaDTu19/lcnHu3DkcDgft2rXzuO5vvvkGlUrlHrZZVJv8/Py4/fbb+fzzz1EUBbjwXH7xxRcMGjSoxLlnq1atIjMzk6FDh3rcYxqNho4dO5Z6j100ZcoUVq1a5fETHh5epn1Lo1KpWLlyJdOmTSMgIIDPP/+ccePGUadOHe699173EFeXy8Xy5cu57bbbaNeuXZHHgQvJNTp06ECXLl3c23x8fBg7dixJSUns37/fY7/L79tdu3aRkJDAfffdR0ZGhvs5y8vLo3fv3vz222+4XK5yuXYhRM0nwwKFEKIUtWrVKjIhwb59+3jxxRdZs2aNe17GRVlZWQAcP34cgPr163ts1+l0xMbGepQlJCRw4MCBQh/ML7qYhOLcuXMe80lMJhN+fn7uczVs2NBjP71eT2xsrHt7ZYmJiSlUdnE+0axZs0hMTPSYn3b5XJqifPHFF9jtdlq3bs2RI0fc5R07dmTx4sWMGzcO+HvezqXD5i5XljpXo6jrBliwYAHvvPMOBw8exG63F1n/6NGjREZGEhgYWOI5hg8fzhdffMGGDRvo1q0bv/76K2lpaQwbNqzE/RISEgDcAfvlzGZziftf1Lx5c/r06VOmulfDYDDwz3/+k3/+85+kpKSwfv163n//fb788kt0Oh2LFi3i7NmzZGdnl/r6HT9+nI4dOxYqv5gF9Pjx4x7HuPz1u/icjRgxothzZGVlERAQUObrE0JcvyS4EkKIUhQ1MT8zM5Pu3btjNpt55ZVXiIuLw2g0smPHDp577rmr+ibb5XLRvHlz3n333SK3X5yfdeedd7J+/Xp3+YgRI4pNPFGVinreXn/9df71r38xevRoXn31Vfc8nQkTJpTpOVu8eDFQdLIDuJC44PKg9VqpVCp3D9GlikpcAkVf96JFixg5ciSDBg3i2WefJTQ0FI1Gw/Tp091B3pWIj48nLCyMRYsW0a1bNxYtWkR4eHipAc/F53jhwoVF9jSVNL+pqkRERDBkyBAGDx5M06ZN+fLLLyv0fr/89bv4nL311lvFLqp9+XxDIcSNq/q9iwohRA2wbt06MjIyWLZsGd26dXOXJyYmetS7ODk/ISHBo7fAbreTmJjokfUtLi6O3bt307t37xKHpb3zzjucP3/e/fhiEoiL5zp06JBHgGGz2UhMTCz3noarGTr39ddf07NnT+bOnetRnpmZSXBwcIn7JiYm8scffzB+/Hi6d+/usc3lcjFs2DCWLFnCiy++SFxcHAB79+71WMfrUpfWKem5CQgI4NixY4XKr6Qn8OuvvyY2NpZly5Z5PG+XD/+Li4tj5cqVnDt3rsTeK41Gw3333cf8+fN54403WL58OQ899BAajabEdly85tDQ0ArteaoIOp2OFi1akJCQQHp6OqGhoZjNZvbu3VvifnXq1OHQoUOFyi8O3b34d1Oci8+Z2Wyucc+ZEKLyyZwrIYS4Chc/xF7ao2Gz2Zg1a5ZHvXbt2hESEsKcOXM8hvLNnz+/0Ho999xzD6dPn+bjjz8udL6CggLy8vIAaNu2LX369HH/NGnSBIA+ffqg1+v54IMPPNo1d+5csrKyGDBgwLVd9GW8vb3dwx/LSqPRFOoF+uqrrzh9+nSp+17stZo0aRJ33XWXx88999xD9+7d3XX69u2Lr68v06dPx2KxeBzn4vnbtGlDTEwMM2bMKPRaXNrGuLg4Dh48yNmzZ91lu3fvvqI03EXdL1u2bGHTpk0e9QYPHoyiKLz88suFjnH58zZs2DDOnz/Pww8/TG5ubqnrs8GFHi+z2czrr7/uMTTxokuvsaokJCRw4sSJQuWZmZls2rSJgIAAQkJCUKvVDBo0iP/+97/uNPWXuvh83XLLLWzdutXjuc7Ly+Ojjz6ibt267r+f4rRt25a4uDjefvttcnNzC22vDs+ZEKL6kJ4rIYS4CjfddBMBAQGMGDGCJ554ApVKxcKFCwt9ANbpdEybNo2HH36YXr16ce+995KYmMi8efMKDV8bNmwYX375JY888ghr166lc+fOOJ1ODh48yJdffsnKlSuLnLh/UUhICJMnT+bll1+mX79+DBw4kEOHDjFr1izat29fpg/fV6Jt27Z88cUXTJw4kfbt2+Pj48Ntt91W4j633norr7zyCqNGjeKmm25iz549LF68uExD+RYvXkyrVq080tdfauDAgTz++OPs2LGDNm3a8N577/Hggw/Svn177rvvPgICAti9ezf5+fksWLAAtVrN7Nmzue2222jVqhWjRo0iIiKCgwcPsm/fPlauXAnA6NGjeffdd4mPj2fMmDGcOXOGOXPm0LRp00Jz7Uq67mXLlnHHHXcwYMAAEhMTmTNnDk2aNPH4wN6zZ0+GDRvGBx98QEJCAv369cPlcrFhwwZ69uzJ+PHj3XVbt25Ns2bN3IlQ2rRpU2o7zGYzs2fPZtiwYbRp04YhQ4YQEhLCiRMn+PHHH+ncuTMffvhhma6pJMePH2fhwoUA7sBn2rRpwIWeopLmhu3evZv77ruP/v3707VrVwIDAzl9+jQLFiwgOTmZGTNmuIPV119/nV9++YXu3bu7ly9ISUnhq6++4vfff8ff35/nn3+ezz//nP79+/PEE08QGBjIggULSExM5Jtvvik1fbxareaTTz6hf//+NG3alFGjRlGrVi1Onz7N2rVrMZvN/Pe//73m50wIcZ2omiSFQghR/RSXir1p06ZF1t+4caPyj3/8QzGZTEpkZKQyadIkZeXKlQqgrF271qPurFmzlJiYGMVgMCjt2rVTfvvttyJTfNtsNuWNN95QmjZtqhgMBiUgIEBp27at8vLLLytZWVlluo4PP/xQadSokaLT6ZSwsDDl0UcfLZRqvDxSsefm5ir33Xef4u/vrwDutOQXU1p/9dVXhfaxWCzK008/rURERCgmk0np3LmzsmnTpmLTnV+0fft2BVD+9a9/FVsnKSlJAZSnnnrKXfb9998rN910k2IymRSz2ax06NBB+fzzzz32+/3335Wbb75Z8fX1Vby9vZUWLVoo//nPfzzqLFq0SImNjVX0er3SqlUrZeXKlcWmYn/rrbcKtc3lcimvv/66UqdOHcVgMCitW7dWfvjhh0LHUJQLadvfeustpVGjRoper1dCQkKU/v37K9u3by903DfffFMBlNdff73Y56Uoa9euVeLj4xU/Pz/FaDQqcXFxysiRI5U///yz1P2Ke22LqlfUT0mvs6IoSlpamvLvf/9b6d69uxIREaFotVolICBA6dWrl/L1118Xqn/8+HFl+PDhSkhIiGIwGJTY2Fhl3LhxHqnvjx49qtx1112Kv7+/YjQalQ4dOig//PDDFV3bzp07lTvvvFMJCgpSDAaDUqdOHeWee+5RVq9eXeL1CCFuLCpFKWKWrhBCCCGqvffff5+nnnqKpKQkoqOjq7o5Qghxw5PgSgghhKiBFEWhZcuWBAUFlXl9KiGEEBVL5lwJIYQQNUheXh7ff/89a9euZc+ePXz33XdV3SQhhBD/Iz1XQgghRA2SlJRETEwM/v7+PPbYY7z22mtV3SQhhBD/I8GVEEIIIYQQQpSDarHO1cyZM6lbty5Go5GOHTuydevWMu23dOlSVCoVgwYN8igfOXIkKpXK46dfv34V0HIhhBBCCCGEuKDKg6uLa6S89NJL7Nixg5YtWxIfH8+ZM2dK3C8pKYlnnnmGrl27Frm9X79+pKSkuH8+//zzimi+EEIIIYQQQgDVYFhgx44dad++vXvRQpfLRVRUFI8//jjPP/98kfs4nU66devG6NGj2bBhA5mZmSxfvty9feTIkYXKroTL5SI5ORlfX19UKtVVHUMIIYQQQghR8ymKQk5ODpGRkaUuPF6l2QJtNhvbt29n8uTJ7jK1Wk2fPn3YtGlTsfu98sorhIaGMmbMGDZs2FBknXXr1hEaGkpAQAC9evVi2rRpBAUFFVnXarVitVrdj0+fPk2TJk2u8qqEEEIIIYQQ15uTJ09Su3btEutUaXCVnp6O0+kkLCzMozwsLIyDBw8Wuc/vv//O3Llz2bVrV7HH7devH3feeScxMTEcPXqUF154gf79+7Np0yY0Gk2h+tOnT+fll18uVP7JJ5/g5eV1ZRclhBBCCCGEuGoOxcGajDUA9ArqhVZVtatH5efn8+CDD+Lr61tq3Rq1zlVOTg7Dhg3j448/Jjg4uNh6Q4YMcf/evHlzWrRoQVxcHOvWraN3796F6k+ePJmJEye6H2dnZxMVFcWgQYMwm83lexHXCbvdzqpVq7j55pvR6XRV3RxRTcl9IspC7hNRVnKviLKQ+6Tmy7PlMfzt4QC8N/o9vPXe5X6OK7lPsrOzefDBB8s0XahKg6vg4GA0Gg1paWke5WlpaYSHhxeqf/ToUZKSkrjtttvcZS6XCwCtVsuhQ4eIi4srtF9sbCzBwcEcOXKkyODKYDBgMBgKlet0OvmjLIU8R6Is5D4RZSH3iSgruVdEWch9UnPplL9ft4p+Hcty/Cs5f5VmC9Tr9bRt25bVq1e7y1wuF6tXr6ZTp06F6jdq1Ig9e/awa9cu98/AgQPp2bMnu3btIioqqsjznDp1ioyMDCIiIirsWoQQQgghhBA3tiofFjhx4kRGjBhBu3bt6NChAzNmzCAvL49Ro0YBMHz4cGrVqsX06dMxGo00a9bMY39/f38Ad3lubi4vv/wygwcPJjw8nKNHjzJp0iTq1atHfHx8pV6bEEIIIYQQ4sZR5cHVvffey9mzZ5kyZQqpqam0atWKFStWuJNcnDhxotSUh5fSaDT89ddfLFiwgMzMTCIjI+nbty+vvvpqkUP/rpaiKDgcDpxOZ7kdsyax2+1otVosFssN+xxUBxqNBq1WK0sGCCGEEEJUA1UeXAGMHz+e8ePHF7lt3bp1Je47f/58j8cmk4mVK1eWU8uKZrPZSElJIT8/v0LPU50pikJ4eDgnT56UD/ZVzMvLi4iICPR6fVU3RQghhBDihlYtgquaxOVykZiYiEajITIyEr1ef0MGFy6Xi9zcXHx8fK6oZ1GUH0VRsNlsnD17lsTEROrXry+vhRBCCCFEFZLg6grZbDZcLhdRUVE39BpYLpcLm82G0WiUD/RVyGQyodPpOH78uPv1EEIIIYSoyQxaAz8M/cH9e00iwdVVkoBCVBdyLwohhBDieqJVaxnQYEBVN+OqyKcyIYQQQgghhCgH0nMlhBBCCCGEqDbsTjuL9ywG4P7m96PT1JzFoCW4EkIIIYQQQlQbNqeNUd9dWPP27iZ316jgSoYF3kB+++03brvtNiIjI1GpVCxfvty9zW6389xzz9G8eXO8vb2JjIxk+PDhJCcnl3jMl19+GZVK5fHTqFGjCr4SIYQQQgghqh8JrqqItcBGRmomJw+nkJGaibXAVuHnzMvLo2XLlsycObPQtvz8fHbs2MG//vUvduzYwbJlyzh06BADBw4s9bhNmzYlJSXF/fP7779XRPOFEELcYPLsVtItOaRbcsh3VPz/SSGEuFYyLLAKWAtsbPt1D28+PBebxY7eqGPS/42hfZ/mGEwVtxBs//796d+/f5Hb/Pz8WLVqlUfZhx9+SIcOHThx4gTR0dHFHler1RIeHl6ubRVCCHHjyndYSSnI4qOEdWxJP4pKBd1CGzGi7k1V3TQhhCiR9FxVgdysfHdgBWCz2HnrkU/Jy8qv4pZ5ysrKQqVS4e/vX2K9hIQEIiMjiY2N5f777+fEiROV00AhhBDXnQKHjQ1nDnPXbx+yMmUPmfZ8ztvy+e7UDh74fQ4ANqejilsphBBFk+CqCuRnF7gDq4usBTbycgqqqEWFWSwWnnvuOYYOHYrZbC62XocOHZg/fz4rVqxg9uzZJCYm0rVrV3JyciqxtUIIIa4XWfYCXtj1NQpKoW1OxQVAtsNS2c0SQogykWGBVcDLbEJv1HkEWAaTHm9fUxW26m92u5177rkHRVGYPXt2iXX79+/vXsS2RYsWdOzYkTp16vDll18yZsyYymiuEEKI64RLcfHj6V3uIKo4W9OPcrtPu0pqlRBClJ30XFUBHz8vJv3fGPTGC2klDSY9z84ZjbefVxW37O/A6vjx46xatarEXqui+Pv706BBA44cOVJBLRRCCHG9sjod/HX+VKn1dp0/iaIU7tkSQlwfDFoDX971JV/e9SUGraGqm3NFpOeqChhMetr3ac78Ha+Tl1OAt68Jbz+vCk1mURYXA6uEhATWrl1LUFDQFR8jNzeXo0ePMmzYsApooRBCiOuZVq3BrDOWWs+sM6JSqSqhRUKIqqBVa7m76d1V3YyrIj1XVcRg0hMY7k9U/QgCw/0rJbDKzc1l165d7Nq1C4DExER27drFiRMnsNvt3HXXXfz5558sXrwYp9NJamoqqamp2Gx/p7/t3bs3H374ofvxs88+y/r160lKSuKPP/7gjjvuQKPRMHTo0Aq/HiGEENcXnVrDPXU6lFrv1lotK6E1Qghx5aTn6gby559/0rNnT/fjiRMnAjBixAimTp3K999/D0CrVq089lu7di09evQA4OjRo6Snp7u3nTp1iqFDh5KRkUFISAhdunRh8+bNhISEVOzFCCGEuC7V9gqkdUAddp4/XmydAL1PJbZICFHZHC4H3x74FoA7Gt+BVl1zQpaa01JxzXr06FHiGPWyjF9PSkoCwOW6MNn4888/dye0EEIIIa5VgMGbt9sO4c19P7IqZR+u/2UN1Ko03FqrFRwHP331SAAlhKgYVoeVe76+B4Dcyblo9TUnZKk5LRVCCCHEDSHI4MOLzW9nYuN+HM09ixqI9Q3F4NKw7vivVd08IYQolgRXQgghhKh2fHVGfHVGwkx+7jK73V7CHkIIUfVkPJcQQgghhBBClAPpuRJCCCGEEBUi35GNxZnLwezNWJx5RJrqU8urASaND1p11S5BI0RFkOBKCCGEEEKUu1z7eZafmsGxvJ0e5Qa1F7fVepxYn9YYNV5V1DohKoYMCxRCCCGEEOUqz5HF0hPTCgVWAFZXPl+ffINT+QfLlKlYiJpEeq6EEEIIIUS5Oms9SXJBQol1VqR8xMjYf+Oj9a+cRokaQ6/RM+/2ee7faxIJroQQQgghRLmxOvPZnP5tqfXO2ZKxOHMluBKF6DQ6RrYaWdXNuCoyLFAIIYQQQpQbh2Inx36+THXzHdkV3BohKpf0XAkhhBBCiHKjVenw0fmDpfS6XlrfCm+PqHkcLgcrj6wEIL5ePFp1zQlZpOeqCp3PyifjfB7ns/Ir5Xy//fYbt912G5GRkahUKpYvX+6xfeTIkahUKo+ffv36lXrcmTNnUrduXYxGIx07dmTr1q0VdAVCCCGEqO4MGi86Bd1Rar1AfQRGtQRXojCrw8qtn9/KrZ/fitVhrermXBEJrqpAXr6VP3cf5+mXv2bwQ//H0y9/zZ+7j5OXX7E3T15eHi1btmTmzJnF1unXrx8pKSnun88//7zEY37xxRdMnDiRl156iR07dtCyZUvi4+M5c+ZMeTdfCCGEEDVEiDGaCGNciXX6RYzFW+tXSS0SonJIcFUFDiSk8tTUr0hIPIPT6SIh8QxPTf2KAwmpFXre/v37M23aNO64o/hvkwwGA+Hh4e6fgICAEo85Y8YMHnroIUaNGkWTJk2YM2cOXl5efPrpp+XdfCGEEELUEN5aP4bWmUJd7xaFtunVJgZHPUuUV2NUKlUVtE6IilNzBjBeJ85n5TNrwfoit81asJ53XrqLAL+qW1Bv3bp1hIaGEhAQQK9evZg2bRpBQUFF1rXZbGzfvp3Jkye7y9RqNX369GHTpk2V1WQhhBBCVEM+ugDuippEgTOXA9kbsTjziTTVI9qrCUaNN1p1zUqxLURZSHBVyVwuhWMn0ovclngyHZer6hbT69evH3feeScxMTEcPXqUF154gf79+7Np0yY0Gk2h+hkZGTidTsLCwjzKw8LCOHjwYGU1WwghhBDVlJfWjJfWTJeQu6u6KUJUCgmuKplarSI2OpiExMJzkmKiglGrq657fMiQIe7fmzdvTosWLYiLi2PdunX07t27ytolhBBCCCFETSBzripZgJ8Xj43oXuS2x0Z0r9IhgZeLjY0lODiYI0eOFLk9KCgIjUZDWlqaR3laWhrh4eGV0UQhhBBCCCGqDQmuqkDj+uG8N/Vu6seEotWqqR8TyntT76Zx/eoVkJw6dYqMjAwiIiKK3K7X62nbti2rV692l7lcLlavXk2nTp0qq5lCCCGEEOI6otfo+bD/h3zY/0P0mpo1N0+GBVYBby8D7VrW4Z2X7sLlUlCrVZXSY5Wbm+vRC5WYmMiuXbsIDAwkMDCQl19+mcGDBxMeHs7Ro0eZNGkS9erVIz4+3r1P7969ueOOO3jssccAmDBhAqNGjaJdu3Z06NCBGTNmkJeXx6hRoyr8eoQQQgghxPVHp9ExrsO4qm7GVZHgqgpV9hDAP//8k549e7ofT5w4EYARI0Ywe/Zs/vrrLxYsWEBmZiaRkZH07duXV199FYPB4N7n6NGjpKf/nZDj3nvvJSMjgylTppCamkqrVq1YsWJFoSQXQgghhBBCXO8kuLqB9OjRA0UpPhvhypUrSz1GUlIScGH430Xjx49n/Pjx19w+IYQQQgghnC4nG05sAKBrdFc06sJZq6srCa6EEEIIIYQQ1YbFYaHnggujrXIn5+Kt967iFpWdJLQQQgghhBBCiHIgwZUQQgghhBBClAMJroQQQgghhBCiHEhwJYQQQgghhBDlQIIrIYQQQgghhCgHki1QCCGEENctp8tFgdOKWqXGS2sofQchhLgG1aLnaubMmdStWxej0UjHjh3ZunVrmfZbunQpKpWKQYMGeZQrisKUKVOIiIjAZDLRp08fEhISKqDlQgghhKiOLA47aZZMFiau44Xdi5ny1+dsOLOfc9acqm6aEKIUOo2ON/u8yZt93kSn0VV1c65IlfdcffHFF0ycOJE5c+bQsWNHZsyYQXx8PIcOHSI0NLTY/ZKSknjmmWfo2rVroW1vvvkmH3zwAQsWLCAmJoZ//etfxMfHs3//foxGY0VejhBCCCGqWIHTxp/njjB59yKcyt+L3m9MP0iEKYBZ7cYSbgrw2MfmtJPjKOCcLQeX4iLIYMZbY8Kk1Vd284W44ek1ep7t/GxVN+OqVHnP1bvvvstDDz3EqFGjaNKkCXPmzMHLy4tPP/202H2cTif3338/L7/8MrGxsR7bFEVhxowZvPjii9x+++20aNGCzz77jOTkZJYvX17BVyOEEEKIqpZhzSkUWF2UUnCep3fO55w1112WZctjYdKvPLDpDR7a+h4Pb3ufoRtf5z8Jyzlvk54uIUTZVWnPlc1mY/v27UyePNldplar6dOnD5s2bSp2v1deeYXQ0FDGjBnDhg0bPLYlJiaSmppKnz593GV+fn507NiRTZs2MWTIkELHs1qtWK1W9+Ps7GwA7HY7drvdo67dbkdRFFwuFy5X4TftsrA5nWRbrKw/ksjBtHQahQXTrV4M/kYDOo3mqo5ZFr/99htvv/02O3bsICUlhW+++cZjSKWmmHO/8cYbPPPMMx5liqIAMHXqVF599VWPbQ0bNmT//v3l23hRLJfLhaIo2O32Yl/DqnLx7+fyvyMhLiX3iSirstwrVqeNz4/+hsalQkPR74mnctJJzTuHr9pAjj2f/zvyI2vP7AZAf3EfBX49vZ2T2Wd4qdkD+Om9y/diRIWR95Saz+lysjN1JwCtw1ujUZf/55sruU+u5F6q0uAqPT0dp9NJWFiYR3lYWBgHDx4scp/ff/+duXPnsmvXriK3p6amuo9x+TEvbrvc9OnTefnllwuV//LLL3h5eXmUabVawsPDyc3NxWazFXm8kigqNUcys3hwyXIsDoe73KjV8sl9g6jn74eqiG/aysPZs2dp1KgRQ4YMYdiwYRQUFLgDSaDQc/7rr7/y+OOP07dvX496l7LZbDRq1MijV1Cr1RZbX5Q/m81GQUEBv/32G45L7qnqZNWqVVXdBFEDyH0iyqq0e6UBaibSocQ6CRt2ksCFD29N8KYJNxVdMQ82pq6/qnaKqiXvKTWXxWlhyJ4LHSJLmy/FqKm4aT1luU/y8/PLfLwqn3N1JXJychg2bBgff/wxwcHB5XbcyZMnM3HiRPfj7OxsoqKi6Nu3L2az2aOuxWLh5MmT+Pj4XNX8rfS8/EKBFYDF4eDBJcv5ZfwoQrx9ru5CSjF48GAGDx4MwLBhwzCZTB7Xd/m1Tpo0iZ49e9KiRYtCx1IUhZycHPR6PQaDgfr161dIm0XpLBYLJpOJbt26Vbs5hXa7nVWrVnHzzTej09WsCami8sh9IsqqLPfKOVsuT2z/hNSCzBKPNSq2F0Oiu/DR0Z/4IXlziXVb+cfxQpOheOuq13usKJq8p9R8ebY82HPh9/j4eLwroOf4Su6TK+k0qNLgKjg4GI1GQ1pamkd5Wloa4eHhheofPXqUpKQkbrvtNnfZxaF5Wq2WQ4cOufdLS0sjIiLC45itWrUqsh0GgwGDoXB6Vp1OV+jJdjqdqFQq1Go1avWVT1lbfySpUGB1kcXhYMORJO5q3eyKj3s1SrqGtLQ0fvrpJxYsWFBknYvPu0qlIiEhgdq1a2M0GunUqRPTp08nOjq6Qtsu/qZWq1GpVEXer9VFdW6bqD7kPhFlVdK94q/2oUNIA74+Wfz0AoDO4U1Aq+KMPRObylli3bOObBQNcn/WMPKeUnPplL9ft4p+Hcty/Cs5f5UmtNDr9bRt25bVq1e7y1wuF6tXr6ZTp06F6jdq1Ig9e/awa9cu98/AgQPp2bMnu3btIioqipiYGMLDwz2OmZ2dzZYtW4o8ZmU7mHb2mrZXlgULFuDr68udd95ZYr0OHTowf/58VqxYwezZs0lMTKRr167k5MgEYCGEEJXPoNExpE5XtKri52g08I0kWO+LTqWllldIqccMNwagr2HpoIUQVaPKhwVOnDiRESNG0K5dOzp06MCMGTPIy8tj1KhRAAwfPpxatWoxffp0jEYjzZp59ur4+/sDeJRPmDCBadOmUb9+fXcq9sjIyELrYVWFRmElv4mXtr2yfPrpp9x///2lDjPr37+/u2erRYsWdOzYkTp16vDll18yZsyYymiqEEII4SHY4Ms7bUby7M4F2Fyeo0XqeIXwVusRBBguDMEfWOsffHFiXYnHG1q3J95aGRIohChdlQdX9957L2fPnmXKlCmkpqbSqlUrVqxY4U5IceLEiSsefjdp0iTy8vIYO3YsmZmZdOnShRUrVlSL+Sjd69XFqNUWOTTQqNXSrV7dym/UZTZs2MChQ4f44osvrnhff39/GjRowJEjRyqgZUIIIUTpDBodLf3r8k2XSaw5s4c/M46iV2sZVLsDsT7hBBr+ntts1nlxT3R3vjxRdNKKTkFNiPYKK3KbEEJcrsqDK4Dx48czfvz4IretW7euxH3nz59fqEylUvHKK6/wyiuvlEPrype/ycTc++9gzOJvC2ULnHv/nfibTFXYugvmzp1L27Ztadmy5RXvm5uby9GjRxk2bFgFtEwIIYQoG4NGh0Gj466oTtxWqz0a1Og1hT/2+Oq8eKBub6K9QliQuIqz1iwAzFovBkd35fZaN+EvadiFEGVULYKrG4leq6F5ZDirxo/ityNJHEw7S6OwELrVq4u/yYReW3HrFOXm5nr0KCUmJrJr1y4CAwPdCSiys7P56quveOedd4o8Ru/evbnjjjt47LHHAHj22WcZOHAgderUITk5mZdeegmNRsPQoUMr7DqEEEKIslKr1Jg0+hLrmHVe9I/swE3BTbE4bSgoGDV6zDovdGr5qCREZdNpdLzU/SX37zWJvGNUAYNWS6ivT6VlBbzozz//pGfPnu7HF9PPjxgxwt0DuHTpUhRFKTY4Onr0KOnp6e7Hp06dYujQoWRkZBASEkKXLl3YvHkzISHVY+6YEEIIURYalZpAg29VN0MIAeg1eqb2mFrVzbgqElzdQHr06IGiKCXWGTt2LGPHji12e1JSEvB3KvbPP//8qlLSCyGEEEIIcb2R4EoIIYQQQghRbbgUFwfOHgCgcUhj1Kqa80W+BFdCCCGEEEKIaqPAXkCz2Remz+ROzsW7BiWVqTlhoBBCCCGEEEJUYxJcCSGEEEIIIUQ5kOBKCCGEEEIIIcqBBFdCCCGEEEIIUQ4kuBJCCCGEEEKIciDBlRBCCCGEEEKUA0nFLoQQQgghhKg2dBodz3R6xv17TSLBVRUpcNjIcVhxuJxo1Rp8tQZMWn1VN0sIIYQQQogqpdfoeavvW1XdjKsiwwIrmaIopBZk8fqen4lf9T43r5pB/Kr3eX3Pz6QWZKEoSoWcd/r06bRv3x5fX19CQ0MZNGgQhw4d8qhjsVgYN24cQUFB+Pj4MHjwYNLS0kq9nilTphAREYHJZKJPnz4kJCRUyDUIIYQQQghRnUlwVcnSLNkM+e0Tlp3Yic3lAMDmcrDsxE6G/PYJaZacCjnv+vXrGTduHJs3b2bVqlXY7Xb69u1LXl6eu85TTz3Ff//7X7766ivWr19PcnIyd955Z4nHfeutt/jggw+YM2cOW7Zswdvbm/j4eCwWS4VchxBCCCFKZ3PayLJlsuP8n6xM+YmtGZvJtJ3H6rRWddOEKJVLcZGUmURSZhIuxVXVzbkiMiywEhU4bMw8uI6zxQRQZy05zDq4jsnN+5X7EMEVK1Z4PJ4/fz6hoaFs376dbt26kZWVxdy5c1myZAm9evUCYN68eTRu3JjNmzfzj3/8o9AxFUXh/fff58UXX+T2228H4LPPPiMsLIzly5czZMiQcr0GIYQQQpQuz5HH1nOb+ebUUmwum7tcq9IxIOI2eoT2xlvrU4UtFKJkBfYCYt6PASB3ci7eeu8qblHZSc9VJcpxWPnh1J4S6/xw6i9yHBX/rVJWVhYAgYGBAGzfvh273U6fPn3cdRo1akR0dDSbNm0q8hjHjx8nNTXVYx8/Pz86duxY7D5CCCGEqDgOl4Od5//k8xOfeQRWAA7FznfJy/jt7FrpwRKigkhwVYkcLqd7KGBxrC4HDpezQtvhcrmYMGECnTt3plmzZgCkpqai1+vx9/f3qBsWFkZqamqRx7k4HyssLKzM+wghhBCi4uQ58/jm1Bcl1vkx5XsszoJKapEQNxYJriqRVq1Bry55JKZBrUWr1lRoO8aNG8fevXtZunRphZ5HCCGEEJXrrCWNPGdeiXVsLhvH8o5WUouEuLFIcFWJfLUGbq3dvMQ6t9Zuga/WUGFtGD9+PD/88ANr166ldu3a7vLw8HBsNhuZmZke9dPS0ggPDy/yWBd7rC7PKFjSPkIIIYSoODmOsiXGyrJnVmxDhLhBSXBViUxaPeMa9SDE6Fvk9hCjL4816lEh610pisL48eP59ttvWbNmDTExMR7b27Zti06nY/Xq1e6yQ4cOceLECTp16lTkMevUqUN4eLjHPtnZ2WzZsqXYfYQQQghRcQL1gWWqF2IIreCWCHFjkuCqkoUZzSzt9hCDo9u4hwga1FoGR7dhabeHCCsm8LpW48aNY9GiRSxZsgRfX19SU1NJTU2loODCmGs/Pz/GjBnDxIkTWbt2Ldu3b2fUqFF06tTJI1Ngo0aN+PbbbwFQqVQ8+eSTTJs2je+//549e/YwfPhwIiMjGTRoUIVchxBCCCGKF6APxF8XUGIdL40Xtb2iK6lFQtxYJBV7JVOpVISbzExu3o/xjXvicDnRqjX4ag0V0mN10ezZswHo0aOHR/m8efMYOXIkAO+99x5qtZrBgwdjtVqJj49n1qxZHvUPHTrkzjQI8Oyzz5Kfn8/YsWPJzMykS5curFixAqPRWGHXIoQQQoii+Wh9GV53NP9JeBcFpcg6Q6KH4aXxquSWCVF2WrWWx9o95v69JqlZrb2OmLT6Cg2mLqcoRb/BXspoNDJz5kxmzpxZ6nFcrgsLuqlUKl555RVeeeWV8mmoEEIIIa6aWqWmnk99nqj/NIuOzyfDlu7e5q8L4N6o+2lsbopOravCVgpRMoPWwMwBxX8erc4kuBJCCCGEuI4YNSYam5syufEUsu1ZZNozMWvN+OkD8NH6oFFVbFZiIW5kElwJIYQQQlxn1Co1Zp0fZp0ftZH5VaJmURSF9PwLva7BXsGoVKoqblHZSXAlhBBCCCGEqDby7fmEvn0ho2Xu5Fy89d5V3KKyk2yBQgghhBBCCFEOJLgSQgghhBBCiHIgwZUQQgghhBBClAMJroQQQgghhBCiHEhwJYQQQgghhBDlQIIrIYQQQgghhCgHkoq9ClicOdiceRzKXkOWPQU/XQQNzb3Qa7wxanyrunlCCCGEqOEsTgt2lw21So231qeqmyPEFdGqtYxoOcL9e00iPVeVLN+RxYa0Ocw79gB/pH/Kvqyf+SP9U+Yde4ANaXMocGRVyHmnT59O+/bt8fX1JTQ0lEGDBnHo0CH39nPnzvH444/TsGFDTCYT0dHRPPHEE2RlldyeUaNGoVKpPH769etXIdcghBBCiJLl2nM5npfEgqRPeffw28w88h+2ndtKlr1iPl8IUREMWgPzB81n/qD5GLSGqm7OFalZoWANZ3HmsPHMR+zPXlnk9v3ZK0Glomvow+Xeg7V+/XrGjRtH+/btcTgcvPDCC/Tt25f9+/fj7e1NcnIyycnJvP322zRp0oTjx4/zyCOPkJyczNdff13isfv168e8efPcjw2GmvVHIIQQQtRU+Y58bC77/x4prEpbwcq0FR51DuceIlgfzLMNJxNkCKr8RgpxA5HgqhLZnHnFBlYX7c9aScegB8o9uFqxwvONdv78+YSGhrJ9+3a6detGs2bN+Oabb9zb4+LieO2113jggQdwOBxotcXfKgaDgfDw8HJtrxBCCCGKl+/IJ816huWnvmNf9n5cKDT0bUDv0J7cHx3K5ycW4cLlrp9uS+f9hHd4puHzmHXmKmy5EKVTFIV8ez4AXjovVCpVFbeo7GRYYCU6lL2mDLUUDmWvrfC2XBzuFxgYWGIds9lcYmAFsG7dOkJDQ2nYsCGPPvooGRkZ5dpWIYQQ5c/mdHDemk+2zVLVTRFXqMBZwMaMTUzd9yq7sv7CrjhwKk72Zx/gP0dmkVKQxpCo+wvtl2xJ5pztXBW0WIgrk2/Px2e6Dz7TfdxBVk0hwVUlyrKnlKledhnrXS2Xy8WECRPo3LkzzZo1K7JOeno6r776KmPHji3xWPHx8Xz22WesXr2aN954g/Xr19O/f3+cTmdFNF0IIcQ1yrNbOZFznjd3rWPMui95fOO3rD6VQIYlr6qbJsrovC2TRceXFLt9ZdqvBOgDCTGEFNq27dzWimyaEDc8GRZYifx0EWWqZy5jvas1btw49u7dy++//17k9uzsbAYMGECTJk2YOnVqiccaMmQIavWFGL158+a0aNGCuLg41q1bR+/evcu76UIIIa5Bnt3GmtNHeGrT97gUxV2+ISWRJgFhzOtxLyEmySxXnVmdVn5OWVFqvTVn1tEtuAffnP7Ko9yh2IvZQwhRHqTnqhI1NPcqQy0VDc09K6wN48eP54cffmDt2rXUrl270PacnBz69euHr68v3377LTqd7oqOHxsbS3BwMEeOHCmvJgshhCgn56z5hQKri/afT+Ot3evIs9uqoGWirCxOCwdzDpda73DOESJNhf/PNzU3rYhmCSH+R4KrSqTXeNPEHF9inSZ+8eg13uV+bkVRGD9+PN9++y1r1qwhJiamUJ3s7Gz69u2LXq/n+++/x2g0XvF5Tp06RUZGBhERFdv7JoQQ4srYnQ4WJewoMrC66PukfeTarZXYqurN6rRx3pZLli0Pl+IqfYdKoi7D5H6NSoOC52vtrfGmjnfh//9CiPIjwVUlMmp86Rw6liZ+/YDL3xhVNPHrR5eQhypkIeFx48axaNEilixZgq+vL6mpqaSmplJQUAD8HVjl5eUxd+5csrOz3XUunT/VqFEjvv32WwByc3OZNGkSmzdvJikpidWrV3P77bdTr1494uNLDiKFEEJUrnyHnV3pp0usY3M5yXNIz1WevYCTeWeYmfA9z+6aw+S/PuG/pzeRYc2u8iDLS+tFW//WpdZr7teUo7l/jyLRqXQ8Uf8pfDQy7FOIiiRzriqZl9aPrqEP0zHoAQ5lryXLnoKfLoKG5p7oNd4VElgBzJ49G4AePXp4lM+bN4+RI0eyY8cOtmzZAkC9evU86iQmJlK3bl0ADh065M40qNFo+Ouvv/jss8/IzMwkMjKSvn378uqrr8paV0IIUc1o1WrM+tJHJOjVmkpoTfWV6yhgVep2Pki48EVigM6HPuFtcChO1qTtpFtIc0KNAVWWGlqn1tErrCcr0lbhVIpOHqVCRe+wnsw5+iFalZYOgR25NeJ2AvQBaG7w11eIiibBVRUwanwxanxpHzy00s6plDAMBC4EXaXVufQ4LpcLk8nEihUr3AkthBBCVF/eOgMjG7Zj9emEYus08g/BpL2yubbXm5SCDD5I+Ba1Ss2j9W4j1juCn1O2sjvzGN5aI15aIx2DGhFs8KuyNpp1Zp6q/wTvJXxQKMBSoeLh2AcJNYQypcnLqFQqjGojBs2VD/UXoqpo1BruanKX+/eapFp8Kp45cyZ169bFaDTSsWNHtm4tPk3osmXLaNeuHf7+/nh7e9OqVSsWLlzoUWfkyJGoVCqPn379+lX0ZQghhBDVWiP/UNqGFE5yAKBRqXi1fX+CjOU/77emyHMUsCBxFQCTGt3D6fx0nt41h1/TdnA45xQ7zx/h7YNf8tSOWZy1ZFVZO/VqPQ3NDXirxXQGRtxKlKk2tU21iA+7mTdbTKd1QCv89f746wPw0/lLYCVqHKPWyFd3f8VXd3+FUVuz7t8q77n64osvmDhxInPmzKFjx47MmDGD+Ph4Dh06RGhoaKH6gYGB/POf/6RRo0bo9Xp++OEHRo0aRWhoqMc8n379+jFv3jz3YxmmJoQQ4kYXZPRmdtfBfLh3I18d3U2B80Ja7iYBYbzavh+N/Auvi3QjsTjt7M48ShNzHVwoLD+9sch6pwrSeWXfQqY1H4mfvmrmMOnVeoIMQdxe6zZuDr+w9IlJY0KnvrF7HoWoalUeXL377rs89NBDjBo1CoA5c+bw448/8umnn/L8888Xqn/5nKEnn3ySBQsW8Pvvv3sEVwaDgfDw8DK1wWq1YrX+nR0pOzsbALvdjt3uuR6E3W5HURRcLhcuV/XJHFTZLg4PvPhciKrjcrlQFAW73Y5GU726zi/+/Vz+dyTEpeQ+qVx+Gj3PNO3C2IbtKXDY0ak1GDRaAg1eoFTv16Gi7xWHw4FB0TAw/B98e/x39Erx76mHM0+QacnBS1X1X96aMF34xQl2Z/V9/SqLvKeIsriS++RK7iWVUpaJNhXEZrPh5eXF119/zaBBg9zlI0aMIDMzk++++67E/RVFYc2aNQwcOJDly5dz8803AxeGBS5fvhy9Xk9AQAC9evVi2rRpBAUFFXmcqVOn8vLLLxcqX7JkCV5eXh5lWq2W8PBwoqKi0Ov1V3jFQpQ/m83GyZMnSU1NxeFwVHVzhBBCCCGuicVpYcieIQAsbb4UYxUPbc3Pz+e+++4jKysLs9lcYt0q7blKT0/H6XQSFhbmUR4WFsbBgweL3S8rK4tatWphtVrRaDTMmjXLHVjBhSGBd955JzExMRw9epQXXniB/v37s2nTpiK/2Z88eTITJ050P87OziYqKoq+ffsWegItFgsnT57Ex8fnqtaBul4oikJOTg6+vr5VljFJXGCxWDCZTHTr1q3a3ZN2u51Vq1Zx8803X/GC1OLGIfeJKKvKuFdOF5wly5bPpN0flVr3qYZ30Tus9LToonLJe0rNl2fLgz0Xfo+Pj8dbX/5zQa/kPrk4qq0sqnxY4NXw9fVl165d5Obmsnr1aiZOnEhsbKx7yOCQIUPcdZs3b06LFi2Ii4tj3bp19O7du9DxDAZDkXOydDpdoSfb6XSiUqlQq9U3dJa8i0MBLz4Xouqo1WpUKlWR92t1UZ3bJqoPuU9EWVXkvRKmDkZRZxBjjuRQzskS67YIipN7thqT95SaS6f8/bpV9OtYluNfyfmr9FNxcHAwGo2GtLQ0j/K0tLQS50up1Wrq1atHq1atePrpp7nrrruYPn16sfVjY2MJDg7myJEjxdYRQgghhDBp9ESaghkdW3KW4bYBDfCRLHxCiMtUaXCl1+tp27Ytq1evdpe5XC5Wr15Np06dynwcl8vlkZDicqdOnSIjI4OIiIhraq8QQgghqq8sWz6Ztjys15jUwaDR0dhch4fibilye5xPJJObDMVcAUOVhBA1W5UPC5w4cSIjRoygXbt2dOjQgRkzZpCXl+fOHjh8+HBq1arl7pmaPn067dq1Iy4uDqvVyk8//cTChQuZPXs2ALm5ubz88ssMHjyY8PBwjh49yqRJk6hXr55HNkEhhBBCXB8yrLlsTU/g25PbsDjttAiI5t46NxGo98Zbd3W9S746EwNr3USP0JYsP/UHR3JP460xcmdUV2K8w/GvohTsQojqrcqDq3vvvZezZ88yZcoUUlNTadWqFStWrHAnuThx4oTHnJ68vDwee+wxTp06hclkolGjRixatIh7770