ResNet_for_CIFAR10.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "1a7db511cf154721abff63bd6dbb5848": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_06f5de90e8a24e4f9925f8c23be01b2f",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_1157466b641f40c092036c44bdb111bc",
              "IPY_MODEL_096eba2683d048ccb045c07b1ed1fd5f"
            ]
          },
          "model_module_version": "1.5.0"
        },
        "06f5de90e8a24e4f9925f8c23be01b2f": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          },
          "model_module_version": "1.2.0"
        },
        "1157466b641f40c092036c44bdb111bc": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_13f22dfb9983491aad334d499c4487e4",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "info",
            "max": 1,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 1,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_43516b00ffbc4e2aadbd750a01b55f58"
          },
          "model_module_version": "1.5.0"
        },
        "096eba2683d048ccb045c07b1ed1fd5f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_5df1e446403848e4a4e1eb66f828dd7f",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 170500096/? [00:20&lt;00:00, 52680965.79it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_99f9cd2e2faa47278e32fef4b0af9422"
          },
          "model_module_version": "1.5.0"
        },
        "13f22dfb9983491aad334d499c4487e4": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          },
          "model_module_version": "1.5.0"
        },
        "43516b00ffbc4e2aadbd750a01b55f58": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          },
          "model_module_version": "1.2.0"
        },
        "5df1e446403848e4a4e1eb66f828dd7f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          },
          "model_module_version": "1.5.0"
        },
        "99f9cd2e2faa47278e32fef4b0af9422": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          },
          "model_module_version": "1.2.0"
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/cestclaire/image_classification/blob/main/ResNet_for_CIFAR10.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bVJMlQFxXoVW"
      },
      "source": [
        "### **CIFAR10 Dataset**\n",
        "\n",
        "* 작성자: 나동빈(Dongbin Na)\n",
        "* CNN을 이용해 CIFAR10 데이터셋에 대하여 학습/평가를 진행합니다.\n",
        "* 코드 실행 전에 **[런타임]** → **[런타임 유형 변경]** → 유형을 **GPU**로 설정해야 합니다."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fmxd2hYMX2Hx"
      },
      "source": [
        "#### 필요한 라이브러리 불러오기"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Kn7IHii5WMI1"
      },
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import torch.optim as optim\n",
        "\n",
        "import torchvision\n",
        "import torchvision.transforms as transforms\n",
        "\n",
        "from matplotlib import pyplot as plt\n",
        "import seaborn as sn\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import os\n",
        "\n",
        "device = 'cuda'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pE7mEvkKX89B"
      },
      "source": [
        "#### 데이터셋(Dataset) 다운로드 및 불러오기"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PCj5ub0sX904",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 102,
          "referenced_widgets": [
            "1a7db511cf154721abff63bd6dbb5848",
            "06f5de90e8a24e4f9925f8c23be01b2f",
            "1157466b641f40c092036c44bdb111bc",
            "096eba2683d048ccb045c07b1ed1fd5f",
            "13f22dfb9983491aad334d499c4487e4",
            "43516b00ffbc4e2aadbd750a01b55f58",
            "5df1e446403848e4a4e1eb66f828dd7f",
            "99f9cd2e2faa47278e32fef4b0af9422"
          ]
        },
        "outputId": "61b6b0e8-4de1-400a-a5b0-d3dc77ee675e"
      },
      "source": [
        "import torchvision\n",
        "import torchvision.transforms as transforms\n",
        "\n",
        "normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])\n",
        "    \n",
        "# 일반적으로 많이 사용되는 입력 전처리(preprocessing) 테크닉\n",
        "transform_train = transforms.Compose([\n",
        "    transforms.RandomCrop(32, padding=4), # 패딩 이후 랜덤하게 잘라서 뽑기\n",
        "    transforms.RandomHorizontalFlip(), # 좌우 반전 (MNIST 말고 CIFAR10에서는 효과적)\n",
        "    transforms.ToTensor(),\n",
        "    normalize, # 실험 결과, 입력 정규화(input normalization)가 성능에 크게 영향을 미치지는 않음\n",
        "])\n",
        "\n",
        "transform_test = transforms.Compose([\n",
        "    transforms.ToTensor(),\n",
        "    normalize,\n",
        "])\n",
        "\n",
        "train_dataset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transform_train)\n",
        "test_dataset = torchvision.datasets.CIFAR10(root='./data', train=False, download=True, transform=transform_test)\n",
        "\n",
        "train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=128, shuffle=True, num_workers=4)\n",
        "test_loader = torch.utils.data.DataLoader(test_dataset, batch_size=100, shuffle=False, num_workers=4)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data/cifar-10-python.tar.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "1a7db511cf154721abff63bd6dbb5848",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=1.0, bar_style='info', max=1.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting ./data/cifar-10-python.tar.gz to ./data\n",
            "Files already downloaded and verified\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "H4d386S9dHDG"
      },
      "source": [
        "#### 학습(Training) 및 평가(Testing) 함수 정의"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9NhdDmhodIfk"
      },
      "source": [
        "def train(net, epoch, optimizer, criterion, train_loader):\n",
        "    print('[ Train epoch: %d ]' % epoch)\n",
        "    net.train()\n",
        "    train_loss = 0\n",
        "    correct = 0\n",
        "    total = 0\n",
        "    for batch_idx, (inputs, targets) in enumerate(train_loader):\n",