XAI1Gw19//cWCBQvIzMwkMjKSvn378uqrr8paV1egR48etGrVihkzZpR5H5VKxbfffsugQYNISkoiJiaGnTt30qpVqwprpxBCiBtbSsF5xm7+mFRLprvsQPZpvjy+mX82u4M+4c2uOsDy0Zrw0Zp4KO4WrE47GpUak/b6/yzhcDnId+ZyxnKa8/Z0zNoAIkzRmDTeso6WEKWoFpkIxo8fz/Hjx7FarWzZsoWOHTu6t61bt4758+e7H0+bNo2EhAQKCgo4d+4cf/zxhzuwAjCZTKxcuZIzZ85gs9lISkrio48+KpSR8EYzcuRIVCoVjzzySKFt48aNQ6VSMXLkSHfZsmXLePXVV6/6fFFRUaSkpNCsWbOrPkZxpk6dikqlKvTz66+/ArBv3z4GDx5M3bp1UalUZQ4QP/74Y1q2bImPjw/+/v60bt26xLl8QghRU+TYrGRZC3C4nFXdlHJ13prL09sXegRWAN5aA3dFX/gscc6Wh911bctU6NRafHSmGyKwynfksTNzI28cfIaZR19lyYnZzDn2OtMPPMWWjDXkO3KruoniBqBRa7il/i3cUv8WNOrqtYZnaaq850pUnqioKJYuXcp7772HyXRhwUGLxcKSJUuIjo72qBsYGHhN59JoNGVexLk4Nput2LXEmjZt6g6mLrrY5vz8fGJjY7n77rt56qmnynSuTz/9lAkTJvDBBx/QvXt3rFYrf/31F3v37r2mayhJSdcnhBDlIb0gj01px/nq6F/YnA56RMYxKLYZQQYvdNVs0fGrkWHLJSEn1aPsjqgO3FqrDT+c3sHCpN9YnLSBHmFNuDOqI4EGH/Rq+ehTHIfLwd6sP/n85OxC2yyuAr45PQ8XLv4R2Au95voPNEXVMWqN/Hjfj1XdjKtSLXqurht5ecX/WCxlr1tQUHrdq9CmTRuioqJYtmyZu2zZsmVER0fTurXnOh09evRgwoQJ7scpKSkMGDAAk8lETEwMS5YsoUWLFrz//vtFnispKQmVSsWuXbvcZevXr6dDhw4YDAYiIiJ4/vnnPRa97dGjB+PHj2fChAkEBweXOEfu4mLOl/5cDFTat2/PW2+9xZAhQ8o8FPT777/nnnvuYcyYMdSrV4+mTZsydOhQXnvtNY96n376KU2bNnVfw/jx493bTpw4we23346Pjw9ms5l77rnHIxPm1KlTadWqFZ988gkxMTHuNakyMzN58MEHCQkJwWw206tXL3bv3l2mdgshRHGS87K5c+VnPPH7d2xISWTLmZO8sWsdvb//P/afT8PurPm9WLvPH/d4PLB2W1oF1OHhrR+x/NRWTuSlk5R3lvnH1nP3hnc5kHUau1MWWy9OvjOX5cmflVjnh+TPsbjyK6lFQtQ8ElyVJx+f4n8GD/asGxpafN3+/T3r1q1buM5VGj16NPPmzXM//vTTT93JQ0oyfPhwkpOTWbduHd988w2ffPIJ6enpZT7v6dOnueWWW2jfvj27d+9m9uzZzJ07l2nTpnnUW7BgAXq9no0bNzJnzpyyX9g1Cg8PZ/PmzRw/frzYOrNnz2bcuHGMHTuWPXv28P3331OvXj3gQsbK22+/nXPnzrF+/XpWrVrFsWPHPIasAhw5coRvvvmGZcuWuQPPu+++mzNnzvDzzz+zfft22rRpQ+/evTl37lyFXa8Q4vp23lrAxI3fczI3s9C2fIedYauXkmkrKLxjDaNT/d37plNruDOqA6/u/Qan4ipU1+py8OSf88hy1PzrriiplpMUOEv+Ateu2DiWd6iSWiREzSN94zeYBx54gMmTJ7uDiI0bN7J06VLWrVtX7D4HDx7k119/Zdu2bbRr1w6Ajz76iIYNG5b5vLNmzSIqKooPP/wQlUpFo0aNSE5O5rnnnmPKlCnupCX169fnzTffLPV4e/bsweeSILNJkyZs3bq1zO253EsvvcSdd95J3bp1adCgAZ06deKWW27hrrvucrdt2rRpPP300zz55JPu/dq3bw/A6tWr2bNnD4mJiURFRQHw2Wef0bRpU7Zt2+auZ7PZ+OyzzwgJCQHg999/Z+vWrZw5c8bdy/b222+zfPlyvv76a8aOHXvV1ySEuHHl2q1sOVP8Arg5divbzpzkljqNK7FV5a9dUKz79x5hTVmXtr/IwOqifKeNLekJDKjVpjKaV+Ocs50tU70M65kKbom40eXZ8gh9OxSAM8+cwbsGLXsgwVV5yi1hkuflY9vPlPDGpL6sQzEp6aqbdLmQkBAGDBjA/PnzURSFAQMGEBwcXOI+hw4dQqvV0qbN3/+M6tWrh7+/f5nPe+DAATp16oRKpXKXde7cmdzcXE6dOuWe89W2bdsyHa9hw4Z8//337sfXmgkyIiKCTZs2sXfvXn777Tf++OMPRowYwSeffMKKFStIT08nOTmZ3r17F7n/gQMHiIqKcgdWcCHg8/f358CBA+7gqk6dOu7ACmD37t3k5uYSFBTkcbyCggKOHj16TdckhLhxncjJLLXO1rSaH1z5aI30CGvCurT9xPqEsuPcsVL32ZR+mL4RLdHVsEnylcGsCyhTPX/dtc3LFqIs8u01c/ipBFflyfsKouqKqlsGo0ePds8VmjlzZrke+1p5l/Fa9Xq9e0heeWrWrBnNmjXjscce45FHHqFr166sX7/e3WN3rS6/vtzcXCIiIorsObyS4FUIIS5l1pf+hVOQqWq/Cc605mNXnKhVKgL13h5fvpWVWe/F5KaDsLkcOFxOdGVIVqFXa1Fz5ee6EdQ2xaBXG7C5rMXW0ai01PNtUomtEqJmkeDqBtSvXz9sNhsqlapMCys3bNgQh8PBzp073T1LR44cITMzs8znbNy4Md988w2Korj/gW7cuBFfX19q1659VddR0Zo0ufDPIy8vD19fX+rWrcvq1avp2bNnobqNGzfm5MmTnDx50t17tX//fjIzM93HKUqbNm1ITU1Fq9VSt27dCrkOIcSNJ9zLlyCDFxnW4r/5vb1u1XxAPm/NZ2/maT46vIHjuecIMngzPO4fdAtvQJDhygO+QIMPL7e4mwKHjajzQWxKP1xi/dtrt0dz+QgRAYBJ40W/8Lv5PnlRsXV6hd6GUW2qxFYJUbNIcHUD0mg0HDhwwP17aRo1akSfPn0YO3Yss2fPRqfT8fTTT2Mymcr8TeNjjz3GjBkzePzxxxk/fjyHDh3ipZdeYuLEiR6LRJcHm83G/v373b+fPn2aXbt24ePjU2xv16OPPkpkZCS9evWidu3apKSkMG3aNEJCQujUqRNwIdvfI488QmhoKP379ycnJ4eNGzfy+OOP06dPH5o3b87999/PjBkzcDgcPPbYY3Tv3r3EXq8+ffrQqVMnBg0axJtvvkmDBg1ITk7mxx9/5I477ii3HjMhxI3FX2/i1Q7xPLbh2yK331e/NX76q1tY91qct+Yzdfd/WZV8wF2Wbs3lnzu/o4E5jE9uGkaw8cqTNvnrvfHXe9NJoyPM6EeaJavIenE+YUR5BRW5TYBOradjYA+cipOVqV/jUOzubRqVhl6hA+kecgsGjQRXQhRHvrq5QZnNZsxmc5nrf/bZZ4SFhdGtWzfuuOMOxowZg4+PT5nnOtWqVYuffvqJrVu30rJlSx555BHGjBnDiy++eLWXUKzk5GRat25N69atSUlJ4e2336Z169Y8+OCDxe7Tp08fNm/ezN13302DBg0YPHgwRqOR1atXu+dDjRgxghkzZjBr1iyaNm3KrbfeSkJCAgAqlYrvvvuOgIAAunXrRp8+fYiNjeWLL74osa0qlYqffvqJbt26MWrUKBo0aMCQIUM4fvz4Db/wtRDi6uk0GrpE1GVBr3up5/d3MBFs9OKfbXrxTKvu+Bkq/wPyhrQEj8DqUoez0/jw4FoKHLarPn6gwYf/6zCWut4hhbY1Ntfig3ajCDRcfcbdG4GX1oeuwfG82OQDHogeT//wuxka9Sj/avwfeocOxFvrW9VNFKJaUymKolR1I6qb7Oxs/Pz8yMrKKhSAWCwWEhMTPdYpuhGdOHGCOnXq8Msvv3DzzTdXdXNuaNX5nrTb7fz000/ccsst6HS6qm6OqKbkPqlY6QV5WJwOXIoLk1ZHgMELbRUMi8uw5DLi9/kcyy1+GQ+DWsvKm58k1FT0B/iy3ivnrLmkWjLZdPYwapWKrqGNCdL7ECCB1Q1B3lNqvjxbHj7TL/y95k7OrZBsgVdyn5QUG1xOhgWKMlmzZg25ubk0b96clJQUJk2aRHR0NN26davqpgkhhChBcBUnrrjIhVJiYAUX1qKyOK++5+qiQIMPgQYfmvhVzzm9VyPHnovVZcPhsqPX6DGqDXhpvaq6WUJUCLVKTfc63d2/1yQSXIkysdvtvPDCCxw7dgxfX186derErFmz5BshIYQQZaICTBodBU57ifW0kiLdg8VpJbkghflJS0nIvZBqXoWKVv7NGF7nXkIMQWjkORPXGZPOxLqR66q6GVdFgitRJvHx8R6ZBV0uF9nZ2VXYIiGEENXBpVlgS+KjNTAwqiVfJP1ZbJ0mfhEYNfKl3UVOxcnR3CReP/AeLv5eHFlBYWfmHg7mJDCt2QtEmsKrsJVCiEvVrH42IYQQQlS5AoedtIJsFh/9kxe2/5cP96/nVF4mWTZLsfsowH2xHfDVFT03VI2KZ5v1RS+9MG459lxmHvnEI7C6VIHTwsfHFpJjz63klgkhiiM9V0IIIYQoszy7lfWpR3h263Lsyt8f+j/Y/xuj6/+DRxp1wb+ITIR5Dhtv7/2FmR2H8sbelezLTHZvi/IKYHKLfiTmnmX7uSTuiGqDWW/EW1u2jLTXq1TLGc7bi04rf9HBnASsLiu+SLIOcf3Is+VR9/26ACQ9mVQhCS0qigRXQgghhCizU/lZPLVlGUWlGv40YTNx5mAG122F+rKhggoKG84cIaUgi5H1biLGJxiry45BreWcLY8FR//gz4wkAP7v0HpeajWQvhFN8Cmmp+tGcLogpUz1Mm3ZBBtk/S5xfUnPLzkBTnUlwZUQQgghyiTXbuU/+9cVGVhd9OH+3+gZUb/QYsBqlZr65lASss/w4s7vGBHXCYNGw5LEzYWSXLhQeGnXdzQ2h9PYP/KK2njOmovN5UBRFPQaLQF67xqXbewiH23Zvq03am7sHj4hqhMJroQQQpSbXKuVArsDAG+9Hi+9JCe4nhQ47fyRllRinZSCbCxOR6HyIIM3jzToxtN/fg1AfK0mjN20oMTsgTMPrmV62zvx1ZW+4HGOvYDd548z8/AvHMlJBSDC5M/I2B70Cm+Kfw0aVnRRfd84NCoNTsVZbJ1AvT8+WhkSKER1UTO/yhEVrkePHkyYMOGK9lGpVCxfvhyApKQkVCoVu3btKve2CSGqH6vDQWLGOSZ/9wu93p9L/IfzeG3FWpKzsnG6ip6ML2ogBTRlyAxYXI1/hMYyMKoFRo2OPIe11LTsWzISsRYRqF0uz2Hhh1M7eGr7Z+7ACiClIJPp+5bzwcEVZNnySz1OdeOlMdIvvFeJdYZGD8ask+BKiOpCgqsbxMiRI1GpVDzyyCOFto0bNw6VSsXIkSPdZcuWLePVV1+96vNFRUWRkpJCs2bNrvoYpTl16hR6vb5CzyGEKJvEjPMMnLOIXw4eweZ0kmez8/Wufdzx0WJSsyWT2fXCW6enT2TDEuvE+QYXm049QO/Fc8368XGnB8qUcl1TxuF8uQ4r7x/6udjt/z29ndSCzDIdqzoxaozcHtmffuG9UF0WsupUWkbUHUJr/2Y1dtijENcj+Wu8gURFRbF06VIKCgrcZRaLhSVLlhAdHe1RNzAwEF9f36s+l0ajITw8HK326kee2my2ErfPnz+fe+65h+zsbLZs2XLV5ykPTqcTl3w7L25Q5/MKeOnH1dichYcuZRZY+M+6TeSX8vcsagYvrZ5HG3dBV0K69Gea9yLIWPwQvACDF22D6xDlFYC5lGQVPcIaYtLoS23XmpS9OJWS34MXJf2OpZSesurIV+fD3bUH8p/W0xlZdyi3RcbzSOxI3m/9Oj1COuNdxnlZQojKIcHVDaRNmzZERUWxbNkyd9myZcuIjo6mdevWHnUvHxaYkpLCgAEDMJlMxMTEsGTJElq0aMH7779f5LmKGha4fv16OnTogMFgICIigueffx6H4+/hHj169GD8+PFMmDCB4OBgj0WLL6coCvPmzWPYsGHcd999zJ07t1CdjRs30qNHD7y8vAgICCA+Pp7z588DFxZBfvPNN6lXrx4Gg4Ho6Ghee+01ANatW4dKpSIzM9N9rF27dqFSqUhKSgIuBHb+/v58//33NGnSBIPBwIkTJ9i2bRs333wzwcHB+Pn50b17d3bs2OHRrszMTB5++GHCwsIwGo00a9aMH374gby8PMxmM19//bVH/eXLl+Pt7U1OTk6xz4cQVcnicLDrVPFZzX7cd4g8W837UCuKFmoys6DrAwToPedBGdRaprbuT/vgOmU6ToDBizH1uha7XatS80jD7njrSk7WoCgKR3PTSj3f6fwMrM7KDfKz7bmkFqSzN/MIx3JPcd6WjdNV/Pyp4nhpvQgyBBIf3pP7ogfTPfQmAvT+kshCXLfUKjXtItvRLrJdjeuZlYQW5SjPllfsNo1ag1FrLFNdtUqN6ZLJu0XVvdp8/6NHj2bevHncf//9AHz66aeMGjWKdevWlbjf8OHDSU9PZ926deh0OiZOnEh6etlTZJ4+fZpbbrmFkSNH8tlnn3Hw4EEeeughjEYjU6dOdddbsGABjz76KBs3bizxeGvXriU/P58+ffpQq1YtbrrpJt577z28vS88L7t27aJ3796MHj2a999/H61Wy9q1a3H+75v1yZMn8/HHH/Pee+/RpUsXUlJSOHjwYJmvByA/P5833niDTz75hKCgIEJDQzl27BgjRozgP//5D4qi8M4773DLLbeQkJCAr68vLpeL/v37k5OTw6JFi4iLi2P//v1oNBq8vb0ZMmQI8+bN46677nKf5+Lja+lJFKIiKUpJuePAIb261xWjRkvLoFr80PcRDmedISH7LGFGX9oGR+GrM2DSlt7TBKBTa7mzThvSrTksOrYF5ZIchF4aPe+1H0KY0VzqcVQqFZGmgFLrBRvM6NWV87HH6XKSYknn3UOLOZST5C4P0PkyImYgnYKa46PzqpS2CFETmXQmtj20raqbcVUkuCpHPtOLn1B6S/1b+PG+H92PQ98OJd9e9OTa7nW6s27kOvfjuu/XLZTrX3mp5A8zxXnggQeYPHkyx48fBy707ixdurTE4OrgwYP8+uuvbNu2jXbt2gHw0Ucf0bBhyePuLzVr1iyioqL48MMPUalUNGrUiOTkZJ577jmmTJmCWn3hW4n69evz5ptvlnq8uXPnMmTIEDQaDc2aNSM2NpavvvrKPW/szTffpF27dsyaNcu9T9OmTQHIycnh/fff58MPP2TEiBEAxMXF0aVLlzJfD4DdbmfWrFm0bNnSXdarl+fE448++gh/f3/Wr1/Prbfeyq+//srWrVs5cOAADRo0ACA2NtZd/8EHH+Smm24iJSWFiIgIzpw5w08//cSvv/56RW0TojLptRpigwI4lnG+yO2dY6MxaIsfRiZqHp1aQ4jRhxCjD53DYkvfoRj+ei8ebdiDYbGdWJVygHRLDk39a9EmKBqzzoihDPOyAPpFtmJ2wqoS69wf0wVTJS1KfNaayVM73yHfafEoP2/PYcbhxVji7uLm8H9gLMOQRyFEzVKz+tnENQsJCWHAgAHMnz+fefPmMWDAAIKDg0vc59ChQ2i1Wtq0aeMuq1evHv7+/mU+74EDB+jUqROqS7JMde7cmdzcXE6dOuUua9u2banHyszMZNmyZTzwwAPusgceeMBjaODFnqvi2mK1WovdXlZ6vZ4WLVp4lKWlpfHQQw9Rv359/Pz8MJvN5ObmcuLECXe7ateu7Q6sLtehQweaNm3KggULAFi0aBF16tShW7du19RWISpSkLcXL8T3KHKbTq3muZu7YTZWj4Vg82w2Mi0W7GXIQHe9cJXSs1jVfHUmIrz8GR7XiYlN+xJfqykhRt8yB1YXjmHkgbrFf0H2j+D6RHsX/l/nUlyct+aSYc0hx15QxJ5XLt9hYUHSfwsFVpeal/gd+Y7yOZ8QonqRnqtylDu5+IxYmssm/5555kyxdS8fW5r0ZNI1tetyo0ePZvz48QDMnDmzXI99rS4O6yvJkiVLsFgsdOzY0V2mKAoul4vDhw/ToEEDTKbi10QpaRvg7kW7dKiT3V54vojJZPIIFgFGjBhBRkYG77//PnXq1MFgMNCpUyd3co7Szg0Xeq9mzpzJ888/z7x58xg1alSh8whRnahUKlrXjuCjoYN4beU6jp/LBKBJeCgvD+hNVIBf1TYQOJefz6GMdObu2EGWxUK7yFrc36IFwd7eGK8h8U51dd5SQLoljy+P7CbTZqFzeF1uiqhDoMGEtoRkFDWVr87EyLgeRJgCmHt0LedsF/4fe2sNDI7qyAOxXQm4bDj9OWsOv53dx/KTm8l2FFDXO5QRMb2o6x2K3zWsiVXgtLIxfVeJdawuO/uzE+kS0uqqzyPE9Szfnk+TmU0A2D9uP141aBjt9fcfpQpdyTyoiqpbFv369cNms6FSqUpMGnFRw4YNcTgc7Ny5092zdOTIEY+ED6Vp3Lgx33zzDYqiuAOFjRs34uvrS+3ata+o/XPnzuXpp5/2SB0P8Nhjj/Hpp5/y73//mxYtWrB69WpefvnlQvvXr18fk8nE6tWrefDBBwttDwkJAS4k8QgIuDCOv6zrdW3cuJFZs2Zxyy23AHDy5EmPuWktWrTg1KlT7iCwKA888ACTJk3igw8+YP/+/e6hi0JUZz5GA93q1aVpxL1YHHbUKhVGrZZA76r/h3iuoIDnV/3Cr8eOucu2Jyfz6Y7tzLvjTtrVikSvuX7+HZ6z5POvLb/w44m/55F+dXQPvjoDi28eQmP/UHSa6h1g5dqtWP6XfEKn1uCnL/0+8tN7cWd0B3pHNCPPYcWlKPhojfgWMbwwrSCTR/+cTZol8+8ySyZbMg5zS0RbxjcYcNUBlkNxlJq5EOCM5dxVHV+IG4GiKBzPOu7+vSa5fv6biDLTaDQcOHDA/XtpGjVqRJ8+fRg7diyzZ89Gp9Px9NNPF9lzU5zHHnuMGTNm8PjjjzN+/HgOHTrESy+9xMSJE909RWWxa9cuduzYweLFi2nUqJHHtqFDh/LKK68wbdo0Jk+eTPPmzXnsscd45JFH0Ov1rF27lrvvvpvg4GCee+45Jk2ahF6vp3Pnzpw9e5Z9+/YxZswY6tWrR1RUFFOnTuW1117j8OHDvPPOO2VqX/369Vm4cCHt2rUjOzubZ5991qO3qnv37nTr1o3Bgwfz7rvvUq9ePQ4ePIhKpaJfv34ABAQEcOedd/Lss8/St2/fKw4+hagqKpWKYJ+qD6YupSgKaxOPeQRW0X5+dK8bg16jZu727dQPCiLE+/r4d2h1OJh38E+PwOqiHLuVob98zprbxxLmVT0XnS1w2EguyGT2obWsTTuI0+WidVAdxjXsRUNzGMZSPrZo1RqCDL4EGYpPAJRpy+OlPUs8AqtL/ZSynQ5BDbg5otVVXYNWpUWr0uBQSs4KGGYMuqrjCyGqN5lzdYMym82YzaVnYbros88+IywsjG7dunHHHXcwZswYfHx8MBjKNjm4Vq1a/PTTT2zdupWWLVvyyCOPMGbMGF588cUravfcuXNp0qRJocAK4I477nAngGjQoAG//PILu3fvpkOHDnTq1InvvvvOve7Wv/71L55++mmmTJlC48aNuffeezlz5sJQTZ1Ox+eff87Bgwdp0aIFb7zxBtOmTStz+86fP0+bNm0YNmwYTzzxBKGhoR51vvnmG9q3b8/QoUNp0qQJkyZNcmcxvGjMmDHYbDZGjx59Rc+PEMLTuYICPv7zTwDMBgMzb72Vyd26k221cCo7m851oimw22vcN6PFybJbmH9we7Hb8xw2Vp48VO7zsPLsVs5b88l3XH2qc5vTwc5zJ7hr/Sx+SdmH3eXEhcL2jCRG//EpXx/fTq7des1tzXEUsOd/34gXZ37ias5Zy774tdPlxOa6MI/PS2Oga0jrEusb1Hoa+8WU+fhCiJrj+viqTpRq/vz5JW5fvny5x+PLswdGRETw008/uR+fOHGCs2fPUq9ePXfZpR9O6tatW+jDSvfu3dm6dWuxbSgtHTzAf/7zn2K3hYeHewQp3bt3Lzalu1qt5p///Cf//Oc/i9zeuXNn/vrrL4+yS69n5MiRhYYlArRu3Zpt2zxTh16aVh0uLND86aefFnsdcCF1fVBQELfffnuJ9YQQJXMqCsk5Oeg1GubcNpD/bNnMplMn3dtXHj3CnD+38cXd9xITUHo67+ou324np5QAZN3pYwyObY637toz1Z235pOUk8EnhzeRWpBNbS9/xjTsRLR3AP6GK+vFzLYX8PT2L4odUvfegV/oGeI5nDrHXkCew8q6Mwc4Z82lsV8tmvtHYdaaih3qeSwntdS2JOWdwVlKzxNApi2HUwVnWZGyGavLRkv/+nQKasaIugPZdm4fucUkrXgo9g68NdUjyYsQonxJcCXKZM2aNeTm5tK8eXNSUlKYNGkS0dHRksWunOXn55OSksK///1vHn74YfR6SdMrxLXQqFRE+/nRIDiY344neQRWF6Xn5/PQ98tZeve9BHtV7LBGi8NBts1CWt6FXpEwbx/MegNGbdkz45VEoy59qLZRo0VTDklyzlvzeXnnz/x0ar+7bM/5FH4+fYDBdVoyqUUfAq4gwErKzWBKi4FEmPxwKgoqYGXyXpaf3EnB/+ZffXPiTy5+pZdly+ejhDV8cWKzxxpZPloDb7a+j5b+dYp8XnVlWOtKharUYe/p1kxe+GsOx/P/DtZ+O7uLOUe+ZUrT0fynzXO8e2gRe7KOuLcHG/wZEzOINoGNMEgadiGuSxJciTKx2+288MILHDt2DF9fXzp16sSsWbPQ6crnA4G44M033+S1116jW7duTJ48uaqbI0SNF+TlxWMdOuKj1zNx5c/F1jt2/jzp+XkVGlydtxSw5MBuZu3aTN7/MpB6aXWMbdmeEU3bEGAsPZtoaUwaHbHmQI5lF06WoFWpqe3jx5jG7YsMOvLsNqxOBwaNttReLZeisOLUfo/A6lLfHN/NTWGx3BbdrEztzrVbyHVY+b/D6zmac/Z/16LntqgWzPnHcCb+uZQMay5Hs89SD39sTjtLT25i6YlNhY/lsPLEnwtY2uVxYnxCC22v7xtR6pyoDkH1MaiL//+WacvhX3s+8gisLrIrDl7a+wkftX+eF5uMId9pJdOWg1FjwFfnhVnng0YlszKEuF7JX7cok/j4ePbu3Ut+fj5paWksW7aM6Ojoqm7WdWfq1KnY7XZWr16Nj0/1nHAuRE3TKSqKIC8vzhWUvK7QyaysCmtDgd3Oon27eGvbBndgBZDvsDNj+x98umc7+UUs+XClgoxeTG3fx6PMpNXxVMsufN3vAZ5s0Zksu4XU/Gz3/KXz1gJ2ZSTz7Jb/Mnr9Fzy/5Qf+ykgh01r883XOmsdHh/4osS2zDmwgw5JX7PYMSy7plhzOFuRwJPsMT21b6g6sAAqcNr5M+pO39q7gtVZ3AhcWHQbIcVpZmPh7scd2KC7mJKwmz1F4iKS31sjAWh2K3VeNikfq9cdXV3ywe8Z6nmN5ycVud+Hi8+Mr0ai0hBoDaWCuQ7R3OAF6swRWQpSBSqWiSUgTmoQ0qXHL0UjPlRBCiOtagMmE1eHAR68n11Z8woXwCvxCI8duZdbuLcVu/+ivrQxr0gqvaxwNoFKpaB1ci096DObFrb+QbbPwSc/BfHN0D3et/Azn/+aOalVq7oprzsSW3ZixZwOfH93pPsZf51L48eRBhsS1YlLLnvgbCgcZDpeL0/klB6NHc9KLnD91zprH72eOMDdhIydyzxFk8Ob26JbM7zyaF3d+S1Jehkf9PZmnSS3Iopl/Le6t24Gk5F2k5me6hwoWZ13aASY1seGt9Uy85KU1MCauD3kOCytTd3psM6h1TG0+lNpeJWfyW5NWfNKQizam7+HB2NsxacuW+EkI8TcvnRf7HttX1c24KhJcCSGEuO4FmEwMadacT3YU/aE40teXCJ/i03dfqxPZmRQ4iu+ZsjqdJGRmEOp97QGer95Az1px/PeWkVgcdt7evZ7vkzyH7zkUF0uP7MbqdNAooPDQOYClR3fRLSKWflGFs7OqVSrUqHBRfNZBnUpd6Bvnc9Y8Jv35DX+c/Ts1fnJBFrMP/caPp/byZrs7GbdlMZm2fI/9vj+5m2ExnYj08icJsJch2YRTcVFcUkR/vQ8TGg5kTNzNrErdxXlbLo3NUbQLqoeP1oRRU3KQa3OV3stYWip2IcT1Sfqmr5LLVfoCgUJUBrkXhSidQatlbNt2NP7fIuGX8tbp+L/bbiewAudbOcuQ+rw806Nr1GpCTN6oVBQKrC61PHEfHUKjMBSTWe/DfRuLHNpn0ujoGVG/xDb0j2qK1yVBiqIorE456BFYXepE3jkWHt3M4Oi2hbalW3NpFRjtHhYYaQpATclDhZr610ZbwjqKZr0XtbyCGBnbm6ca3U6/yDYEG8ylBlYA7QOblFqnkbkOWnX1XqxZCFH+pOfqCun1etRqNcnJyYSEhKDX62vcWNDy4HK5sNlsWCyWK1oEWJQfRVGw2WycPXsWtVotmQWFKEWwtzef3TGYzadOsvivvyhw2OlZN4Z7mjYjyMsLdQW+l9c1B6BXa7C5iu7N0KrU1A8o/0VlN6aWvJ6TAmw/e5pG/qHszig8h+hg5pkigz5fvZFnm/dmQ9rRIq/JpNHxZJPueOv+HhJ3zpbHJ4eLnycF8EvyARZ0GcncIxs8yqO8AvDR/Z263Eujp1toI9adOVDssR6u1xt/vXeJ57tajczRmLXeZDuKn1M2rE5/zLqKOb8Q17t8ez7tP24PwLaHtuGlq14L1JdEgqsrpFariYmJISUlheTk4iezXu8URaGgoACTyXRDBpfViZeXF9HR0RLkClEGQV5eDGjQkC7RdXApCj56PTpNxfcu+Oj13N+4JfP27Shy+z0Nm+FTDutOXc5RTDB3eZ3iUrP7lDBfqJa3P5/3GMmkbd9xNCfdXd7AHMrbHQYRbvJcqN6lKJzMP19iW+wuJ64i5mmNadAVX50R+/+SfvjojExuejun8s9xJDetUP0RsV1p5l+rxHNdC7PWm9daPMKzu/6DxVV47tfgWj2I86m48wtxvVMUhf1n97t/r0kkuLoKer2e6OhoHA6Hx6K1NxK73c5vv/1Gt27dJB17FdJoNGi1WglwhbhCfsbKXcDVW6fn8TadsLmcLD34l3uYoFql4q4GzXimfVd89OWf+KBjWOlZXVuH1GL2gcIpzQHujmuJWV/4ucq1Wylw2In08mNpj5FkWPPIsOYRbPTBrDMSZCzcY6NChY/WQG4RGfwupVV5Brv9azUn2jsQu8vhUR5s9GV2h9H8lXmCpUmbyHYUEOcTxvCYroQZzZgr8JtujVpDjHckH7efzPenN7Du7A5sLjv1faK4v048UV5h+Nagb9qFEOVHgqurpFKp0Ol0N2xgodFocDgcGI3GG/Y5EEKIKxFo8uL5jt0Z37oTe9NTUYBmwWH46PSYDRUT7AUavGgTXIsd6aeL3N4hNIrTedlk2y3uMoNGy63RjelbuyFNA8KwOOzuOVnZtgJO5WUx+8Dv7MtMxVdnYEhsG26u1ZBYc3CJbfHVGRhcpzULjm4utk6rwCgOZV9YO6qOdxDD4zoR7R3IhG2L6RzSgNtrtbrs+nzoEdaENoExOBUnRrUek7Zyhkjr1BpCjQGMiLmFO2v3ABXoVVp8JKgS4oYmwZUQQghRSXz1Bnz1hgrNTHipQKMXs7rdyUPrvmLPOc8Fb1sGRfBBl0HYFQddw2P4PTWRlkGRvNw2nh9PHmDq9pXkOWw0CwjniWZdiTUHsSxxF2/tXetxnCk7fuaTQ5tZ3GMY4V6eQwEvZdDoGFnvJv57cg/nbIXnKulUGl5scQthJl86hsRwMCuFr45vY2vGhQQY+7KS+eLYJp6mOQ6XEx1/f7FnLmFNqoqmU2sJNBR/3UKIG4sEV0IIIcR1LMzLh3m97iElP4dfTh4GID6qIeFevgQZL/SyfNB5EDangzyHnXt+XUiG9e/gZ9OZ42xac5zxTbvgc9lIhfrmEMJMvuTarUzdsYI3OwwschjhRSFGH5Z2f5CXd/3AH2ePofwvlXtjvwheaXUbsT7B5DgKuGv9h1gvGwYIYHFemHOVZS/AVEG9fUIIcS0kuBJCCCGuc0FGb4KM3jQLDC9yu5/eRKa1gBe2fecRWF3qw32/s7jnfQToTbQMjGRso5s4lZfF8dxzBBq8aBVUi3yHjSx7AQa1Fj+9EcNlac3VKjW1vQN4p/1d5Dtt5NiteGn1mDQ6Ag3eKIrC8lP7iwysLvVX5inCfQKu7skQQogKJMGVEEIIIch32Nhy5kSJdZYl7eG5Fr3x1Rt57I8vybpkrpZOreHZZr3RaTS8tedX7qrbiocbdSHEWHhhZLPehBkT4ZeN5rO6HGxNTyy1rdvSj9G3dvOyXZgQosZRqVTU8avj/r0mqRa5m2fOnEndunUxGo107NiRrVu3Flt32bJltGvXDn9/f7y9vWnVqhULFy70qKMoClOmTCEiIgKTyUSfPn1ISEio6MsQQgghaqzztoJS65zIzaRlYC2e2vyNR2AFF9Kov/7XLxjUWloE1uKzo9sYtWFxkYsQF0ejUmMqwyK+XhpZ10+I65mXzoukCUkkTUiqUWtcQTUIrr744gsmTpzISy+9xI4dO2jZsiXx8fGcOXOmyPqBgYH885//ZNOmTfz111+MGjWKUaNGsXLlSnedN998kw8++IA5c+awZcsWvL29iY+Px2KxFHlMIYQQ4kbnqys9FXyo0Ye951OxF7EW1UWfHP6D+2LboVWpUVD4M/1Emdep0ak1DIxq7VGmUanpEdaI0XHdGFK3IwA3RzYr0/GEEKKyVXlw9e677/LQQw8xatQomjRpwpw5c/Dy8uLTTz8tsn6PHj244447aNy4MXFxcTz55JO0aNGC33+/sOq7oijMmDGDF198kdtvv50WLVrw2WefkZyczPLlyyvxyoQQQoiaw6TR0cg/tMQ6Ixq0Y+7hP0qsczovi/rmYJb2GMnweu05b8vneN45zlvzS22D1enAoNbR2C8SgPiIZnzW+SGa+EWSkJNKSkEWAL4lLG4shBBVqUrnXNlsNrZv387kyZPdZWq1mj59+rBpU9ELGl5KURTWrFnDoUOHeOONNwBITEwkNTWVPn36uOv5+fnRsWNHNm3axJAhQwodx2q1YrX+vahhdnY2cGGh3IurwQtPF58XeX5ESeQ+EWUh90n1oDhdPN+sB09sWo7VWTihRK/IeoTpvUkvyMFQzHezRo2WdzvcycLDm/nu5N+LJQN0CK7DK61vLXIO1kUOp4OfTuxicqP+bEtPoo5vEA9vnOdeQFivqOlCK5778wve6XAfoUZJgS4Kk/eUmq/AXkCvRb0AWPPAGkwVsNzCldwnV3IvqZSy9tVXgOTkZGrVqsUff/xBp06d3OWTJk1i/fr1bNmypcj9srKyqFWrFlarFY1Gw6xZsxg9ejQAf/zxB507dyY5OZmIiAj3Pvfccw8qlYovvvii0PGmTp3Kyy+/XKh8yZIleHnVrHGeQgghhBBC1GQWp4Uhey50iCxtvhSjpmqXXsjPz+e+++4jKysLs7nkL3VqZLZAX19fdu3aRW5uLqtXr2bixInExsbSo0ePqzre5MmTmThxovtxdnY2UVFR9O3bt9Qn8EZlt9tZtWoVN998Mzpd6ZOPxY1J7hNRFnKfVA8JWWe569fPiPQyc1dMC9oG18YF5Ngt7M5IJszkzaeHtzCt3QAe/eMLXHh+N6tXa/iky32M/P0zSvrW9qNOQ2nsF06u04rVaUev0WJU6/DTX/hm+pwtj8VHN6GoFL5I8vySVa+oeVZpxVuqXdhULhZ3eZi6PiHl/EyImk7eU2q+PFse7Lnwe3x8PN5673I/x5XcJxdHtZVFlQZXwcHBaDQa0tLSPMrT0tIIDy96LQ64MHSwXr16ALRq1YoDBw4wffp0evTo4d4vLS3No+cqLS2NVq1aFXk8g8GAwVB4/LZOp5M/ylLIcyTKQu4TURZyn1StYG8zOq2WxPxM3tr3m8e251v25MeUg5y0ZrP2zFGmtL2Fl3b+hM3ldNdp7B/BgZwzWCg+2QXA16d209GSySt//Yjzf4kx2gZF82KLWwjUe6PRaBnWoDNPbluCVVXEsRSwqVxYVS52Zp2kfkDktV+8uC7Je0rNpVP+ft0q+nUsy/Gv5PxVmtBCr9fTtm1bVq9e7S5zuVysXr3aY5hgaVwul3vOVExMDOHh4R7HzM7OZsuWLVd0TCGEEOJG4qc3MqZRxyK3GTQ68uw2AOYnbGXL2eMs7j6CSc37cG9MGyY268k7He7AVsriv3AhacXRnLPuwApge8YJhqz/hGO56Tzz51f8fHofzze7lQB9yUPz1SrNFVxh+cq05ZJmOcfp/LNkWLMocFhL30kIcd2r8mGBEydOZMSIEbRr144OHTowY8YM8vLyGDVqFADDhw+nVq1aTJ8+HYDp06fTrl074uLisFqt/PTTTyxcuJDZs2cDFxYamzBhAtOmTaN+/frExMTwr3/9i8jISAYNGlRVlymEEEJUawaNlmH123LOks+iI9s9hvads+bTPDCCA1kXRpp8e3wP3x3fS/uQaMJNZvacS+GcJZ9BdUtf2LdVYG02nT1aqNzqcjDz4DriazXh9T0/0zYwmn+3voeHt8wv9lgdg2Ov9DKvWb7DwpHc08w68i1Hc08DoFVp6B7SigfjbiXY4F/pbRJCVB9VHlzde++9nD17lilTppCamkqrVq1YsWIFYWFhAJw4cQK1+u8Otry8PB577DFOnTqFyWSiUaNGLFq0iHvvvdddZ9KkSeTl5TF27FgyMzPp0qULK1aswGis2slwQgghRHUWYPBiYovuPNT4H6xNPkKWzUKLwAjCvXxRUPgqcZc76HKhsOXscfe+t0Y1JVDvRUNzKIeyi16r0lurp11wNO/uX1Xk9j8zjvN005sB2H7uBAnZZ+gQFMvWjGOF6nYJqY+vruT/69n2fPIcVk7lZ6BXa6llCsRXZ8JQhoWKi2J3OtideZSX9s5FuST8dChOVp/Zzl9ZR/lPmwkEGfyu6vhCiJqvyoMrgPHjxzN+/Pgit61bt87j8bRp05g2bVqJx1OpVLzyyiu88sor5dVEIYQQ4oZg1hsx643cGdOcg5lpqFDx1l9raeIfxvMt+/Dv3b8WSljR0C+UCc268/7+9bzWZiBPbPmK5P+tSXWRl0bHjA6DmX1wvUdgcjnVJb9/dXwHjzbsVii4ahNQl1da3Yl/CcMGz1iyeG3vN2zJSHCXmTR6Hojpxl1R/8DvKibI5zjzeevgkmLbf9aayVcn1zI6ZgD6qwzghBAXBHsFV3UTrkq1CK6EEEIIUb1k2ywk5pzj+W0/4lIUVicn8EjjTizuMZxvk/7iYFYaPjoD98e1I8TozfDfFpJSkM2e8ym83f4ODmefYW1qAnaXk57h9bk5shGzDq7jtzMJxZ5Tq1K7swYCnMw7T+vAaD5sP4w/zibgpdLBkQJebX0nQYbi18vKsObw2LaPOZWf4VFe4LTx8ZFfyXNYGRPXC2/tlY1oOZ6XSo6j5MWQf07ZzD1RvQiU4EqIq+at9+bss2eruhlXRYIrIYQQooJZHA6ybRZOZGdhdTqI9QvEW6fDbKi+w9ULnHambF+B65LlMOcc2MSSIzsZWKcJg+q04ObIBiw/8RePb/nKXedw9hmGrl9Am6DadAiuw21RzYg1B5Njs9DQr/hMwAB9I5twaadQsMEHrUpD17AGdA1rgN1u56cjP5XYY6UoChvPHCwUWF3qi+MbubdO5ysOrk4XpJdaJ99pxaE4S60nhLg+SXAlhBBCVKAsawHfJhzgra0byLPbAVCrVNwW15B/3dSLYFP1XKx+29mTWJyFs/9l2y0sOrIDgG7hccw8uKHI/XdknGJHxil2nTvNB/+4CxcKHULq0iowil3nThaqH2HyY0z9LlwaXd0X24GAS3qyyiLTlsfXJzeVWMepuNh09hCDojpc0bGD9KWvfalTadCoqjQZsxCiCslfvxBCCFFBnC4Xq5KOMnXjGndgBeBSFL47cpDn16/kvKWgCltYvOO550utU+C0eax1VZRDWWnYnU7sLidPbf2KCU1681STPtT2CgAgQO/FyHqdmNHhHibv+NadxCraO5DbardAo76ydOsuFLLtpT+n52y5V3RcgAa+0RjV+hLr9AhtjZfmynrECpxWztuyOG/LIt9hueJ2CXG9KbAX0GN+D3rM70FBGf6eqxPpuRJCCCEqyDlLAW9t/b3Y7b8eP0qW1UKA8cp6ZypDtLd/qXUM6tI/RvjqjKhUgAKJuemM/n0BXcPq83TTPgQbfMh1WFlxeh/DN8zD6nKQbbdwR3Qrnmjcm2Bj8fOqiqNXa4n2DiHVkllivQa+EVd8bB+tkdGxA5h15Nsit3tpjAyv2x+T1lCm41mcVjKsmSw79QvbM/fiUhSamutzT1R/Qo1BeGur330hRGVwKS7WH1/v/r0mkeBKCCGEqCD5dhtp+SX3kPx+6jh1/QIqqUVl1z0iDp1ajd1V9Acbo0aLSaujqX84+zJTiz3OkJg26FQajuWcJcLkR0pBFuvTDrM+7XChulqVmkC9Ny80vwUvbck9RMXx1ZkYFduTrRnFJ87w0Rpp7Ff7io9t0OjpE9YOg1rH3MQfybbnubc18I3iuUb3E1LGNOwWp5Xt5/fx7qFPcV0yFHLzuV1sPreLUXUH0zvsH3hrq+ewUSFE0SS4EkIIISqMqvQapVepEmadkX+17suU7SuK3P5q2/4E6L2Z2voWhqybh1MpnJ48wmTm9jot2HX+JF8l7eCuum34z4G1xZ6zT0QjfHXGqw6sLor1CaV/RGt+TtlZaJtGpeb1lvdhvspeIV+dF33DO/CPoKactWaS57AQZgzAW2vCX1/2nrZsex7vHprnEVhdal7SN7T0byjBlRA1jMy5EkIIISqIl05HuHfJH7i71KpbOY25Ql46PbdFN2VB96E0C/g7y1/LwEiW9HyAm2s1wKjVUt8cwuLuI6jnG+Kuo0ZFj/B6fN5jJAa1ho8Pb2Rd6mHaB9WlqV9kkecLMfrybLN4fHRlG1JXEj+9NxMaDeCV5vdS1zvU3aauIY1ZeNMTtAiog05z9d8va9UaAg1mGpqjaRPYgFpeIVcUWNldDlambsBFycOdvjn1CwUyB0uIGkV6roQQQogKEmg08VyHbjy19qcit8fXrYfZcO3BREUx6410CY+liX84NpcDFSp0ag2Bxr97U7y0etoERbGw2zDyHTbynTb89Ca8NHrMeiPpllyS8zNxKgoTt37Nux3uIiU/C2+dHj+diQKnnXRLLu2C6xBuKj0bX1n56b3pG9mKdkH1cCouVCowqnX46Kp+HlOB08JfWQdLrbcvOwGLy4aJ6puyXwjhSYIrIYQQooJo1Gp61Ynl9a438+8tv5Fts14oV6m4o34TJv+je7VMZnG5S4Op4gQZvQnCu1C5Rq0mzGQmpSAbp+JCQSHdmst/Dm7jVP55AvRe3FmnNR1DYlBVwBjJwBIWG646KjSq0rMgqlFDMcMGhRDVkwRXQgghRAXyMxi5q0FTbq5bj9O52VgdDuqY/fHR6/HRV99eq/KiKDC9zSAyrHn4G7z4+NAGvj+12739vC2fuQkb+enUXhZ3G024qWwJIWoyb62Rm4LacCgnscR6HQNb4K2p/sG3EBXBS1cz5xtKcCWEEEKUgwK7nWybFYvDjl6jwaTV4f+/Xim9VkuIVkuIV+GenetVts3C3vPJvLt/DXvPJwPQOjCKMQ06UccnkP8c9ExskVKQxVt7f+GVVgPxLod5V9WZRqWhS0gbPj/5AxantZg6am6t1Qu95tqSewhRE3nrvcl7Ia/0itWQBFdCCCHENUrPz+O9P//g64P7sDodAHSKjOL17jcTbfZHo76x8kfl2a0sP7Gb1/9a6VG+89xJxm8+yQst4rm7Tlu+Or7dY/uq5AM81yweb52BDGsux3MzWJN6ABUq+kY2JewKkkZUd2atL680fZIp+94vFGBpVRomN36YAF35zUETQlQOCa6EEEKIa5BRkM/YFcvZkZbiUb4p+SS3f7OYFfeOINLnxvqQnGO38O+/fil2+9t7f2VRt5F8c3yHRypyh+LC4rRzOv88Yzct4GT+Ofe2Bcc20sQnjKFc+fpU1ZFWrSHGuzYz27zE5oyd/JG+ExcKbfyb0CO0I2att/RaCVEDSXAlhBBCXIOj588VCqwuyrZZmbNjKy/c1AOj9sb5l/vz6f3Frt8EYHM52ZFxgtZB0WzPOO6xTavWMOqPuaQUZBXa72jOWaA25215hOr8y7nVlU+r1hCo9yM+vBvdQjqAomDSmG64nk4hLmdxWBj85WAAvrnnG4zampMx88Z5pxdCCCHKmd3pZMn+3SXW+e7IQca37VQhwdV5SwF2xYkK8Ncbr2ntpvKiKAoJ2WdLrZecn0WQwXMOWvvguuTarUUGVp77ZhLq7X8tzaxWNCo1PrJYsBBuTpeTnxJ+cv9ek1T9u7AQQghRQzkVBauz5H/8VqcDVOWbTjvLWsC+c2d4d9cG9mSkYdYZuLt+c0Y0akuoybvUlOZWp4MsmwWLw45OrcGo1RFgKJ+sdCqVilifoFLrhZvMHMpJdT82qLW82OIWlp/YXsJeF/ySso9WIXWvpZlCCFEhJLgSQgghrpJRqyU+ph4/HztcbJ2bakVjLMcepWyrhfkHtvPe7o3usrNOB7P2bOarI3tYfsswavkUn848w5LH3IPbWHR4Bzn2C4kU2gTX4pX2fYk1B2HS6q65jQOimvHuvjXFDg3UqdS0D67Dewd+BaBdUB3+1XIAtUx+pFqySz2+U3EVuy3XbsHqsqMAOpUGP730CAkhKo8EV0IIIcQ16Fy7DiEmL84W5BfapgKe6dAFs6H85gucsxZ4BFaXOluQx6vbVvNW5wH4FrGGVoYlnyd//46NaZ7znHakn+aOlQv4uu9wWgRFXHMbfXRGnmrai3f2rS5y+wst+xFm8uXH3uPRazQYNTr8/xcE9YlozKqUfSUev0doo0Jl+Q4rKQXnmZPwK7+fOYRLcdEyoA6PNLiZ+r7h+OrK3jPnUlxk2vI4XZDOnswkjBod/whqjI/WiFl/46TTF0JcOQmuhBBCiGsQbPLiqzuGMvLHb0jKynSX++r1vNOrP3XM5bcortPlYvGhnSXWWXXyCHkOW5HB1dHs9EKB1UV2l4t/bl3Bgp73Emi8tt4eX52Be2PaEOcbzIz9azmcfQaApv4RTGzaixaBtfDVFR1wtg+KwV/vRaatcLB6UYw5xOOxxWnnz4xjPLtjkUdv2c7zSTy85WPGNejL4OiOZQqwnC4nx/PP8PyuuZyxZrrLP2A5XUKa8XSjuwi4jlLCCyHKlwRXQgghxDVQqVTU9Qvgq0FDSc3N4dC5dEK9vGkUFIq/wYC+HBNZ2FxOTuRmlljnwjwwR6HyfIeNeQe3lbjv3nOpFDjt19JEN7PeRK/IhrQMrI3N5UClUqFXawg0lNzzE2Dw5tNOoxn1x1yy7AUe24KNvlAAQZf1HuXaC3hh1+fFDkOcefgXeoY3LVNwlWHL4fE/Z5LntBTa9vvZvThdTp5vMgQ/6cESQhRBgishhBCiHIR4eRPi5U3z0PAKO4dBraGeXxArSSi2jk6tLnKOl93l4ry1cMBwOVspCTquVJDxyoIQjUpNjG8w3/Z4nC3pR1mVsh+1SsWAWi1o6hvJltXrC+2zKT0Bq6twQHmpzxP/YGLjW9Brip9TZnc5WH5qY5GBlftcGQfIduRLcCWEKJIEV0LcYHJyLVgsdhxOF0aDFn8/r1Iziwkhqge1Ws2QBi2ZuWdzsXVuqdMQH13hIYFeWh0tgiLYcuZEsfvq1ZpyTb5xtTQqNcFGHwbUbknviCYAGDU67Paie9X+Ol/8NV2UkJNCgdNeYnCVbc/nl9TSsxWuTdvN8Jg+pdYTQlwdb703ykvlm2W1ssgqdULcQDLO5fL6Wz9wz/DZDB35fzzxzOccPpKG3V6z1pAQ4kYWoDcxtUPvIrdF+fgxuW1PvHX6Qtt0ag0P1G9NSV+l9I9uyI70U5y3Fj/fqbIZNTqMJQREAIGG0udA+epMaFWlf+yxOG2l1sl1FJRaRwhxY5LgSogbRFZ2Aa+/9QObth5D+d+XQSdPn+PJZz8nK7v6fJASQpTMR2/gzrhm/HfACPpE1SPU5EOcOZAX2/Vi2S3DCPf2LXbfQKMX73S6tcgAq5F/CCMatuPFbStYceIQLqXmfGs8oFbrUuvcW6cT3sUk0bhIp9YQ5xNZ6rFa+MWUuW1CiBtL1ff9CyEqhdVqZ/uuwkNnLFY7m7cd49Z+LaugVUKIq2HWG2keHM67XW7F4rCjVqkIMJjQqEv+ztRHZ6BHZBzL+41kScIO9p0/g49Oz8A6TajrG8jjv39Lls3CzH0b6VO7PiGmmpEVz0/nRd+IFvyS8leR2xuYI2hoLj1oMuu8GR5zM0/v/L9i6/hoTTTxq3PVbRVClM7isDDs22EALLxjIUZt+S1nUdEkuBLiBmErYehfekZuJbZECFFezHoD5iJSrpfEobh4bvMPdAiN5s6YZhQ47Xyd+Bc700+76yTnZ2N31Zzhwn56L55pfBv+em++PbnV3XY1KrqFNua5ZreXaeggQH2fSAZEduTH5C2FtulUGl5vMQpfnSxMLERFcrqcfL3/awDm3z6/ahtzha4quHI4HKxbt46jR49y33334evrS3JyMmazGR+fmvEtlxA3Gi+jDj+zif9v787Dm6rSP4B/b/al+95CaVllLSBLRUSQrQg6ougAMsMig84IjtpRHBzHgjiCyjg4o4IbqL+RzV3RYasUUSoIWGQXkLKVLhS6JU1yk9zfH0ggNE2TkjRJ+/08T58nPffNvW/Tk9O8veeeW1lV91qBgTd0CEBGRBQIAgRYJTv+74j7hRtkHlyfFEyi1XrM7DQSU9sNRqGhDDa7HW3D4qFTqL26gXCESo8HOozGoPhuePf4JvxcfRpKQYEhiRn4ffpwxKkjoZTJ/fiTEFEo87q4OnHiBEaNGoWTJ0/CbDZjxIgRCA8Px/PPPw+z2YylS5f6I08iukbh4Vr89S+j8dS8j2GzX76WYuSwbkiIr/8aDSJqXqLUGoxrl4EXCjbXGzMwMR1aD1YNNIoWVFvN2Fp8FIU159E+PA4DE9shXKmBVuF+EQp/0CnU0CnUiNdEXNN+IpR63BDXFZ0j2sAuSZAgQSfXQKuou1AIEdGVvC6uHn74YfTt2xd79uxBbGyso/3OO+/EjBkzfJocEfmOUilHr4xU/PftGfhqw15UVdVi5LBuaN0qBpERnv9Xl4hCm0yQIav1dVh+aAfKTIY62xWCDH/uMQjyBs5cVVlMWHf6AOb9+D+Ikt3RrpLJ8Vyf2zE0pZPLJeFDSZSKs3GIyDten/PfunUrnnrqKahUzv+9SU9Px5kzZ+p5FhEFA61GheSkKEyfPAiPzhqJbl1asbAiamEumI14aU8elt58N/rFpzptaxsegyU3j8MHxwpgsLpfkvxAxVk8tftLp8IKACx2Gx774VMcqzrn89yJiIKd12eu7HY7bC7u3n769GmEh3NqERE1XmV1LaqNZhSXVSEpPgLhOjUiw1n8EfmSXZLwY3kRHt32GaZ3zsRfew9FaW0NolRaXDAb8fqBfOwsO41HMm6udx/nzUYs2vu12+P8c9/X+PcNdyNKzfcwEbUcXhdXI0eOxOLFi/HGG28AAARBQE1NDXJycjB69GifJ0hELcP5SgOeXbIO+XsKHW2ZPdLw9MzRiInkylxEvqKVK5ERm4x1pw4jZ+d6KAQZwlVqGEQLLL+uspegDXO7aINot+GnC0Vuj/N9WSFEu9WnuRMRBTuvpwUuWrQI3333Hbp27QqTyYR7773XMSXw+eef90eORNTMmcwiln/8vVNhBQDb957Amx9sQ61ZDExiRM2MJEmw2K14sNuNjjarZMcFc62jsAKA+7tkIkbNf2oQUWDolDrUzKlBzZwa6ELs1gdeF1epqanYs2cP/va3v+HRRx9F7969sXDhQvz4449ISEjwR45E1MzVGM34Im+fy23/+2Y/DEZzE2dE1PxUmGvxyfF9uHPde9hadByP9nA97S+r9XW4M72H2xsSq2RyZES7vylvZnwalDLeTpOIvCcIAvQqPfQqPQRBCHQ6XvFq1BNFEZ07d8batWsxadIkTJo0yV95EVELYpckmC2upw+ZRSvsVywdT0Teq7KYsGR/Pt44cPHGuC8WbMFDPQZi9fDf49Pj+/BLdTniNXpM75yJ1LAoRGvc/6c4Wq3DYz2GYvI3/6035rHuw3i9FRG1OF6duVIqlTCZTP7KhYhaKJVCjs7tEl1u65SeAKWSN+wkuhaVFpOjsAIACcC/936H6Zs/gARgRKtOeLrPSPSMS0FMA4XVJd2ikjH/+jFQXrVku0omx6J+Y9E+Is6HPwERtSRmqxlTP52KqZ9OhdkaWrNXvD5fP3PmTDz//PN46623oFDwdD8RXbuoCB1mTx+OB3JWQbRevu5DIZdh9vThiI4IrfnWRMHmk19cT7utFs1YeaQAAHBjUjritHqP9xmu0uD21O4YktwRW4uP4kTNBbQLjw3oTYSJqHmw2q14d8+7AIBXR78KNULnnnleV0c//PADcnNzsWHDBvTo0QN6vfNA/PHHH/ssOSJqOdq1jsP7L07B/32+A0dOlKFDahwm35GJhFje4oHoWhUZKhuMuWCu9Xq/OqUKOqUKd7ft3Zi0iIiaHa+Lq6ioKIwbN84fuRBRC6ZWKZCaFI3sKUNhMovQqJXQqPmfbyJf6BXXCquP/eQ2JkUf0UTZEBE1X14XV8uXL/dHHkREAMCiisgPhrbqAJVM7rTc+pV6xaYgXBk6026IiIKV10uxX1JWVoZvv/0W3377LcrKynyZExEREflQhEqD1wePg0Ko+2c/XqvHyzfd4fFCFkREVD+vz1wZDAY89NBDeO+992C32wEAcrkckydPxn/+8x/odByciYiIgolGoUBmYhvk/uZ+vH1wB74tLoRarsC4dj1we3pXxGs8X8iCiIjq53VxlZ2djS1btuCLL77AwIEDAQDffvst/vznP+Mvf/kLlixZ4vMkiYiI6NpoFUq0CY/GX68fCoNogSAAUSqt25sFExGRd7wurj766CN8+OGHGDJkiKNt9OjR0Gq1+O1vf8viioiIKIhpFUouk05EQU2n1KH0sVLH41Di9b+rjEYjEhPr3uwzISEBRqOxUUm8+uqrSE9Ph0ajQWZmJnbs2FFv7JtvvolBgwYhOjoa0dHRGD58eJ34qVOnQhAEp69Ro0Y1KjciIiLyr2qLCSW1VThjrECZqRpmmxjolIgogARBQLw+HvH6eAiCEOh0vOJ1cTVgwADk5OTAZDI52mprazFv3jwMGDDA6wRWr16N7Oxs5OTkYPfu3ejZsyeysrJQWlrqMj4vLw8TJ07E5s2bkZ+fj9TUVIwcORJnzpxxihs1ahTOnj3r+Fq5cqXXuREREZH/WGxWHK8+h9m7PsKwDf/CyI2LMSb3Ffz74GacM9UEOj0iIq95PS3w5ZdfRlZWFlq3bo2ePXsCAPbs2QONRoP169d7ncBLL72EGTNmYNq0aQCApUuX4ssvv8SyZcvw17/+tU78+++/7/T9W2+9hY8++gi5ubmYPHmyo12tViMpKcnrfIiIiKhpFBrKMfGbt2C64kyVwWrGO8e2YVvZMbw14PeI1YQFMEMiCgSz1Yzs9dkAgJeyXoJaETq3ivC6uOrevTuOHDmC999/H4cOHQIATJw4EZMmTYJWq/VqXxaLBbt27cKcOXMcbTKZDMOHD0d+fr5H+zAajRBFETExMU7teXl5SEhIQHR0NIYOHYpnn30WsbGxLvdhNpthNpsd31dVVQEARFGEKHJqgiuXXhe+PuQO+wl5gv2kZaqwGDFv9+eQbDaoXUykOVFVhg2n9uGutN6Q/bqEPPsKeYL9JPTVWmrx2s7XAAD/GPIPyCTfL7zjTT/xpi8JkiRJjc7qGhUVFaFVq1bYtm2b05TC2bNnY8uWLdi+fXuD+3jwwQexfv167N+/HxqNBgCwatUq6HQ6tG3bFseOHcOTTz6JsLAw5OfnQy6X19nH3LlzMW/evDrtK1as4NLyRERERERNyGQzYcLeCQCAVT1WQSPXBDQfo9GIe++9F5WVlYiIiHAb6/WZqwULFiAxMRH33XefU/uyZctQVlaGJ554wttdNtrChQuxatUq5OXlOQorAJgwYYLjcY8ePZCRkYH27dsjLy8Pw4YNq7OfOXPmIDs72/F9VVWV41quhl7AlkoURWzcuBEjRoyAUslVp8g19hPyBPtJy3SkqgS/+3aZ2xiVoMAnQ/+EOPXFqYHsK+QJ9pPQZ7AYgL0XH2dlZUGv8v29+LzpJ5dmtXnC6+Lq9ddfx4oVK+q0d+vWDRMmTPCquIqLi4NcLkdJSYlTe0lJSYPXSy1atAgLFy7Epk2bkJGR4Ta2Xbt2iIuLw9GjR10WV2q1Gmp13bmcSqWSb8oG8DUiT7CfkCfYT1qWaG04zLC7jUnRR0ClqNsv2FfIE+wnoUspXf69+fv36Mn+vTm+1xMYi4uLkZycXKc9Pj4eZ8+e9WpfKpUKffr0QW5urqPNbrcjNzfX7cqDL7zwAubPn49169ahb9++DR7n9OnTKC8vd5k3ERERNT2NXIle0aluY6a2H4AYte//Y01E5C9eF1epqan47rvv6rR/9913SElJ8TqB7OxsvPnmm3j33Xdx8OBB/OlPf4LBYHCsHjh58mSnBS+ef/55/P3vf8eyZcuQnp6O4uJiFBcXo6bm4pKtNTU1ePzxx/H999+jsLAQubm5uOOOO9ChQwdkZWV5nR8RERH5XrRah2d6/wY6hcrl9m5RKRia3Dnk7nFDRC2b19MCZ8yYgUceeQSiKGLo0KEAgNzcXMyePRt/+ctfvE5g/PjxKCsrw9NPP43i4mL06tUL69atc9yo+OTJk5DJLteAS5YsgcViwd133+20n5ycHMydOxdyuRw//fQT3n33XVRUVCAlJQUjR47E/PnzXU79IyIiosBoo4/Bx0P+iJcP5GLj2YOwSnZEq3SY1K4/fpvel2etiCjkeF1cPf744ygvL8eDDz4Ii8UCANBoNHjiiSeczjB5Y9asWZg1a5bLbXl5eU7fFxYWut2XVqtt1P22iIiaUpXRBLPVBoVMhugw725jQdRcKGVypOpjMK/XbzC7xyjY7PaL7wmVDgpZ3dV9iahl0Cq1OP7wccfjUOJ1cSUIgmNq3sGDB6HVatGxY0eeFSIi8kCNyYzCkgt45cvvcPjMOSRHh+MPI/ujT/tWiNSH1h8QCg3nag2otYmwSxK0CiWi1Voog6xw0SvV0Cv5OYKILpIJMqRHpQc6jUbxuri6JCwsDP369cOJEydw7NgxdO7c2Wn6HhERObPabNh++BT+suwLXLrD4PlqIx596wtMH9EP9w3vhzAtP2CSb1RbzNhVdhrzd+biWNV5AECMWos/drsB93TIQLSaxTwRka95XA0tW7YML730klPb/fffj3bt2qFHjx7o3r07Tp065fMEiYiaiwqDCf9YkwtXt25fvmknakyWpk+KmiXRZsPWs8cx9esPHIUVAJw31+K53Zux6MctqDSbApghEVH9LDYLHt/wOB7f8DgsttD62+hxcfXGG28gOjra8f26deuwfPlyvPfee/jhhx8QFRWFefPm+SVJIqLmwGi2oLza6HKbXZKw94R3t7Mgqk+FpRZP79hQ7/b3jxSg0lLr9zxqRBOqRBNskvv7WRERXUm0iViUvwiL8hdBtImBTscrHk8LPHLkiNM9pT777DPccccdmDRpEgDgueeecyyfTkREdckaWFJapWj0TG0iJyW1NThncl3IX/J54UHM6nGjX45/zlSD7WWF+OTkHoh2O4YkdcDo1t0Qo9JDKQ+u672IiHzJ47/ktbW1iIiIcHy/bds2TJ8+3fF9u3btUFxc7NvsiIiaEa1KiTbxUThZVlFnm0ohR9fUhKZPipqlGou5wZjzDRRfjXXaUIHfffMOztZWOdq+LzuOlw/k4d1Bv0fXyGQWWETUbHk8LTAtLQ27du0CAJw7dw779+/HwIEDHduLi4sRGRnp+wyJiJqJmHAdFky+FWpl3Q+Wfx8/jItZkM+0DotqMOaGxDY+P265yYA/5a9yKqwuqbWJmPbtf3HB4p+ijogoGHh85mrKlCmYOXMm9u/fj6+//hqdO3dGnz59HNu3bduG7t27+yVJIqLmQBAEdEqJx6d/m4rV3xRg78lipMZG4fdDr0dSdDi0KmWgU6RmIkypwk3J6fj2bKHL7VEqDXrHt/L5cUtN1fi5qrTe7QarBVuKj+Cettf7/NhERMHA4+Jq9uzZMBqN+Pjjj5GUlIQPPvjAaft3332HiRMn+jxBIqLmRKWUIyUmArNuGwijWYRaqYBGxWutyLei1FosunEMJm1c6bRaIACEK9V4f8RExPhhKfY95880GLO15Bh+0yYDajn7PRE1Px6PbDKZDM888wyeeeYZl9uvLraIiKh+SoUckQped0L+k6QLx+qRk1BQXoRVR/bAYrNhWOsOGNXmOsRqtJD74d6UWkXDZ181ckWDi7sQEYUq/tuIiIiomYrT6jG8dUcMSEyDJEnQKVV+LWz6x6VBBgF2uLiZ268mtOsDpYz/WCCi+mmVWuz70z7H41Di+39bERERUVDRK1UIU6n9fsYoTKnGeDfXU3WLSkZaWKxfcyCi0CcTZOiW0A3dErpBJoRWuRJa2RIREVHQCldq8HDXWzC5fX8or/pANCixPV6/cSJi1foAZUdE5H+cFkhEREQ+E63W4ZGut+APnW7EocoSiHYbOkcmIVypRqQqtKb3EFFgWGwWPLf1OQDAk4OehEquCnBGnmNxRURERD6lV6qhV6qRqI0IdCpEFIJEm4h5W+YBAB6/8fGQKq58Ni3w1KlTuO+++3y1OyIiIiIiopDis+Lq/PnzePfdd321OyIiIiIiopDi8bTAzz//3O32X3755ZqTISIiIiIiClUeF1djx46FIAiQpPrvXSHwpoBERERERNRCeTwtMDk5GR9//DHsdrvLr927d/szTyIiIiIioqDmcXHVp08f7Nq1q97tDZ3VIiIiIiIias48nhb4+OOPw2Aw1Lu9Q4cO2Lx5s0+SIiIiIiKilkmj0GDHH3Y4HocSj4urQYMGud2u1+sxePDga06IiIiIiIhaLrlMjn6t+gU6jUbxeFrgL7/8wml/RERERERE9fC4uOrYsSPKysoc348fPx4lJSV+SYqIiIiIiFomi82CF797ES9+9yIsNkug0/GKx8XV1WetvvrqK7fXYBEREREREXlLtImYvWk2Zm+aDdEmBjodr3hcXBEREREREVH9PC6uBEGoc5Ng3jSYiIiIiIjoIo9XC5QkCVOnToVarQYAmEwm/PGPf4Rer3eK+/jjj32bIRERERERUQjwuLiaMmWK0/e/+93vfJ4MERERERFRqPK4uFq+fLk/8yAiIiIiIgppXNCCiIiIiIjIBzw+c0VERERERORvGoUGm6dsdjwOJSyuiIiIiIgoaMhlcgxJHxLoNBqF0wKJiIiIiIh8gGeuiIiIiIgoaIg2EW/segMAcH+f+6GUKwOckedYXBERERERUdCw2CyY9b9ZAICpvaaGVHHFaYFEREREREQ+wOKKiIiIiIjIB1hcERERERER+QCLKyIiIiIiIh9gcUVEREREROQDLK6IiIiIiIh8gEuxExERERE1EUmywy6ZIUABmSx0lhhvSmqFGmsnrnU8DiUsroiIiIiI/Ey0VcEmVaOo+mPUiAchF3RICR8PnaIt1Iq4QKcXVBQyBcZ0GhPoNBolKKYFvvrqq0hPT4dGo0FmZiZ27NhRb+ybb76JQYMGITo6GtHR0Rg+fHideEmS8PTTTyM5ORlarRbDhw/HkSNH/P1jEBERERHVIdouoLDiVWw7PQyFla/gnDEXJYYv8GPx77CreDyM4glIkhToNMkHAl5crV69GtnZ2cjJycHu3bvRs2dPZGVlobS01GV8Xl4eJk6ciM2bNyM/Px+pqakYOXIkzpw544h54YUX8O9//xtLly7F9u3bodfrkZWVBZPJ1FQ/FhERERERrHYjTlYtx6nqdwDULaBM1tPYXXwvLLZzTZ5bsBJtIt4peAfvFLwD0SYGOh2vBLy4eumllzBjxgxMmzYNXbt2xdKlS6HT6bBs2TKX8e+//z4efPBB9OrVC507d8Zbb70Fu92O3NxcABfPWi1evBhPPfUU7rjjDmRkZOC9995DUVERPv300yb8yYiIiIiopbPZa3Cy0vXn2ksstnMoNnwOu2RroqyCm8VmwbTPpmHaZ9NgsVkCnY5XAnrNlcViwa5duzBnzhxHm0wmw/Dhw5Gfn+/RPoxGI0RRRExMDADg+PHjKC4uxvDhwx0xkZGRyMzMRH5+PiZMmFBnH2azGWaz2fF9VVUVAEAURYhiaFXLTeXS68LXh9xhPyFPsJ+Qp9hXyBPB1k/KDNtgt8kAuF+Y4dSFNYhTj4FKHts0iQWxK393oihCFHz/u/Smn3jTlwJaXJ07dw42mw2JiYlO7YmJiTh06JBH+3jiiSeQkpLiKKaKi4sd+7h6n5e2XW3BggWYN29enfYNGzZAp9N5lEdLtXHjxkCnQCGA/YQ8wX5CnmJfIU8ETz8RAMxvMKoGwKY92/2eTSgw2S5fyrN+/Xpo5Bq/HcuTfmI0Gj3eX0ivFrhw4UKsWrUKeXl50Gga/6LPmTMH2dnZju+rqqoc13JFRET4ItVmRxRFbNy4ESNGjIBSyWVEyTX2E/IE+wl5in2FPBFs/aSo+kMcvbCgwTilLBp9kldBJefKgQaLAdh78XFWVhb0Kr3Pj+FNP7k0q80TAS2u4uLiIJfLUVJS4tReUlKCpKQkt89dtGgRFi5ciE2bNiEjI8PRful5JSUlSE5Odtpnr169XO5LrVZDra57qlapVAbFmzKY8TUiT7CfkCfYT8hT7CvkiWDpJ4kRg3Gs6hkAdrdxKZG/gVYVy3tfAVBKl18Df/8ePdm/N8cP6IIWKpUKffr0cSxGAcCxOMWAAQPqfd4LL7yA+fPnY926dejbt6/TtrZt2yIpKclpn1VVVdi+fbvbfRIRERER+ZpcCEOizv09m2SCFq3DJ0ImUzVRVuQvAZ8WmJ2djSlTpqBv377o378/Fi9eDIPBgGnTpgEAJk+ejFatWmHBgounU59//nk8/fTTWLFiBdLT0x3XUYWFhSEsLAyCIOCRRx7Bs88+i44dO6Jt27b4+9//jpSUFIwdOzZQPyYRERERtUBKeQQ6xj4J0X4e503f1dkuF3TolfQOVPL4AGRHvhbw4mr8+PEoKyvD008/jeLiYvTq1Qvr1q1zLEhx8uRJyGSXT7AtWbIEFosFd999t9N+cnJyMHfuXADA7NmzYTAYcP/996OiogI33XQT1q1bd03XZRERERERNYZKHoOu8YtQaz2FExWvw2g9DrmgQ1LYnUjQjYRSHg2ZwOmAl6gVaqy5e43jcSgJeHEFALNmzcKsWbNcbsvLy3P6vrCwsMH9CYKAZ555Bs8884wPsiMiIiIiujYqeczFIivuBdglMyAIUMoiIQjyQKcWdBQyBe7pdk+g02iUoCiuiIiIiIhaAoU8DEBYoNMgP2FxRRRA1dW1sFntCI/QQi4P6PoyREREREHBarfik4OfAADu7HInFLLQKVlCJ1OiZqSmuhZHfi7Bine+RXV1LW4Z0Q1Zo3siKtr393EgIiIiCiVmqxm//fC3AICaOTVQqEKnZAmdTImaCZvNju+3HcXzz3zmaDt2pATbvzuCp/9xNwssIiIiohDFeUhETayywoi3Xvu6TvvePadwvrwmABkRERERkS+wuCJqYpIkofxctcttRWcuNHE2REREROQrLK6ImphcLkN6W9c3CmzXPqGJsyEiIiIiX2FxRdTEoqL1ePSvY6BQOL/9bht7PcIjtAHKioiIiIiuFRe0IAqA9h0TsXzVg/jfFwW4UF6DkaMz0CY9jsUVERERUQhjcUUUAGq1EknJUZh2/xDY7RJkMiHQKREREREFBZVcheV3LHc8DiUsrogCjIUVERER0WVKuRJTe00NdBqNwmuuiIiIiIiIfIBnroiIiFqwGtGMaosZ/zt1CEcrz6FTVBxGpXZGuFIDvbJppuMYrRYYrGbYJAkKQYYolRYKmbxJjk1Ewcdqt2L90fUAgKwOWVDIQqdkCZ1MiYiIyKeqLSZ8Vrgf83ZtgE2SHO3/2J2Lf/S7Fbe26YJwldpvx7fZbSg2VePVQ3n46vQ+WOxWxGvCMaX9DRjbphdi1Hq/HZuIgpfZasZtK28DANTMqYFCFTolC6cFEhERtVCHK8rw9M71ToUVANgkCX/d8RV+qS736/FPGStw1+al+PRkASx2KwCgzFSNRfs34s/bV+G82eDX4xMR+RqLKyIiohaowlyLRT9tcRvzzz15qDTX+uf4FiP+/uPnqBZNLrfvPn8KW0p+hnRV4UdEFMxYXBEREbVAot2GnWWn3MZsLz0Ji93ml+PXWkXsKj/hNmb5kW08e0VEIYXFFRERUQulENx/DFD6cVGJaqvrM1ZXOmOsgB08c0VEoYPFFRERUQukUygxKvU6tzG3pnaGXuGfFQP1ioYXykjQhEMG3guQiEIHiysiIqIWSK9U45GMm6Gq5+yURq7AQz1ugs5Py7Hr5Cp0j0pxG/P79jcgWq3zy/GJiPyBxRUREVELlayLwKrhv0NrfaRTe5uwKKwe/nskacP9duxotQ7ze/8G6nruX9MhPAFZKV0ha2DqIhE1Pyq5Cq/c+gpeufUVqORNc789XwmdReOJiIjIp9RyBTJiU/DxyKkoNdWg2FiNZF044jVhiNXoIAj+nZLXNiwOH93yABbuXYfvSn+BBAk6hQrj0q7HjI43IVYT5tfjE1FwUsqVmNl/ZqDTaBQWV0RERC2YTBAQp9UjTqtH1+jEJj22Sq5Au/B4vNj3bphsIkS7DSq5ApFKDdRyZZPmQkTkCyyuiIiIKKAiVVpEQhvoNIgoSNjsNmw9uRUAMKjNIMj9uHKpr7G4IiIiIiKioGGymnDLu7cAAGrm1ECv0gc4I8/xKlEiIiIiIiIf4JmrIFZRWokf1hfg5KEzGD7pZsSnxkIXzmkTRERERETBiMVVkKoorcRTty/E4R+OAgBWL/wUizbPRcbNXQOcGRERERERucJpgUGq6nyNo7ACAEmS8N7cNai+UBPArIiIiIiIqD48cxWkXN1bRCaXQYB/7zlCRETeE21W1IgiFIIM4Wp1oNMhIqIAYXEVpMJjwtBjUGfs3XoIwMXCaur8CQiLDp3VUoiImjuT1YryWiPe37cH3585hTCVClN69EavpGTEanWBTo+IiJoYi6sgFRUfgac/eAz783/G6Z+LcNOdmYhJjgp0WkRE9CuLzYqCkrOY/PlHsNhtjvZvTp3A9YnJeHP0WMTqWGAREXlLKVfiheEvOB6HEhZXQSwqIRID7+gX6DSIiMiFSrMZ07/8xKmwumR3yVm8t68AM/tkQiUPnZtfEhEFA5VchccHPh7oNBqFC1oQERE1ws6zZ2AQxXq3v/vTj6gwmZowIyIiCjSeuSIiImqEI+fL3W6vMJtgk+xNlA0RUfNhs9uw++xuAMD1yddDLgudGQAsrijoSJKEyvMGSHYJEdF6yBU8wUpEwadtVLTb7REqNeQCxy8iIm+ZrCb0f6s/AKBmTg30qtBZ0I3FFQWV6kojCr47glWvbITZJOLWiQMw/O5+iOQqiUQUZDJTWkOrUKDWanW5/XfdeyKqmSzLLkkSzpkMOFBRjM1FR6FRKDE2rTsSNOGI0XDRDiKiS1hcUVA5sPM4npv5ruP7t577HIYqIybMGgGVOrRWiyGi5i1SrcGSUb/BH776FFa78/S/bnEJuK/n9VApQv/PrF2SUFh9HpM2/xelpss3sn/z0Pe4ISEN/x5wJ2I1/AcYERHABS0oiFRdMOCDpV/Xaf/sna0wVPGicCIKLmqFApkprZF77zT8rltPXBcThz5JKXh5xGi8d/s4xOmaR8FRbjJgwtf/51RYXfJ96QnM270eVRaO0UREAM9cUZCx2epe/G23SwCkpk+GiKgBWqUSaZFR+NvAwTCIIuSCgGitNtBp+dQPZadQbjbUu/1/pw9hTq/hiFBpmjArIqLgxDNXFDQiovW4c/rgOu2j770RunD+0Sai4KVVKhGn0zW7wspis2H96UNuY+yShJ8ry5ooIyKi4MbiioJK74Gd8OcF9yCxdQyi48IxYeZw/PZPQ6HWqAKdGhFRiyMIgFzW8EcFBVdFJCICwGmBFGTCo3QYeU9/3DCsOyRICI/UQaliNyUiCgSlTI5xbTPw2Yl99caoZHJ0jIxrwqyIqLlTypXIGZzjeBxK+KmVgo5cLkd0fHig0yAiIgDXRSagfXgsjlW7vmny5I59oVc0jyXniSg4qOQqzB0yN9BpNErAz+O/+uqrSE9Ph0ajQWZmJnbs2FFv7P79+zFu3Dikp6dDEAQsXry4TszcuXMhCILTV+fOnf34ExARETVfcRo93htyL3rHtnJqlwsCJnfsiwe63Ai9klO3iYiAAJ+5Wr16NbKzs7F06VJkZmZi8eLFyMrKwuHDh5GQkFAn3mg0ol27drjnnnvw6KOP1rvfbt26YdOmTY7vFc3gPiNERESBkqSLwBs3/RYXLLXYc/4MNHIlro9thTClGmFKnrUiIt+yS3YcLDsIAOgS3wWyELquM6BVx0svvYQZM2Zg2rRpAIClS5fiyy+/xLJly/DXv/61Tny/fv3Qr18/AHC5/RKFQoGkpCT/JE1ERNQCxWh0iNHo0D4iNtCpEFEzVyvWovuS7gCAmjk10KtC576BASuuLBYLdu3ahTlz5jjaZDIZhg8fjvz8/Gva95EjR5CSkgKNRoMBAwZgwYIFaNOmTb3xZrMZZrPZ8X1VVRUAQBRFiKJ4Tbk0V5deF74+5A77CXmC/YQ8xb5CnmA/CX1X/u5EUYQo+P536U0/8aYvBay4OnfuHGw2GxITE53aExMTceiQ+3tquJOZmYl33nkH1113Hc6ePYt58+Zh0KBB2LdvH8LDXS+SsGDBAsybN69O+4YNG6DT6RqdS0uwcePGQKdAIYD9hDzBfkKeYl8hT7CfhC6TzeR4vH79emjk/rvfqSf9xGg0ery/Zncx0q233up4nJGRgczMTKSlpWHNmjWYPn26y+fMmTMH2dnZju+rqqqQmpqKkSNHIiIiwu85hyJRFLFx40aMGDECSmVoLZFJTYf9hDzBfkKeYl8hT7CfhD6DxQDsvfg4KyvLL9MCveknl2a1eSJgxVVcXBzkcjlKSkqc2ktKSnx6vVRUVBQ6deqEo0eP1hujVquhVte9IFepVPJN2QC+RuQJ9hPyBPsJeYp9hTzBfhK6lNLl35u/f4+e7N+b4wds6Q2VSoU+ffogNzfX0Wa325Gbm4sBAwb47Dg1NTU4duwYkpOTfbZPIiIiIiKiqwV0WmB2djamTJmCvn37on///li8eDEMBoNj9cDJkyejVatWWLBgAYCLi2AcOHDA8fjMmTMoKChAWFgYOnToAAB47LHHcPvttyMtLQ1FRUXIycmBXC7HxIkTA/NDEhERERFRixDQ4mr8+PEoKyvD008/jeLiYvTq1Qvr1q1zLHJx8uRJyGSXT64VFRWhd+/eju8XLVqERYsWYfDgwcjLywMAnD59GhMnTkR5eTni4+Nx00034fvvv0d8fHyT/mxERES+VG0xo0a04EB5CSQA3WITEaZUIVzF+0wRUfOilCvx2IDHHI9DScAXtJg1axZmzZrlctulgumS9PR0SJLkdn+rVq3yVWpERERB4bzJiBd3bcXqn3+C7de/g3JBwG879sDsvjcjRhMcK9tWiyYYRAt2lJ1AtWhGRmwKkrURiNOEBTo1IgohKrkKL458MdBpNErAiysiIiKqX7XFjEW7tmLF4T1O7TZJwsqffwIA/K3/LQE/g3XBbMTrh7/Du0e2OwpAAEgPi8EbAyciLSwagiAEMEMiIv8L2IIWRERE1LAa0YJVvxZRrqw5shc1oqUJM6rLbBPx/rGdWPbz906FFQAU1pzHxLx3cM5sCFB2RBRq7JIdhRWFKKwohF2yBzodr7C4IiIiCmIHykvrFCxXskkS9peX1Lu9KVRazHjr8LZ6t5ebDfi66OcGp/YTEQFArViLti+3RduX26JWrA10Ol5hcUVERBTiAl2yFNdWwmgT3cZ8dnIvqiymJsqIiCgwWFwREREFsa6xCZC7uVZJLgjoHpvYhBnVZbU3PG3HarfBHvAykIjIv1hcERERBbEwpQrjO2XUu/23HXsgTKnyaF+SJOFcrQFFhioUG6tRYfbNdJtW+igoBfcfKYYmd0K4UuOT4xERBSuuFkhERBTEwlVqPNZnECRJwpoje10uxe7JSoEV5lrkFf2CxXu+RWH1BQBAZkIq/t53GNpFxEDnYYHmSphChTvSeuDDwj0ut2vkCoxNy4BC1vD/dK12G2ySBLWcH1GIKPRw5CIiIgpysRod/tb/FjzceyAOlJdCguTVTYQrLSa8sX87Xtv/vVP79tJTuON/7+L94RPQP7ENZI1cKl2vVOMv3YehpLYaW0t+cd6mUGHZoEmIVevd7uO82YBThgtYfXwnjDYLMuPSMSylCyKVGqhD7CaiRNRysbgiIiIKAeEqNcJVaiTrw71+bpXFVKewusQmSXhs21f49NbJiNO6L4DcidXosaj/nSitrcaawh9RbTHjxsS2GJjQDlEqLZRyeb3PPWeqwUPbV2HPhdOOtg1FB/DCvg1YOuBe9IpJZYFF1AJI9hpAqoFk+ORymzkfkrwnIIuBINQ/jgQLFldERETN3