        "        inputs, targets = inputs.to(device), targets.to(device)\n",
        "        optimizer.zero_grad()\n",
        "\n",
        "        benign_outputs = net(inputs)\n",
        "        loss = criterion(benign_outputs, targets)\n",
        "        loss.backward()\n",
        "\n",
        "        optimizer.step()\n",
        "        train_loss += loss.item()\n",
        "        _, predicted = benign_outputs.max(1)\n",
        "\n",
        "        total += targets.size(0)\n",
        "        correct += predicted.eq(targets).sum().item()\n",
        "\n",
        "    print('Train accuarcy:', 100. * correct / total)\n",
        "    print('Train average loss:', train_loss / total)\n",
        "    return (100. * correct / total, train_loss / total)\n",
        "\n",
        "\n",
        "def evaluate(net, epoch, file_name, data_loader, info):\n",
        "    print('[ Evaluate epoch: %d ]' % epoch)\n",
        "    print(\"Dataset:\", info)\n",
        "\n",
        "    net.eval() # Dropout을 적용하는 경우 필수임\n",
        "    test_loss = 0\n",
        "    correct = 0\n",
        "    total = 0\n",
        "\n",
        "    for batch_idx, (inputs, targets) in enumerate(data_loader):\n",
        "        inputs, targets = inputs.to(device), targets.to(device)\n",
        "        total += targets.size(0)\n",
        "\n",
        "        outputs = net(inputs)\n",
        "        test_loss += criterion(outputs, targets).item()\n",
        "\n",
        "        _, predicted = outputs.max(1)\n",
        "        correct += predicted.eq(targets).sum().item()\n",
        "\n",
        "    print('Accuarcy:', 100. * correct / total)\n",
        "    print('Average loss:', test_loss / total)\n",
        "    return (100. * correct / total, test_loss / total)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FYVmZYLVdLDP"
      },
      "source": [
        "#### 혼동행렬(Confusion Matrix) 함수 정의\n",
        "\n",
        "* 학습이 완료된 모델을 불러와 혼동 행렬을 NumPy 객체로 반환합니다.\n",
        "* 별도로 데이터 로더(data loader)를 파라미터로 이용해 해당 데이터셋에 대하여 평가를 진행합니다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6tueW8Y7dLnE"
      },
      "source": [
        "def get_confusion_matrix(net, num_classes, data_loader):\n",
        "    confusion_matrix = torch.zeros(num_classes, num_classes)\n",
        "\n",
        "    net.eval() # Dropout을 적용하는 경우 필수임\n",
        "\n",
        "    for batch_idx, (inputs, targets) in enumerate(data_loader):\n",
        "        inputs, targets = inputs.to(device), targets.to(device)\n",
        "\n",
        "        outputs = net(inputs)\n",
        "        _, predicted = outputs.max(1)\n",
        "\n",
        "        for t, p in zip(targets.view(-1), predicted.view(-1)):\n",
        "            confusion_matrix[t.long(), p.long()] += 1\n",
        "\n",
        "    return confusion_matrix"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "l9CgN4tfYBHx"
      },
      "source": [
        "#### ResNet 모델 정의\n",
        "\n",
        "* [ResNet 논문](https://arxiv.org/pdf/1512.03385.pdf)에서 제안된 CIFAR-10 전용 아키텍처를 따릅니다.\n",
        "  * CIFAR-10 데이터셋을 위한 아키텍처의 파라미터가 훨씬 적습니다.\n",
        "* ImageNet 데이터셋을 위한 아키텍처와는 차이가 있습니다."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pRNWfoomX3O-"
      },
      "source": [
        "class LambdaLayer(nn.Module):\n",
        "    def __init__(self, lambd):\n",
        "        super(LambdaLayer, self).__init__()\n",
        "        self.lambd = lambd\n",
        "\n",
        "    def forward(self, x):\n",
        "        return self.lambd(x)\n",
        "\n",
        "\n",
        "# ResNet을 위한 BasicBlock 클래스 정의\n",
        "class BasicBlock(nn.Module):\n",
        "    def __init__(self, in_planes, planes, stride=1):\n",
        "        super(BasicBlock, self).__init__()\n",
        "\n",
        "        # 3x3 필터를 사용 (너비와 높이를 줄일 때는 stride 값 조절)\n",
        "        self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)\n",
        "        self.bn1 = nn.BatchNorm2d(planes) # 배치 정규화(batch normalization)\n",
        "\n",
        "        # 3x3 필터를 사용 (패딩을 1만큼 주기 때문에 너비와 높이가 동일)\n",
        "        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1, padding=1, bias=False)\n",
        "        self.bn2 = nn.BatchNorm2d(planes) # 배치 정규화(batch normalization)\n",
        "\n",
        "        self.shortcut = nn.Sequential() # identity인 경우\n",
        "        if stride != 1: # stride가 1이 아니라면, Identity mapping이 아닌 경우\n",
        "            self.shortcut = LambdaLayer(lambda x: F.pad(x[:, :, ::2, ::2], (0, 0, 0, 0, planes // 4, planes // 4), \"constant\", 0))\n",
        "\n",
        "    def forward(self, x):\n",
        "        out = F.relu(self.bn1(self.conv1(x)))\n",
        "        out = self.bn2(self.conv2(out))\n",
        "        out += self.shortcut(x) # (핵심) skip connection\n",
        "        out = F.relu(out)\n",
        "        return out\n",
        "\n",
        "\n",
        "# ResNet 클래스 정의\n",
        "class ResNet(nn.Module):\n",
        "    def __init__(self, block, num_blocks, num_classes=10):\n",
        "        super(ResNet, self).__init__()\n",
        "        self.in_planes = 16\n",
        "\n",
        "        # 16개의 3x3 필터(filter)를 사용\n",
        "        self.conv1 = nn.Conv2d(3, 16, kernel_size=3, stride=1, padding=1, bias=False)\n",
        "        self.bn1 = nn.BatchNorm2d(16)\n",
        "        # 첫 레이어를 제외하고는 너비와 높이를 줄이기 위해 stride를 2로 설정\n",
        "        self.layer1 = self._make_layer(block, 16, num_blocks[0], stride=1)\n",
        "        self.layer2 = self._make_layer(block, 32, num_blocks[1], stride=2)\n",
        "        self.layer3 = self._make_layer(block, 64, num_blocks[2], stride=2)\n",
        "        self.linear = nn.Linear(64, num_classes)\n",
        "\n",
        "    def _make_layer(self, block, planes, num_blocks, stride):\n",
        "        strides = [stride] + [1] * (num_blocks - 1)\n",
        "        layers = []\n",
        "        for stride in strides:\n",
        "            layers.append(block(self.in_planes, planes, stride))\n",
        "            self.in_planes = planes # 다음 레이어를 위해 채널 수 변경\n",
        "        return nn.Sequential(*layers)\n",
        "\n",
        "    def forward(self, x):\n",
        "        out = F.relu(self.bn1(self.conv1(x)))\n",
        "        out = self.layer1(out)\n",
        "        out = self.layer2(out)\n",
        "        out = self.layer3(out)\n",
        "        out = F.avg_pool2d(out, out.size()[3])\n",
        "        out = out.view(out.size(0), -1)\n",
        "        out = self.linear(out)\n",
        "        return out\n",
        "\n",
        "\n",
        "def ResNet20():\n",
        "    return ResNet(BasicBlock, [3, 3, 3])\n",
        "\n",
        "\n",
        "def ResNet32():\n",
        "    return ResNet(BasicBlock, [5, 5, 5])\n",
        "\n",
        "\n",
        "def ResNet44():\n",
        "    return ResNet(BasicBlock, [7, 7, 7])\n",
        "\n",
        "\n",
        "def ResNet56():\n",
        "    return ResNet(BasicBlock, [9, 9, 9])\n",
        "\n",
        "\n",
        "def ResNet110():\n",
        "    return ResNet(BasicBlock, [18, 18, 18])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IhvQewGicnCv"
      },
      "source": [
        "#### 결과 확인: ResNet20 결과 분석"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bI-tFeKpbmfZ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6bde2e47-1ca4-4c7f-9708-5bba87257932"
      },
      "source": [
        "# 학습 과정 중에 천천히 학습률(learning rate) 감소 시키기\n",
        "def adjust_learning_rate(optimizer, epoch):\n",
        "    lr = learning_rate\n",
        "    if epoch >= 100:\n",
        "        lr /= 10\n",
        "    if epoch >= 150:\n",
        "        lr /= 10\n",
        "    for param_group in optimizer.param_groups:\n",
        "        param_group['lr'] = lr\n",
        "\n",
        "\n",
        "net = ResNet20()\n",
        "net = net.to(device)\n",
        "\n",
        "epoch = 200\n",
        "learning_rate = 0.1\n",
        "file_name = \"ResNet20_CIFAR10.pt\"\n",
        "\n",
        "criterion = nn.CrossEntropyLoss()\n",
        "optimizer = optim.SGD(net.parameters(), lr=learning_rate, momentum=0.9, weight_decay=0.0002)\n",
        "\n",
        "train_result = []\n",
        "test_result = []\n",
        "\n",
        "train_result.append(evaluate(net, 0, file_name, train_loader, \"Train\"))\n",
        "test_result.append(evaluate(net, 0, file_name, test_loader, \"Test\"))\n",
        "\n",
        "for i in range(epoch):\n",
        "    adjust_learning_rate(optimizer, epoch)\n",
        "    train(net, i, optimizer, criterion, train_loader)\n",
        "    train_acc, train_loss = evaluate(net, i + 1, file_name, train_loader, \"Train\")\n",
        "    test_acc, test_loss = evaluate(net, i + 1, file_name, test_loader, \"Test\")\n",
        "\n",
        "    state = {\n",
        "        'net': net.state_dict()\n",
        "    }\n",
        "    if not os.path.isdir('checkpoint'):\n",
        "        os.mkdir('checkpoint')\n",
        "    torch.save(state, './checkpoint/' + file_name)\n",
        "    print('Model Saved!')\n",
        "\n",
        "    train_result.append((train_acc, train_loss))\n",
        "    test_result.append((test_acc, test_loss))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[ Evaluate epoch: 0 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 10.0\n",
            "Average loss: 0.018115994958877564\n",
            "[ Evaluate epoch: 0 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 10.0\n",
            "Average loss: 0.023148727083206175\n",
            "[ Train epoch: 0 ]\n",
            "Train accuarcy: 31.914\n",
            "Train average loss: 0.014278708560466766\n",
            "[ Evaluate epoch: 1 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 38.676\n",
            "Average loss: 0.012749296979904175\n",
            "[ Evaluate epoch: 1 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 39.76\n",
            "Average loss: 0.016184935748577118\n",
            "Model Saved!\n",
            "[ Train epoch: 1 ]\n",
            "Train accuarcy: 43.962\n",
            "Train average loss: 0.01187758123397827\n",
            "[ Evaluate epoch: 2 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 43.654\n",
            "Average loss: 0.012331243920326232\n",
            "[ Evaluate epoch: 2 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 45.97\n",
            "Average loss: 0.014994074857234954\n",
            "Model Saved!\n",
            "[ Train epoch: 2 ]\n",
            "Train accuarcy: 51.088\n",
            "Train average loss: 0.010549728014469146\n",
            "[ Evaluate epoch: 3 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 53.728\n",
            "Average loss: 0.009957269558906556\n",
            "[ Evaluate epoch: 3 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 53.67\n",
            "Average loss: 0.0126382932305336\n",
            "Model Saved!\n",
            "[ Train epoch: 3 ]\n",
            "Train accuarcy: 56.558\n",
            "Train average loss: 0.009404542071819306\n",
            "[ Evaluate epoch: 4 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 58.52\n",
            "Average loss: 0.008971089273691177\n",
            "[ Evaluate epoch: 4 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 57.04\n",
            "Average loss: 0.011907819229364396\n",
            "Model Saved!\n",
            "[ Train epoch: 4 ]\n",
            "Train accuarcy: 60.498\n",
            "Train average loss: 0.008590157047510148\n",
            "[ Evaluate epoch: 5 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 60.652\n",
            "Average loss: 0.008596913784742356\n",
            "[ Evaluate epoch: 5 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 59.5\n",
            "Average loss: 0.011344692957401276\n",
            "Model Saved!\n",
            "[ Train epoch: 5 ]\n",
            "Train accuarcy: 63.646\n",
            "Train average loss: 0.00792343925356865\n",