OfHD7jdftpQiXKz8ZqKKwCIVusQrdbhqZ5ZsEsS5B5MA6yy1GJewVqnwuoSs92K+/Pfx/+GP4QUXdQ15UZEwU2yV0CqeRswvgW5TcSfukYCAORVf4RUmwAh+g1Iik4QhMZPYW4KXNCCiIiomSsyVDcYU2n23TLpgiB4VFgBQI3Vgq+LD9e7XbTbsOr4D7DYrL5Kzy2TzYJaq4X35CJqQpK9GlLNm4DxdQA2qOUyvDIoAa8MSoBaLgPs5yCVTwRspwKdaoNYXBERETVzveNTGoxpzHRDX/i5qhhSA0u0f1NyFDVWs99ykCQJ5eZqbC7Zi7l7VyJn7wp8fmYHzpmrYLXb/HZcIvqVVA0Y324gyAypagEke1WTpNRYnBbojsEAuJojLpcDGo1zXH1kMkCrbVys0QjU958zQQB0usbF1tYC7u5Jotc3HCuKkJuu+i+nyQTY3PwRunK/DcXqdBfzBgCzGbC6+Y+lN7Fa7cXXGQAsFkB0c9NLb2I1mst9xZtYUbwYXx+1GlAovI+1Wi++FvVRqQCl0vtYm+3i764+SuXF+EuxBsPFfmIwXN6Hq1i7/WJf82S/DcUqFBdfC+Die8Jo9E2sN+97jhGuY+t73zdmPOEYcZGHY8TNEQnQQ4Dh1yJGbrNBZb38+l4flwK9WQSEX/tfU40RJhOURhO0prp5W+VyiMqLr4NCki6+N2Si6zHlGsYIyWDAmdpyPLrrbZy3XD7Dt+vUHryu0uDFAX9Cp/BWF5ea99MYUV1RCqtkhQRAIcgRoQy7HMsxwvtYUXT+WThGeB/bhJ8jJMkOybAGMFoBhQCohIu3jTDYAIuEOLUMwqXfj3ELoDh38ee/1s8RYj3jiavPBu7ed1eTqI7KykoJgFR58SWt+zV6tPMTdDrXcYAkDR7sHBsXV39s377OsWlp9cd27eoc27Vr/bFpac6xffvWHxsX5xw7eHC9saJaLVkslsuxo0fXv9+ru9rdd7uPram5HDtlivvY0tLLsQ8+6D72+PHLsY895j52377LsTk57mN37Lgc+8IL7mM3b74c+8or7mPXrr0cu3y5+9g1ay7HrlnjPnb58suxa9e6j33llcuxmze7j33hhcuxO3a4j83JuRy7b5/72Mceuxx7/Lj72AcfvBxbWuo+dsqUy7E1Ne5j775bcuIulmPExS+dzjm2gTHCaTzhGHGRD8eIg6+/InX47/NS2nsLpD/Nutf9foNgjHhl/GCpyyc5UpdPcqQ1a991v18/jRFfjuguDd30N6m0tsIvY0SNaJQOVf4imTWq+mM5Rlz+ulIDY8QXq1ZdHlM4RlwUAp8jbH+Pk2xnO0pVJ9tLmAsJcyHVKOuJb8LPEZWABECqrKyUGsJpgURERC1Au4gYfH3H/ZjepR9a6SICnY7HwhRq3JJ0XcCOb7KL2HD2R9h8PD3QaK3F5tIdmP3TP2GHmzM7RBRSBEmSpEAnEWyqqqoQGRmJyqIiRES4+APEKT8QRRHr169H1l13QXnpVCqnBXof2wxO5zu5asqPWF19sZ9kZV3uJ65iOS2wcbFBPEa4jK3nfd+o8YRjxEWNGCPMNitqao0QzGZEqbWu723VhNMCgYv3uFq0bwPWF11e1dAqlyMyLBJv3vg7tNfHQmG2XO4rV48pjRwjbDYrnt/5X+SWuL75sU0ug6hSoGtEKhb1mopIq5v7gHk5RhRL1Xhg1zwAgLrWdX9Y0OMRtI9Mb/FjhLexoijiq7w8jB4z5mI/4RjhfWwTf46QTLmQKrMd0wINoh0Ry44BAKomtIVeeel8kA5C3OcQ5PHX/Dmi3vHExWeDqqoqRKakoLKy0nVtcAVec+WOXu/8RnYX580+PXXlQObL2CsH3sbGiiJsV/4RAZz/qDTEm1i1+nIn92WsSnX5jRaoWKWy7rVIvohVKC4PkL6Mlcs978O/xto0movPcZe7TOb5fr2JFQT/xALBERvMY4Qr9b3vr3U84RjhVaxaroA6LAIIazD0Ij+PEQAQo9fjiRvG4U/iaGwsOoga0Yy+cWnoFJGAaLUeckEGKJSX+4q7McWLMUISBIg6NUxa978PCRIkH44Rol3E2sItju/NWtc/y5rzW/FwXDqc3r0tcYzwNlYULxdLAMeIxsQ29ecIzQDAGglILlY21cmAS8WV7l4gLBmQXfU7asznCE/Gk0vve3dF/1VYXBEREVHARaq0iFRpcV/HgU12TIVMjlsSM7ChuMBt3E3xXRGm8OKDfwNqbRYcri5sMO6o4RTMdhE6eFGgEIUiWRSE6CWQzk8FUM+ZQ0UnCPo/QLi6sAoyvOaKiIiIWqxukW0Qrar/NJ5KpsDolD5QyFysHtxIMkGAWtbwGQSV4OFZBqIQJwhKQNEDQuyHgLLPVRu1gG4yhOh3IMjjApOgF1hcERERUYsVrdLjP33uR6Sy7rQ4lUyBf10/HVFuiq/GCFPoMCzhhgbjhiT0Q7jCi+l6RCFMkGkhKLtCiFoCIfary+2xX0AIezwkCiuA0wKJiIioBZMJMqTp4/F/A7KxvfwwNp4tgFWy48b4zhie2BNRSj2UMt9/XOoV3QXRyghcEF3fEFUr12B44gAo/HBsomAmyKOgVGkxpecUAIBSmRz0UwGvxHcsERERtWgyQYZYdThuTe6Dm+O7AQC0chXkPpwKeLUoZTiey3gEf9/3H5wzX3DaFq7Q45nuDyFKGe634xMFM7VCjXfGvhPoNBqFxRURERERAEEQEKZsmsUjBEFAijYBi3o+jlPGs8gr/QF2SBgQ2xOdwtMRodRDLvivuCMi/2BxRURERA06ZzKgsPo8vi0+Do1cgazW1yFarUOUmivZXYtoVQSiVRHoHtkRwMWzaEQtnSRJMIoX7z2pU+oguLofX5BicUVERERunay5gMmbV+KUocLR9uJPeRjRqiP+0W80YjVe3HuJXGJRRXSZUTQibMHFhWRq5tRArwqdMYbvZCIiIqrXOZMB9379X6fC6pKNZ47g+YKvUS2amz4xIqIgxOKKiIiI6lVQfgZnjdX1bv/0xD4YWnhxVS0aUWK6gLySAuSW7MYZYxkqLYZAp0VEAcBpgUREROSSaLNh7ckDbmNskoQjleeQpItooqyCywVLNf51+ENsO7cPEiRHe4/Itvhb198jXhMVuOSIqMnxzBURERG5JgACGr6QPJQuNvelCksNnt67DN+d2+tUWAHA3srjyC54DRcs9Z/1I6Lmh8UVERERuaSUyXFneo8GYmToGBnXRBkFlzO153Cg6kS924tqz2HX+Z+bMCMiCjQWV0RERFSvrtGJSAuLrnf7hHa9EaZQNWFGwcFsE/HJ6a0Nxn125ltUWmqaICMiCgYsroiIiKhecRo9/nvLvegcleDULgC4u20G/txjEPRKdWCSCyCrZEO11dhgXLW1FjU2E+ySvQmyImoe5DI57u56N+7uejfkstC6mTYXtCAiIiK3UvSReG/IRJTW1mBn2SloFArcmNgWYUo1IlWaQKcXEGqZEu30ydh5/rDbuFRdPLaW7sHIpH6IUbfMRT+IvKVRaPDBPR8EOo1GYXFFREREDYrV6BGr0aNLdGKgUwkKCpkct7caiDWn8tzG3ZqciSVHP0GiJhq3JF7fJLkRUeBwWiARERFRI0QodZjadlS924cm9Ea1aESx6Ty+Lt2NWmvLvh8YUUvAM1dEREREjRCm0GJM8g1I0sRgzcnN+MVwFgCQrInF2NY3IVkTg+cO/h8AwCbZ6izXTkSuGSwGhC0IAwDUzKmBXqUPcEaeY3FFRERE1Eh6hQaVlmqMb3MLkjQxkABUiDX4smgbdl/42VFQ9Y/pCo285a2qSNTSsLgiIiIiaiS1XIUB8T1w3/bnYK/nzJRapsRN8RmQCbwag6i547uciIiI6BrEKMMxu8skCBDqbFPKFHgu4wFEKnQByIyImhrPXBERERFdA61CjQGx3bC8/xx8eGozfqw4Apkgw8C4HrgtZSCiVeFQyviRi6gl4DudiIiI6BrpFBroFBr8scNYGK1mQLi44AWLKqKWhe94IiIiIh9Ry1VQc+EKohaLxRUREREREQUNuUyO0R1HOx6HEhZXREREREQUNDQKDb6898tAp9EoXC2QiIiIiIjIBwJeXL366qtIT0+HRqNBZmYmduzYUW/s/v37MW7cOKSnp0MQBCxevPia90lEREREROQLAS2uVq9ejezsbOTk5GD37t3o2bMnsrKyUFpa6jLeaDSiXbt2WLhwIZKSknyyTyIiIiIKDSZrNWqs51FjPQ+TtTrQ6ZCfGCwG6J/TQ/+cHgaLIdDpeCWgxdVLL72EGTNmYNq0aejatSuWLl0KnU6HZcuWuYzv168fXnzxRUyYMAFqtdon+yQiIiKi4FZrrcZpw158dmY+3jjye7xx5Hf47Mx8nDLsRa21KtDpkR8YRSOMojHQaXgtYAtaWCwW7Nq1C3PmzHG0yWQyDB8+HPn5+U26T7PZDLPZ7Pi+qurim1QURYii2KhcmrtLrwtfH3KH/YQ8wX5CnmJfaZlMthr8cG4NCiq++LVFACDgTPUBfFA9BxlRY3BD3ASo5WEA2E+agyt/d6IoQhR8/7v0pp9405cCVlydO3cONpsNiYmJTu2JiYk4dOhQk+5zwYIFmDdvXp32DRs2QKfTNSqXlmLjxo2BToFCAPsJeYL9hDzFvtISJaE9ZrjcYgCQi2/qtLOfhC6TzeR4vH79emjkGr8dy5N+YjR6fgaNS7EDmDNnDrKzsx3fV1VVITU1FSNHjkREREQAMwteoihi48aNGDFiBJRKZaDToSDFfkKeYD8JXiariJLaarxx8HtsLT4OQRAwOKkdZnS5AQkaPdSKpv19sa+0PLXWKnx25hmUmY65jYvXtMMdrXKgVUSwnzQDBosB2HvxcVZWFvQqvc+P4U0/uTSrzRMBK67i4uIgl8tRUlLi1F5SUlLvYhX+2qdarXZ5DZdSqeSbsgF8jcgT7CfkCfaT4GKxWbGnogT3bVkNq2R3tK8+sRcfn9qP/7vlXvSObQWlvOlv8Mm+0nKYIKFEPAQ00M1KxMMQFHanfsF+ErqUUtP9Hj3ZvzfHD9iCFiqVCn369EFubq6jzW63Izc3FwMGDAiafRIREbVElRYTHvz2I6fC6hLRbscft36ECkttADKjlkXyQySR/wR0WmB2djamTJmCvn37on///li8eDEMBgOmTZsGAJg8eTJatWqFBQsWALi4YMWBAwccj8+cOYOCggKEhYWhQ4cOHu2TiIiIGlZQXoQaq6Xe7RWWWhyqKEW8NqwJs6KWRiYoEKlMRqV41m1cpDIRcoFXuzQXMkGGwWmDHY9DSUB74fjx41FWVoann34axcXF6NWrF9atW+dYkOLkyZOQyS6/oEVFRejdu7fj+0WLFmHRokUYPHgw8vLyPNonERERNeznirIGY45UnsOg5HZNkA21VDp5JPrHjsfG4sVu4/rHjodOHtk0SZHfaZVa5E3NC3QajRLwEn/WrFmYNWuWy22XCqZL0tPTIUkNn/R1t08iIiJqWCt9wws6tdLzwyz5lyDI0DH8Rhyq2oxTxj0uY1prM9Ax/CYIIXaGg5qngBdXREREFHwGJKVDJZPDYre53K6VK9E7rlUTZ0UtkU4RidtbPYm9FRuw6/xHMNoqLrbLo9An5i70iMqCTsFCn4IDiysiIiKqI1yhRk6fkfjbD/9zuf3ZfqMQrqy70i6RP+gUUegXOw7dI4dDlMwAAKWghlYRARmvtWp2DBYD0l9OBwAUPlzol6XY/YW9kYiIKMSdqzX8ulKahCi1FkrZtS+PrlOqcFubLmgTFoUX9mzG3vPFAIBesSl4otct6BqVBG0T3+eKWjaZIIdeGRPoNKiJnDOeC3QKjcLiioiIKESdN9XimzO/4NW93+NYZTliNTpM7nw9JnTqiXjttf+nN1ylQWZCGl4fdDesdjsECJDLBMRrwiCX8foWIqKrsbgiIiIKQedNRuRs34Qvjh90tJXVGvDPH7fik2P7sfrWide8TPoFcy3WnjiA/+z/FudMBgBAojYMD3cfhFGpnRGl1l7T/omImhv+24mIiCgEHas871RYXemXqvNYfmAXLDZro/dfbTHj7UPbkbNrvaOwAoCS2ho8+cP/8N8ju1Ajmhu9fyKi5ojFFRERUYipsZjxxr7tbmP+e/hHVFpMjT+G1YylB/Pr3f6f/d/CINZ/k2EiopaIxRUREVGIsdhtKDbWuI2psphh9+DekPXJKzrm9vmi3Y7vS080ev9EDTFaq2CwXoDJZmg4mChI8JorIiKiEKORK9EuIgZ7y4vrjUnQhkEmCI0+xnmTscGYC+baRu+fqD411gs4XrMbuy+shclWg0hlIgbE/RZx6jbQK6ICnR41AZkgQ9+Uvo7HoYTFFRERUYjRKZV4oEd/fHb8QL0xf+jWFzFqXaOP0SsupcGY7jFJjd4/kSsXLGfxfuFsVFvLHW3nLWdw3LAbqdruuDP1SRZYLYBWqcUPM34IdBqNElqlIBEREQEAWukj8eeeN7rclpmYinEdelzTcumdIuMRp6l/OfdkXTjSwqIbvX+iqxmsF7DyxJNOhdWVTtXuw6biN2DmNEEKYiyuiIiIQlCkWoP7uvbDZ2MmY0SbjmgbEY2+Ca2w9JY78dqQsYjVNP6sFQDEqHV4d8gEhClUdbZFKDVYPmQCYt0UX0TeKjOfQKVY4jbmYNU3MNs5HZWCF6cFEhERhagotQZR8cl46aYxMNmsUAoyRGl8c+8puUyGjpFx2DjmAXxSuBfrTh0GAIxp0wW/SeuGOI3+mq7pIrqSzS5iz4X1DcZJsOOM8SAiIgc1QVYUKEbRiK6vdgUAHJh5ADrltf2zqCmxuCIiIgpx4So1wqH2+X4VMjkSdeH4Q+cb8Nt2vQAAkSrNNU03JHLFDhtEybOl/UWp8bcYoNAgSRJOVJ5wPA4lHB2JiIjILYVMhhiNDjEaHQsr8guFoEKKppNHsfHqdP8mQ3QNOEISERERUUAJggzdo4ZCaOCjaZQyCRGKuCbKish7LK6IiIiIKODUMj2GJk6vd7sMCvym1ePQcSl2CmK85oqIiIiIAk4t1yEjagTCFbHYXLrcaeXAFO11yEqaiVh1KgQupEJBjMUVEREREQUFjTwMnSNuQht9D5hsBljstdDJI6CUaaBTRAY6PaIGsbgiIiIioqAhCDLoFdHQK3iT6pZKEAR0je/qeBxKWFwREREREVHQ0Cl12P/g/kCn0Shc0IKIiIiIiMgHWFwRERERERH5AIsrIiIiIiIKGkbRiG6vdUO317rBKBoDnY5XeM0VEREREREFDUmScKDsgONxKOGZKyIiIiIiIh9gcUVEREREROQDnBZIREREzcJ5cw3O1lZgXdFPECUbBid0QaeIRMSowkLuXjlEFJpYXBEREVHIKzVV4aEf3sOR6mJH25oT25GgicDS/vehjT4GMoETdojIvzjKEBERUUg7b67Bn68qrC4pNVVh+vdv4rzFEIDMiKilYXFFREREIa3EVImfXRRWl1ywGPB92dEmzIiIroUgCEiLTENaZFrITenltEAiIiIKaRuK9jYY81XRHgxJ7IIwpaYJMiKia6FT6lD4SGGg02gUFldEREQEABDtNihl8kCn4TUr7A3G2CU7JITW/XL8xS7Zef2ZD5ltFbDZTbBLFshkaigEDVTyyECnRQHC4oqIiKgFq7aYUWUx4cNf9uKXqnJ0iIzDXe26I1KpQZhKjWqLGRIk6BUqyGXB+YF8cEJnvH98m9uYQQmdoVOoXW4z20TUWE0AAIUgR6RK5/McA61arEalWIlvz30Lo82Itvq26B3VG1q5Fmq569eF6pIkG8y2ChisZ1Fl+QVyQY0IVTqKDJtxtPJ9WO01iNVcj4zYvyBclQ6FrPn1JXKPxRUREVELVWUxYfXRPViw+2unczpfFO7H0sHjkHf2GD785SeIdhuGpLTHb9K6IUatg1IeXGe32oYlIFkThbOmCpfbNXIlRiR3h/yqszWizYpySzXeL/wGG4v3wGQT0TmiFaa3H4ZO4a2aTZFVaanEa8dew881Pzvatp7bihUnV2BK+hRcH3U9dIrm8bP6k2gzoNz8E3aWzketrcTRLkCB1mHDMDD5FXxf/CjKTbux+czvMSBpMRK0/SGXsXj1Vq1Yi5vfuRkA8M3Ub6BVagOckeeC819QRERE5HcHLpTguasKq9b6SLwwYAymbV6NP3/3Kb45+wvyS05gwY9fY9japThYUQqr3RawnF2JVYdhaeY0xKnD6mzTyJVY2n8aopXOxYNNsuNYTTEmbnsJH57KR6VohNkuYk9FIf686228dWwTKkVjU/0IflMtVtcprC6xSla8ffxtHDccD0BmocUu2VBu2oOtZx9yKqwAQIIVp2rWY8+5f6FPwvxLz8COktmw2KuaPtlmwC7ZsbNoJ3YW7YRdanjabzBhcUVERNQCXTDX4qWCb+q0z+pxI57dvQmnDBV1thmtIn7/9QqcN9c2QYbeaa2LwcqbZuG5Xr/FwPhOuCGuA2Z3HYPPBj+KLhGtoJQ7T9apsBjw2I/vwmQTXe7vw1PbUFhT2hSpN5pdssNqt7qNqRArXBZWV1p9ajWqRBYB7lhsFfih7BnAzXV75ea9MFnLEaXqAgCwSSaUGN1PV6Xmh9MCiYiIWiDRbsPuc2ec2nQKJdpFxNZpv1K1aMaO0pO4La2rv1P0iiAIiFWHYVRKBgYlXAdJkqBVqOpMBbzkhKEM5ZZqt/tcdiwXz/a8F+FBNiWp0lKFIlMJ8kq3wWIX0T2yM66P7gG9QguVTOUU+01Z3QL6aqdqT8Fit3idh9lmgsleC6O1BjJBBq1cD508DApZ8/t4WSOehslW1mDcL1WfIC3iN6g4dxAAUGz8BqlhWZDLuEplS9H8ej8RERE1SACglitgtF4+c5OkC8cvVeUNPndbcWHQFVdX0tezcMWV9leeajDmSM1ZmO1WhPsiKR85Zz6PZw+8jLOmy1PTtpXvhFKmxF86PYAuER2h+XWBCpvdBoPNs5snezv1qkq8gC/PrsbuC9/BKl3sQzq5HjfHj8bAuBEIU0R4tb9gVyX+4mFcIbSK5CtahF+/qKXgtEAiIqIWSK9U4fZ05wLJZLUiTNlwYRKlDq4zOY0Rpmj4TIJOrgqqj8WVlirMP7DYqbC6RLSLeOHQaygzXy6O5TI52uraNrhfhaCAUqb0OI8qsQKvHH0GO87nOQorADDaDFhX/AFWn3wDNc1smqFc8OzMk0LQwC5dPguYrLuZC1q0MCyuiIiIWiCdQoVZ3W+ETnH5