            "[ Evaluate epoch: 6 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 61.908\n",
            "Average loss: 0.008189198353290558\n",
            "[ Evaluate epoch: 6 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 62.02\n",
            "Average loss: 0.010745218205451965\n",
            "Model Saved!\n",
            "[ Train epoch: 6 ]\n",
            "Train accuarcy: 66.156\n",
            "Train average loss: 0.007399168919324875\n",
            "[ Evaluate epoch: 7 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 65.704\n",
            "Average loss: 0.007534655032157898\n",
            "[ Evaluate epoch: 7 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 64.26\n",
            "Average loss: 0.010295533007383346\n",
            "Model Saved!\n",
            "[ Train epoch: 7 ]\n",
            "Train accuarcy: 68.39\n",
            "Train average loss: 0.006968320455551147\n",
            "[ Evaluate epoch: 8 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 68.608\n",
            "Average loss: 0.006937050969600678\n",
            "[ Evaluate epoch: 8 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 67.52\n",
            "Average loss: 0.00922619069814682\n",
            "Model Saved!\n",
            "[ Train epoch: 8 ]\n",
            "Train accuarcy: 69.982\n",
            "Train average loss: 0.006619315494298935\n",
            "[ Evaluate epoch: 9 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 67.468\n",
            "Average loss: 0.007218145072460175\n",
            "[ Evaluate epoch: 9 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 66.14\n",
            "Average loss: 0.009827256613969803\n",
            "Model Saved!\n",
            "[ Train epoch: 9 ]\n",
            "Train accuarcy: 71.32\n",
            "Train average loss: 0.006307624521255493\n",
            "[ Evaluate epoch: 10 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 71.95\n",
            "Average loss: 0.006211489001512528\n",
            "[ Evaluate epoch: 10 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 70.72\n",
            "Average loss: 0.008176967626810075\n",
            "Model Saved!\n",
            "[ Train epoch: 10 ]\n",
            "Train accuarcy: 72.808\n",
            "Train average loss: 0.0060018265187740324\n",
            "[ Evaluate epoch: 11 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 72.33\n",
            "Average loss: 0.0061329620790481565\n",
            "[ Evaluate epoch: 11 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 71.52\n",
            "Average loss: 0.008190153354406357\n",
            "Model Saved!\n",
            "[ Train epoch: 11 ]\n",
            "Train accuarcy: 74.286\n",
            "Train average loss: 0.005739732128381729\n",
            "[ Evaluate epoch: 12 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 72.258\n",
            "Average loss: 0.006155968610048294\n",
            "[ Evaluate epoch: 12 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 71.93\n",
            "Average loss: 0.008248225104808807\n",
            "Model Saved!\n",
            "[ Train epoch: 12 ]\n",
            "Train accuarcy: 75.1\n",
            "Train average loss: 0.0055432010781764985\n",
            "[ Evaluate epoch: 13 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 74.228\n",
            "Average loss: 0.005721332018375397\n",
            "[ Evaluate epoch: 13 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 72.7\n",
            "Average loss: 0.00777958419919014\n",
            "Model Saved!\n",
            "[ Train epoch: 13 ]\n",
            "Train accuarcy: 76.088\n",
            "Train average loss: 0.00535406433403492\n",
            "[ Evaluate epoch: 14 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 73.536\n",
            "Average loss: 0.005790895458459854\n",
            "[ Evaluate epoch: 14 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 71.9\n",
            "Average loss: 0.008079302901029586\n",
            "Model Saved!\n",
            "[ Train epoch: 14 ]\n",
            "Train accuarcy: 76.848\n",
            "Train average loss: 0.005182589043974876\n",
            "[ Evaluate epoch: 15 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 75.782\n",
            "Average loss: 0.005363601488471031\n",
            "[ Evaluate epoch: 15 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 74.11\n",
            "Average loss: 0.007547867637872696\n",
            "Model Saved!\n",
            "[ Train epoch: 15 ]\n",
            "Train accuarcy: 77.752\n",
            "Train average loss: 0.004981698431968689\n",
            "[ Evaluate epoch: 16 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 77.912\n",
            "Average loss: 0.004908369323015213\n",
            "[ Evaluate epoch: 16 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 76.76\n",
            "Average loss: 0.006792174550890923\n",
            "Model Saved!\n",
            "[ Train epoch: 16 ]\n",
            "Train accuarcy: 78.472\n",
            "Train average loss: 0.004836769942641258\n",
            "[ Evaluate epoch: 17 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 75.72\n",
            "Average loss: 0.005424226376414299\n",
            "[ Evaluate epoch: 17 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 74.12\n",
            "Average loss: 0.007684176367521286\n",
            "Model Saved!\n",
            "[ Train epoch: 17 ]\n",
            "Train accuarcy: 78.732\n",
            "Train average loss: 0.004725069047808647\n",
            "[ Evaluate epoch: 18 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 79.144\n",
            "Average loss: 0.004698432938456535\n",
            "[ Evaluate epoch: 18 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 77.41\n",
            "Average loss: 0.006597341632843017\n",
            "Model Saved!\n",
            "[ Train epoch: 18 ]\n",
            "Train accuarcy: 79.626\n",
            "Train average loss: 0.0045665877491235735\n",