Q3WRsQqJ2jCEKVRungmMa5fh7/T8bkBcJwgNlE5jUvogUhk8q+idNJ5Bsan+68DssOODU2tRa718TVzfmL6QC+5Xd+wf0x9auWcFs81uRf65XJSZz9Ybs69qJ0rM9U8tDUXx2t7w5GNzin4wSmt3ALhYkCXoBvg5Mwo2LK6IiIhaqERtOD7OmowuUQmOtjXH9uChHjfV+5zb2nRBrDp4Co7G0is0GJPSp97tkUodbmvVD4oguamyxWbBppKtDcb9cL4Apiuun9LKtZjUZlK98ZHKSIxrPQ4auWdnZgy2amw9t67BuI0ln8BorfFon6FAIejRSn+L2xgBCqSFj8ap6i8hQI7MxEVQyXgz4caK08UhThcX6DS8xmuuiIiIWiilXI7rohPwf8MnoNpiRrnZiDiNHlqFEvGaMLxQsBnFtRcXfdArVPh9pz74Q+fMZjEtMFypxcxOt0IuyLC2aCdsV1xzlKaLxwu9JyPWxdLugWKHHRa765UNr46TrljRTiPXIDMmEzGqGKw5tQZFpiIAgFyQo190P9zd+m5EK6M9zsMm2WCwuV8IBABKTGdgldyvZBhKVPJwXB/3BAxiESosB+tsF6BA/4S5KKz6BNHqbugRl41wZTrkMvdngck1vUqPsscbXkAkGLG4IiIiauFiNXrEavRIR4yj7fa0rhiYlA6D1QKb3Y5wlQYRKjU0cs+vzQl2USo9Zl03Gve1H4Y9F46j9tebCMepIxATRIUVAKhkKnQKb4fdFXvdxiVpEiC7amKSTqFDz6ieaKtvC4vdAptkg0qmglau9fiM1SUyyCBAcCrgXOerbnDaZajRKGIxKPnfKDPtxqELy1EtnoBCpkWqfiQ6RE6ATFAgXns9ZIIGajnPWLVULK6IiIioDrlMhnhtGOIDnYifhSk0CFNoMCK5V6BTcUsmyHBzfCZWn/rcbWEztlUWIpWu1zeMUF77Cn5KmRrXhWfgUPUet3GZMUOgVwRXgeoLGkUMUsOGI17TBxIu3kxbJQvnohXkEBTXXL366qtIT0+HRqNBZmYmduzY4Tb+gw8+QOfOnaHRaNCjRw989dVXTtunTp0KQRCcvkaNGuXPH4GIiIjIr/QKHf7Y/vf1bu8e0Rl9ojMgCP47Y6RT6DEmeaLbs1IauQ59Ym6CrIGFNEKZRhENrSIOWkUcCys/qBVrMeSdIRjyzhDUisF303J3Al5crV69GtnZ2cjJycHu3bvRs2dPZGVlobTU9Wo427Ztw8SJEzF9+nT8+OOPGDt2LMaOHYt9+/Y5xY0aNQpnz551fK1cubIpfhwiIiIiv9DINegf0xvzuz2OLhEdHe1xqhjc13YCHu40HRH1nLXypTh1IqakP+JyFUKdPAwPdchBmILT4qjx7JIdW05swZYTW7y+B1ugBXxa4EsvvYQZM2Zg2rRpAIClS5fiyy+/xLJly/DXv/61TvzLL7+MUaNG4fHHHwcAzJ8/Hxs3bsQrr7yCpUuXOuLUajWSkpKa5ocgIiIiagI6hRadItrjsU5/hMUuwg47FIICEcowyISm+Z+5Rq5Fl4he+HuX/+DHinwcqdkHGWToE30T2oV1QZgioslyIQo2AS2uLBYLdu3ahTlz5jjaZDIZhg8fjvz8fJfPyc/PR3Z2tlNbVlYWPv30U6e2vLw8JCQkIDo6GkOHDsWzzz6L2NhYl/s0m80wm82O76uqLt74ThRFiGLDK/O0RJdeF74+5A77CXmC/YQ8xb5ymRoqqK9Yic5mtcH26zVATUGADDohHAOihqNfxGAAgEquDkguV2M/CX1X/u5EUYQo+P536U0/8aYvBbS4OnfuHGw2GxITE53aExMTcejQIZfPKS4udhlfXFzs+H7UqFG466670LZtWxw7dgxPPvkkbr31VuTn50Mur3sKe8GCBZg3b16d9g0bNkCnC/17efjTxo0bA50ChQD2E/IE+wl5in2FPMF+ErpMNpPj8fr1671e1dIbnvQTo9Ho8f4CPi3QHyZMmOB43KNHD2RkZKB9+/bIy8vDsGHD6sTPmTPH6WxYVVUVUlNTMXLkSEREXPvKOs2RKIrYuHEjRowYAaWy+SzLS77FfkKeYD8hT7GvkCfYT0KfwWIAfr3rQFZWFvQqvc+P4U0/uTSrzRMBLa7i4uIgl8tRUlLi1F5SUlLv9VJJSUlexQNAu3btEBcXh6NHj7osrtRqNdTquiu9KJVKvikbwNeIPMF+Qp5gPyFPsa+QJ9hPQpdSuvx78/fv0ZP9e3P8gF5tqFKp0KdPH+Tm5jra7HY7cnNzMWDAAJfPGTBggFM8cPF0Xn3xAHD69GmUl5cjOTnZN4kTEREREZHf6JQ66JShd3lOwKcFZmdnY8qUKejbty/69++PxYsXw2AwOFYPnDx5Mlq1aoUFCxYAAB5++GEMHjwY//znPzFmzBisWrUKO3fuxBtvvAEAqKmpwbx58zBu3DgkJSXh2LFjmD17Njp06ICsrKyA/ZxERERERNQwvUoPw5OGQKfRKAEvrsaPH4+ysjI8/fTTKC4uRq9evbBu3TrHohUnT56ETHb5BNuNN96IFStW4KmnnsKTTz6Jjh074tNPP0X37t0BAHK5HD/99BPeffddVFRUICUlBSNHjsT8+fNdTv0jIiIiIiLyhYAXVwAwa9YszJo1y+W2vLy8Om333HMP7rnnHpfxWq0W69ev92V6REREREREDeId3oiIiIiIKGiYrCaMWTEGY1aMgclqavgJQSQozlwREREREREBgM1uw1dHvnI8DiU8c0VEREREROQDLK6IiIiIiIh8gMUVERERERGRD7C4IiIiIiIi8gEWV0RERERERD7A1QJdkCQJAFBVVRXgTIKXKIowGo2oqqqCUqkMdDoUpNhPyBPsJ+Qp9hXyBPtJ6DNYDMCvK7BXVVXBpvL9ioHe9JNLNcGlGsEdQfIkqoU5ffo0UlNTA50GEREREREFiVOnTqF169ZuY1hcuWC321FUVITw8HAIghDodIJSVVUVUlNTcerUKURERAQ6HQpS7CfkCfYT8hT7CnmC/YQ84U0/kSQJ1dXVSElJgUzm/qoqTgt0QSaTNViV0kUREREcuKhB7CfkCfYT8hT7CnmC/YQ84Wk/iYyM9Gh/XNCCiIiIiIjIB1hcERERERER+QCLK2oUtVqNnJwcqNXqQKdCQYz9hDzBfkKeYl8hT7CfkCf81U+4oAUREREREZEP8MwVERERERGRD7C4IiIiIiIi8gEWV0RERERERD7A4oqIiIiIiMgHWFxRvRYsWIB+/fohPDwcCQkJGDt2LA4fPuwUM2TIEAiC4PT1xz/+MUAZUyAsWbIEGRkZjpvwDRgwAP/73/8c200mE2bOnInY2FiEhYVh3LhxKCkpCWDGFCgN9RWOJ+TKwoULIQgCHnnkEUcbxxW6mqt+wjGF5s6dW6cPdO7c2bHdH2MJiyuq15YtWzBz5kx8//332LhxI0RRxMiRI2EwGJziZsyYgbNnzzq+XnjhhQBlTIHQunVrLFy4ELt27cLOnTsxdOhQ3HHHHdi/fz8A4NFHH8UXX3yBDz74AFu2bEFRURHuuuuuAGdNgdBQXwE4npCzH374Aa+//joyMjKc2jmu0JXq6ycAxxQCunXr5tQHvv32W8c2v4wlEpGHSktLJQDSli1bHG2DBw+WHn744cAlRUEpOjpaeuutt6SKigpJqVRKH3zwgWPbwYMHJQBSfn5+ADOkYHGpr0gSxxNyVl1dLXXs2FHauHGjU9/guEJXqq+fSBLHFJKknJwcqWfPni63+Wss4Zkr8lhlZSUAICYmxqn9/fffR1xcHLp37445c+bAaDQGIj0KAjabDatWrYLBYMCAAQOwa9cuiKKI4cOHO2I6d+6MNm3aID8/P4CZUqBd3Vcu4XhCl8ycORNjxoxxGj8AcFwhJ/X1k0s4ptCRI0eQkpKCdu3aYdKkSTh58iQA/40limvOmFoEu92ORx55BAMHDkT37t0d7ffeey/S0tKQkpKCn376CU888QQOHz6Mjz/+OIDZUlPbu3cvBgwYAJPJhLCwMHzyySfo2rUrCgoKoFKpEBUV5RSfmJiI4uLiwCRLAVVfXwE4ntBlq1atwu7du/HDDz/U2VZcXMxxhQC47ycAxxQCMjMz8c477+C6667D2bNnMW/ePAwaNAj79u3z21jC4oo8MnPmTOzbt89pnioA3H///Y7HPXr0QHJyMoYNG4Zjx46hffv2TZ0mBch1112HgoICVFZW4sMPP8SUKVOwZcuWQKdFQai+vtK1a1eOJwQAOHXqFB5++GFs3LgRGo0m0OlQkPKkn3BMoVtvvdXxOCMjA5mZmUhLS8OaNWug1Wr9ckxOC6QGzZo1C2vXrsXmzZvRunVrt7GZmZkAgKNHjzZFahQkVCoVOnTogD59+mDBggXo2bMnXn75ZSQlJcFisaCiosIpvqSkBElJSYFJlgKqvr7iCseTlmnXrl0oLS3F9ddfD4VCAYVCgS1btuDf//43FAoFEhMTOa5Qg/3EZrPVeQ7HFIqKikKnTp1w9OhRv31GYXFF9ZIkCbNmzcInn3yCr7/+Gm3btm3wOQUFBQCA5ORkP2dHwcxut8NsNqNPnz5QKpXIzc11bDt8+DBOnjzpdJ0NtVyX+oorHE9apmHDhmHv3r0oKChwfPXt2xeTJk1yPOa4Qg31E7lcXuc5HFOopqYGx44dQ3Jyst8+o3BaINVr5syZWLFiBT777DOEh4c75p9GRkZCq9Xi2LFjWLFiBUaPHo3Y2Fj89NNPePTRR3HzzTe7XA6Vmqc5c+bg1ltvRZs2bVBdXY0VK1YgLy8P69evR2RkJKZPn47s7GzExMQgIiICDz30EAYMGIAbbrgh0KlTE3PXVzie0CXh4eFO1/YCgF6vR2xsrKOd4wo11E84phAAPPbYY7j99tuRlpaGoqIi5OTkQC6XY+LEiX77jMLiiuq1ZMkSABdvwnel5cuXY+rUqVCpVNi0aRMWL14Mg8GA1NRUjBs3Dk899VQAsqVAKS0txeTJk3H27FlERkYiIyMD69evx4gRIwAA//rXvyCTyTBu3DiYzWZkZWXhtddeC3DWFAju+sqpU6c4npDHOK5QQ/gZhQDg9OnTmDhxIsrLyxEfH4+bbroJ33//PeLj4wH4ZywRJEmSfJE8ERERERFRS8ZrroiIiIiIiHyAxRUREREREZEPsLgiIiIiIiLyARZXREREREREPsDiioiIiIiIyAdYXBEREREREfkAiysiIiIiIiIfYHFFRERERETkAyyuiIiIAujw4cNISkpCdXV1o/dx4MABtG7dGgaDwYeZERGRt1hcERFRk8vPz4dcLseYMWMCnUrAzZkzBw899BDCw8MBAIWFhbj55puh1+tx8803o7Cw0Cn+tttuw0cffeTU1rVrV9xwww146aWXmiptIiJygcUVERE1ubfffhsPPfQQvvnmGxQVFQU0F4vFErBjnzx5EmvXrsXUqVMdbX/5y1/QqlUrFBQUIDk5GY899phj2+rVqyGTyTBu3Lg6+5o2bRqWLFkCq9XaFKkTEZELLK6IiKhJ1dTUYPXq1fjTn/6EMWPG4J133qkT88UXX6Bfv37QaDSIi4vDnXfe6dhmNpvxxBNPIDU1FWq1Gh06dMDbb78NAHjnnXcQFRXltK9PP/0UgiA4vp87dy569eqFt956C23btoVGowEArFu3DjfddBOioqIQGxuL2267DceOHXPa1+nTpzFx4kTExMRAr9ejb9++2L59OwoLCyGTybBz506n+MWLFyMtLQ12u93la7FmzRr07NkTrVq1crQdPHgQU6ZMQceOHTF16lQcPHgQAFBRUYGnnnoKr776qst9jRgxAufPn8eWLVtcbiciIv9jcUVERE1qzZo16Ny5M6677jr87ne/w7JlyyBJkmP7l19+iTvvvBOjR4/Gjz/+iNzcXPTv39+xffLkyVi5ciX+/e9/4+DBg3j99dcRFhbmVQ5Hjx7FRx99hI8//hgFBQUAAIPBgOzsbOzcuRO5ubmQyWS48847HYVRTU0NBg8ejDNnzuDzzz/Hnj17MHv2bNjtdqSnp2P48OFYvny503GWL1+OqVOnQiZz/ed269at6Nu3r1Nbz549sWnTJtjtdmzYsAEZGRkAgMcffxwzZ85Eamqqy32pVCr06tULW7du9eq1ICIiH5KIiIia0I033igtXrxYkiRJEkVRiouLkzZv3uzYPmDAAGnSpEkun3v48GEJgLRx40aX25cvXy5FRkY6tX3yySfSlX/ucnJyJKVSKZWWlrrNs6ysTAIg7d27V5IkSXr99del8PBwqby83GX86tWrpejoaMlkMkmSJEm7du2SBEGQjh8/Xu8xevbsKT3zzDNObadPn5bGjBkjpaamSmPGjJFOnz4tbdmyRerbt69UXl4u3XPPPVLbtm2lBx54QDKbzU7PvfPOO6WpU6e6/bmIiMh/eOaKiIiazOHDh7Fjxw5MnDgRAKBQKDB+/HjHtD4AKCgowLBhw1w+v6CgAHK5HIMHD76mPNLS0hAfH+/UduTIEUycOBHt2rVDREQE0tPTAVy8LurSsXv37o2YmBiX+xw7dizkcjk++eQTABenKN5yyy2O/bhSW1vrmJZ4SatWrbB27VrH9VhxcXF48MEHsXTpUjz77LMIDw/H4cOHceTIEbz++utOz9VqtTAajd68FERE5EMsroiIqMm8/fbbsFqtSElJgUKhgEKhwJIlS/DRRx+hsrISwMUCoT7utgGATCZzmmIIAKIo1onT6/V12m6//XacP38eb775JrZv347t27cDuLzgRUPHVqlUmDx5MpYvXw6LxYIVK1bgvvvuc/ucuLg4XLhwwW3Mc889h5EjR6JPnz7Iy8vDuHHjoFQqcddddyEvL88p9vz583WKRiIiajosroiIqElYrVa89957+Oc//4mCggLH1549e5CSkoKVK1cCADIyMpCbm+tyHz169IDdbq930Yb4+HhUV1c73e/p0jVV7pSXl+Pw4cN46qmnMGzYMHTp0qVO0ZORkYGCggKcP3++3v384Q9/wKZNm/Daa6/BarXirrvucnvc3r1748CBA/VuP3jwIFasWIH58+cDAGw2m6NYFEURNpvNKX7fvn3o3bu322MSEZH/sLgiIqImsXbtWly4cAHTp09H9+7dnb7GjRvnmBqYk5ODlStXIicnBwcPHsTevXvx/PPPAwDS09MxZcoU3Hffffj0009x/Phx5OXlYc2aNQCAzMxM6HQ6PPnkkzh27BhWrFjhcjXCq0VHRyM2NhZvvPEGjh49iq+//hrZ2dlOMRMnTkRSUhLGjh2L7777Dr/88gs++ugj5OfnO2K6dOmCG264AU888QQmTpzY4NmurKws5Ofn1ymSAECSJNx///3417/+5TjTNnDgQLz55ps4ePAg3nvvPQwcONARX1hYiDNnzmD48OEN/rxEROQfLK6IiKhJvP322xg+fDgiIyPrbBs3bhx27tyJn376CUOGDMEHH3yAzz//HL169cLQoUOxY8cOR+ySJUtw991348EHH0Tnzp0xY8YMx5mqmJgY/Pe//8VXX32FHj16YOXKlZg7d26DuclkMqxatQq7du1C9+7d8eijj+LFF190ilGpVNiwYQMSEhIwevRo9OjRAwsXLoRcLneKmz59OiwWS4NTAgHg1ltvhUKhwKZNm+pse+ONN5CYmIjbbrvN0TZ37lyYTCZkZmaiQ4cOmDlzpmPbypUrMXLkSKSlpTV4XCIi8g9BunpyOhERETXa/Pnz8cEHH+Cnn37yKP7VV1/F559/jvXr1zf6mBaLBR07dsSKFSuczmYREVHTUgQ6ASIiouagpqYGhYWFeOWVV/Dss896/LwHHngAFRUVqK6uRnh4eKOOffLkSTz55JMsrIiIAoxnroiIiHxg6tSpWLlyJcaOHYsVK1bUmS5IRETNH4srIiIiIiIiH+CCFkRERERERD7A4oqIiIiIiMgHWFwRERERERH5AIsrIiIiIiIiH2BxRURERERE5AMsroiIiIiIiHyAxRUREREREZEPsLgiIiIiIiLygf8H+8XYHmkGsWoAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualizzare graficamente il trade-off tra Accuracy e F1 Score\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "plt.figure(figsize=(10, 6))\n",
    "sns.scatterplot(\n",
    "    data=results_df, \n",
    "    x=\"final_accuracy\", \n",
    "    y=\"f1_score\", \n",
    "    hue=\"tradeoff_score\", \n",
    "    size=\"tradeoff_score\", \n",
    "    palette=\"viridis\", \n",
    "    legend=\"brief\"\n",
    ")\n",
    "plt.title(\"Trade-off tra Accuracy e F1 Score\")\n",
    "plt.xlabel(\"Accuracy (%)\")\n",
    "plt.ylabel(\"F1 Score\")\n",
    "plt.axhline(y=best_combination[\"f1_score\"], color=\"r\", linestyle=\"--\", label=\"Miglior F1 Score\")\n",
    "plt.axvline(x=best_combination[\"final_accuracy\"], color=\"g\", linestyle=\"--\", label=\"Miglior Accuracy\")\n",
    "plt.legend()\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save({\n",
    "    'model': model,\n",
    "    'model_state_dict': model.state_dict(),\n",
    "    'optimizer_state_dict': optimizer.state_dict(),\n",
    "}, 'basic_CNN_edges_38_accuracy.pth')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".env",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}