            "[ Evaluate epoch: 19 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 77.834\n",
            "Average loss: 0.004948977710604667\n",
            "[ Evaluate epoch: 19 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 75.98\n",
            "Average loss: 0.007066182667016983\n",
            "Model Saved!\n",
            "[ Train epoch: 19 ]\n",
            "Train accuarcy: 80.18\n",
            "Train average loss: 0.004473092684149742\n",
            "[ Evaluate epoch: 20 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 79.542\n",
            "Average loss: 0.004572370000481605\n",
            "[ Evaluate epoch: 20 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 78.18\n",
            "Average loss: 0.006479631692171097\n",
            "Model Saved!\n",
            "[ Train epoch: 20 ]\n",
            "Train accuarcy: 80.656\n",
            "Train average loss: 0.004362159276604652\n",
            "[ Evaluate epoch: 21 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 77.322\n",
            "Average loss: 0.005058692629337311\n",
            "[ Evaluate epoch: 21 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 75.14\n",
            "Average loss: 0.007485232675075531\n",
            "Model Saved!\n",
            "[ Train epoch: 21 ]\n",
            "Train accuarcy: 81.12\n",
            "Train average loss: 0.004269265319108963\n",
            "[ Evaluate epoch: 22 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 79.934\n",
            "Average loss: 0.004538850882649422\n",
            "[ Evaluate epoch: 22 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 78.23\n",
            "Average loss: 0.006421920478343964\n",
            "Model Saved!\n",
            "[ Train epoch: 22 ]\n",
            "Train accuarcy: 81.428\n",
            "Train average loss: 0.004151920683383941\n",
            "[ Evaluate epoch: 23 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 81.966\n",
            "Average loss: 0.004048504141569138\n",
            "[ Evaluate epoch: 23 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 80.16\n",
            "Average loss: 0.005782841637730599\n",
            "Model Saved!\n",
            "[ Train epoch: 23 ]\n",
            "Train accuarcy: 81.862\n",
            "Train average loss: 0.004080228493213654\n",
            "[ Evaluate epoch: 24 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 80.002\n",
            "Average loss: 0.0044025241690874095\n",
            "[ Evaluate epoch: 24 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 77.99\n",
            "Average loss: 0.0064284891247749325\n",
            "Model Saved!\n",
            "[ Train epoch: 24 ]\n",
            "Train accuarcy: 82.462\n",
            "Train average loss: 0.0039921763432025905\n",
            "[ Evaluate epoch: 25 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 82.202\n",
            "Average loss: 0.0039843771326541905\n",
            "[ Evaluate epoch: 25 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 80.16\n",
            "Average loss: 0.005878878185153007\n",
            "Model Saved!\n",
            "[ Train epoch: 25 ]\n",
            "Train accuarcy: 82.622\n",
            "Train average loss: 0.003911052045822144\n",
            "[ Evaluate epoch: 26 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 81.358\n",
            "Average loss: 0.00416882479429245\n",
            "[ Evaluate epoch: 26 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 78.39\n",
            "Average loss: 0.006471032491326332\n",
            "Model Saved!\n",
            "[ Train epoch: 26 ]\n",
            "Train accuarcy: 83.002\n",
            "Train average loss: 0.0038322177582979203\n",
            "[ Evaluate epoch: 27 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 83.192\n",
            "Average loss: 0.0037986656576395033\n",
            "[ Evaluate epoch: 27 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 80.27\n",
            "Average loss: 0.005668219435214996\n",
            "Model Saved!\n",
            "[ Train epoch: 27 ]\n",
            "Train accuarcy: 83.162\n",
            "Train average loss: 0.0037749959146976472\n",
            "[ Evaluate epoch: 28 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 82.772\n",
            "Average loss: 0.0038188986307382583\n",
            "[ Evaluate epoch: 28 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 80.45\n",
            "Average loss: 0.00572771335542202\n",
            "Model Saved!\n",
            "[ Train epoch: 28 ]\n",
            "Train accuarcy: 83.678\n",
            "Train average loss: 0.0036687194055318832\n",
            "[ Evaluate epoch: 29 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 82.686\n",
            "Average loss: 0.0039005583429336547\n",
            "[ Evaluate epoch: 29 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 79.69\n",
            "Average loss: 0.006030200585722924\n",
            "Model Saved!\n",
            "[ Train epoch: 29 ]\n",
            "Train accuarcy: 83.95\n",
            "Train average loss: 0.003614337609410286\n",
            "[ Evaluate epoch: 30 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 83.264\n",
            "Average loss: 0.003740271179676056\n",
            "[ Evaluate epoch: 30 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 80.42\n",
            "Average loss: 0.005739199036359787\n",
            "Model Saved!\n",
            "[ Train epoch: 30 ]\n",
            "Train accuarcy: 84.208\n",
            "Train average loss: 0.0035646791172027586\n",
            "[ Evaluate epoch: 31 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 83.834\n",
            "Average loss: 0.0036235645541548727\n",
            "[ Evaluate epoch: 31 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 81.43\n",
            "Average loss: 0.005655310866236687\n",
            "Model Saved!\n",
            "[ Train epoch: 31 ]\n",
            "Train accuarcy: 84.332\n",
            "Train average loss: 0.0035144881999492647\n",
            "[ Evaluate epoch: 32 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 80.472\n",
            "Average loss: 0.004358655834794044\n",
            "[ Evaluate epoch: 32 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 78.18\n",
            "Average loss: 0.00654148510992527\n",
            "Model Saved!\n",
            "[ Train epoch: 32 ]\n",
            "Train accuarcy: 84.676\n",
            "Train average loss: 0.003450044629573822\n",
            "[ Evaluate epoch: 33 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 82.846\n",
            "Average loss: 0.0038836625814437867\n",
            "[ Evaluate epoch: 33 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 79.45\n",
            "Average loss: 0.006145222800970077\n",
            "Model Saved!\n",
            "[ Train epoch: 33 ]\n",
            "Train accuarcy: 84.772\n",
            "Train average loss: 0.0034065197297930717\n",
            "[ Evaluate epoch: 34 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 84.048\n",
            "Average loss: 0.003533858980834484\n",
            "[ Evaluate epoch: 34 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 81.33\n",
            "Average loss: 0.005676734885573387\n",
            "Model Saved!\n",
            "[ Train epoch: 34 ]\n",
            "Train accuarcy: 85.266\n",
            "Train average loss: 0.0033237380036711694\n",
            "[ Evaluate epoch: 35 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 82.692\n",
            "Average loss: 0.0039192993628978726\n",
            "[ Evaluate epoch: 35 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 79.51\n",
            "Average loss: 0.0063598457783460615\n",
            "Model Saved!\n",
            "[ Train epoch: 35 ]\n",
            "Train accuarcy: 85.43\n",
            "Train average loss: 0.0032874657410383226\n",
            "[ Evaluate epoch: 36 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 85.53\n",
            "Average loss: 0.0032553392067551615\n",
            "[ Evaluate epoch: 36 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.93\n",
            "Average loss: 0.005273846654593945\n",
            "Model Saved!\n",
            "[ Train epoch: 36 ]\n",
            "Train accuarcy: 85.53\n",
            "Train average loss: 0.0032422840306162836\n",
            "[ Evaluate epoch: 37 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 85.402\n",
            "Average loss: 0.003262848945260048\n",
            "[ Evaluate epoch: 37 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.69\n",
            "Average loss: 0.005186676728725434\n",
            "Model Saved!\n",
            "[ Train epoch: 37 ]\n",
            "Train accuarcy: 85.938\n",
            "Train average loss: 0.0031772999015450476\n",
            "[ Evaluate epoch: 38 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 85.734\n",
            "Average loss: 0.0032048409801721574\n",
            "[ Evaluate epoch: 38 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 81.98\n",
            "Average loss: 0.005353630113601684\n",
            "Model Saved!\n",
            "[ Train epoch: 38 ]\n",
            "Train accuarcy: 86.056\n",
            "Train average loss: 0.003148924505710602\n",
            "[ Evaluate epoch: 39 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 85.472\n",
            "Average loss: 0.0032343495163321494\n",
            "[ Evaluate epoch: 39 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.25\n",
            "Average loss: 0.005412604171037674\n",
            "Model Saved!\n",
            "[ Train epoch: 39 ]\n",
            "Train accuarcy: 86.196\n",
            "Train average loss: 0.003105404597222805\n",
            "[ Evaluate epoch: 40 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.08\n",
            "Average loss: 0.003101575160622597\n",
            "[ Evaluate epoch: 40 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.24\n",
            "Average loss: 0.005015453073382378\n",
            "Model Saved!\n",
            "[ Train epoch: 40 ]\n",
            "Train accuarcy: 86.444\n",
            "Train average loss: 0.003069850480556488\n",
            "[ Evaluate epoch: 41 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.684\n",
            "Average loss: 0.0029861187505722045\n",
            "[ Evaluate epoch: 41 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.11\n",
            "Average loss: 0.0049849761217832566\n",
            "Model Saved!\n",
            "[ Train epoch: 41 ]\n",
            "Train accuarcy: 86.604\n",
            "Train average loss: 0.0030274417540431023\n",
            "[ Evaluate epoch: 42 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 85.628\n",
            "Average loss: 0.0032901062935590743\n",
            "[ Evaluate epoch: 42 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.0\n",
            "Average loss: 0.005518501003086567\n",
            "Model Saved!\n",
            "[ Train epoch: 42 ]\n",
            "Train accuarcy: 86.916\n",
            "Train average loss: 0.002941332542002201\n",
            "[ Evaluate epoch: 43 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.334\n",
            "Average loss: 0.003103845692574978\n",
            "[ Evaluate epoch: 43 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.33\n",
            "Average loss: 0.0054531288325786595\n",
            "Model Saved!\n",
            "[ Train epoch: 43 ]\n",
            "Train accuarcy: 87.062\n",
            "Train average loss: 0.002927381517291069\n",
            "[ Evaluate epoch: 44 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 84.786\n",
            "Average loss: 0.0034083030539751053\n",
            "[ Evaluate epoch: 44 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 81.4\n",
            "Average loss: 0.005786271515488625\n",
            "Model Saved!\n",
            "[ Train epoch: 44 ]\n",
            "Train accuarcy: 87.204\n",
            "Train average loss: 0.002862615893781185\n",
            "[ Evaluate epoch: 45 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.576\n",
            "Average loss: 0.0030137017580866812\n",
            "[ Evaluate epoch: 45 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.28\n",
            "Average loss: 0.00529744614213705\n",
            "Model Saved!\n",
            "[ Train epoch: 45 ]\n",
            "Train accuarcy: 87.466\n",
            "Train average loss: 0.0028449427735805512\n",
            "[ Evaluate epoch: 46 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.14\n",
            "Average loss: 0.0029060587430000304\n",
            "[ Evaluate epoch: 46 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.31\n",
            "Average loss: 0.005172583395242691\n",
            "Model Saved!\n",
            "[ Train epoch: 46 ]\n",
            "Train accuarcy: 87.376\n",
            "Train average loss: 0.002838315413296223\n",
            "[ Evaluate epoch: 47 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.97\n",
            "Average loss: 0.0029138975301384925\n",
            "[ Evaluate epoch: 47 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.97\n",
            "Average loss: 0.005240482312440872\n",
            "Model Saved!\n",
            "[ Train epoch: 47 ]\n",
            "Train accuarcy: 87.87\n",
            "Train average loss: 0.0027476733607053757\n",
            "[ Evaluate epoch: 48 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.37\n",
            "Average loss: 0.0030093468895554545\n",
            "[ Evaluate epoch: 48 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.84\n",
            "Average loss: 0.00538335163295269\n",
            "Model Saved!\n",
            "[ Train epoch: 48 ]\n",
            "Train accuarcy: 87.848\n",
            "Train average loss: 0.0027511633747816086\n",
            "[ Evaluate epoch: 49 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.424\n",
            "Average loss: 0.0030061326053738594\n",
            "[ Evaluate epoch: 49 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.55\n",
            "Average loss: 0.005462522637844086\n",
            "Model Saved!\n",
            "[ Train epoch: 49 ]\n",
            "Train accuarcy: 87.954\n",
            "Train average loss: 0.002707098355591297\n",
            "[ Evaluate epoch: 50 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.03\n",
            "Average loss: 0.002929427033662796\n",
            "[ Evaluate epoch: 50 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.51\n",
            "Average loss: 0.005358178767561912\n",
            "Model Saved!\n",
            "[ Train epoch: 50 ]\n",
            "Train accuarcy: 88.208\n",
            "Train average loss: 0.0026501551455259324\n",
            "[ Evaluate epoch: 51 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.56\n",
            "Average loss: 0.0029763037663698196\n",
            "[ Evaluate epoch: 51 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.3\n",
            "Average loss: 0.005746329724788665\n",
            "Model Saved!\n",
            "[ Train epoch: 51 ]\n",
            "Train accuarcy: 88.29\n",
            "Train average loss: 0.002630742002427578\n",
            "[ Evaluate epoch: 52 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.452\n",
            "Average loss: 0.002595080433189869\n",
            "[ Evaluate epoch: 52 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.4\n",
            "Average loss: 0.004786145617067814\n",
            "Model Saved!\n",
            "[ Train epoch: 52 ]\n",
            "Train accuarcy: 88.416\n",
            "Train average loss: 0.002602605767548084\n",
            "[ Evaluate epoch: 53 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.358\n",
            "Average loss: 0.0028034222838282585\n",
            "[ Evaluate epoch: 53 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.6\n",
            "Average loss: 0.005370326229929924\n",
            "Model Saved!\n",
            "[ Train epoch: 53 ]\n",
            "Train accuarcy: 88.676\n",
            "Train average loss: 0.0025735993188619613\n",
            "[ Evaluate epoch: 54 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.314\n",
            "Average loss: 0.002605955014228821\n",
            "[ Evaluate epoch: 54 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.0\n",
            "Average loss: 0.004965004199743271\n",
            "Model Saved!\n",
            "[ Train epoch: 54 ]\n",
            "Train accuarcy: 88.84\n",
            "Train average loss: 0.002524123410284519\n",
            "[ Evaluate epoch: 55 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.962\n",
            "Average loss: 0.0027105368530750277\n",
            "[ Evaluate epoch: 55 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.94\n",
            "Average loss: 0.0054596510976552964\n",
            "Model Saved!\n",
            "[ Train epoch: 55 ]\n",
            "Train accuarcy: 88.834\n",
            "Train average loss: 0.0024957217675447463\n",
            "[ Evaluate epoch: 56 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.69\n",
            "Average loss: 0.0025788084539771082\n",
            "[ Evaluate epoch: 56 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.06\n",
            "Average loss: 0.0050074575617909435\n",
            "Model Saved!\n",
            "[ Train epoch: 56 ]\n",
            "Train accuarcy: 88.918\n",
            "Train average loss: 0.0024962318554520605\n",
            "[ Evaluate epoch: 57 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 84.838\n",
            "Average loss: 0.00330804546803236\n",
            "[ Evaluate epoch: 57 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 81.11\n",
            "Average loss: 0.005944780662655831\n",
            "Model Saved!\n",
            "[ Train epoch: 57 ]\n",
            "Train accuarcy: 88.99\n",
            "Train average loss: 0.002470550205409527\n",
            "[ Evaluate epoch: 58 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.626\n",
            "Average loss: 0.0027486799287796022\n",
            "[ Evaluate epoch: 58 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.45\n",
            "Average loss: 0.005352795279026032\n",
            "Model Saved!\n",
            "[ Train epoch: 58 ]\n",
            "Train accuarcy: 89.174\n",
            "Train average loss: 0.002449954245686531\n",
            "[ Evaluate epoch: 59 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.956\n",
            "Average loss: 0.002669978129863739\n",
            "[ Evaluate epoch: 59 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.15\n",
            "Average loss: 0.005345960679650307\n",
            "Model Saved!\n",
            "[ Train epoch: 59 ]\n",
            "Train accuarcy: 89.3\n",
            "Train average loss: 0.002413459642827511\n",
            "[ Evaluate epoch: 60 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.744\n",
            "Average loss: 0.002312819712460041\n",
            "[ Evaluate epoch: 60 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.89\n",
            "Average loss: 0.004611101938784123\n",
            "Model Saved!\n",
            "[ Train epoch: 60 ]\n",
            "Train accuarcy: 89.506\n",
            "Train average loss: 0.0023643518549203873\n",
            "[ Evaluate epoch: 61 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.54\n",
            "Average loss: 0.0030366964837908743\n",
            "[ Evaluate epoch: 61 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.25\n",
            "Average loss: 0.0057821089833974834\n",
            "Model Saved!\n",
            "[ Train epoch: 61 ]\n",
            "Train accuarcy: 89.642\n",
            "Train average loss: 0.0023526397310197353\n",
            "[ Evaluate epoch: 62 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.592\n",
            "Average loss: 0.002496359613239765\n",
            "[ Evaluate epoch: 62 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.49\n",
            "Average loss: 0.005231185582280159\n",
            "Model Saved!\n",
            "[ Train epoch: 62 ]\n",
            "Train accuarcy: 89.628\n",
            "Train average loss: 0.002334553543329239\n",
            "[ Evaluate epoch: 63 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.054\n",
            "Average loss: 0.002681571186482906\n",
            "[ Evaluate epoch: 63 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.08\n",
            "Average loss: 0.005445106053352356\n",
            "Model Saved!\n",
            "[ Train epoch: 63 ]\n",
            "Train accuarcy: 89.964\n",
            "Train average loss: 0.0022792629608511927\n",
            "[ Evaluate epoch: 64 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.552\n",
            "Average loss: 0.0023145886784791945\n",
            "[ Evaluate epoch: 64 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.8\n",
            "Average loss: 0.0048845166131854054\n",
            "Model Saved!\n",
            "[ Train epoch: 64 ]\n",
            "Train accuarcy: 90.008\n",
            "Train average loss: 0.002272453578710556\n",
            "[ Evaluate epoch: 65 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.516\n",
            "Average loss: 0.0030196504309773445\n",
            "[ Evaluate epoch: 65 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.0\n",
            "Average loss: 0.006063640296459198\n",
            "Model Saved!\n",
            "[ Train epoch: 65 ]\n",
            "Train accuarcy: 89.93\n",
            "Train average loss: 0.002275910325050354\n",
            "[ Evaluate epoch: 66 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.416\n",
            "Average loss: 0.0029738569316267965\n",
            "[ Evaluate epoch: 66 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 80.78\n",
            "Average loss: 0.006260870036482811\n",
            "Model Saved!\n",
            "[ Train epoch: 66 ]\n",
            "Train accuarcy: 90.128\n",
            "Train average loss: 0.002236262526363134\n",
            "[ Evaluate epoch: 67 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.424\n",
            "Average loss: 0.002346351039111614\n",
            "[ Evaluate epoch: 67 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.22\n",
            "Average loss: 0.005008280146121979\n",
            "Model Saved!\n",
            "[ Train epoch: 67 ]\n",
            "Train accuarcy: 90.228\n",
            "Train average loss: 0.002196988510787487\n",
            "[ Evaluate epoch: 68 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.766\n",
            "Average loss: 0.0027315559720993044\n",
            "[ Evaluate epoch: 68 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.6\n",
            "Average loss: 0.0056867597609758375\n",
            "Model Saved!\n",
            "[ Train epoch: 68 ]\n",
            "Train accuarcy: 90.27\n",
            "Train average loss: 0.00220116712898016\n",
            "[ Evaluate epoch: 69 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.5\n",
            "Average loss: 0.0023290377041697502\n",
            "[ Evaluate epoch: 69 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.12\n",
            "Average loss: 0.004925738821923733\n",
            "Model Saved!\n",
            "[ Train epoch: 69 ]\n",
            "Train accuarcy: 90.302\n",
            "Train average loss: 0.002181717559695244\n",
            "[ Evaluate epoch: 70 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.158\n",
            "Average loss: 0.0024258035776019097\n",
            "[ Evaluate epoch: 70 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.99\n",
            "Average loss: 0.005072156602144241\n",
            "Model Saved!\n",
            "[ Train epoch: 70 ]\n",
            "Train accuarcy: 90.508\n",
            "Train average loss: 0.0021270765767991543\n",
            "[ Evaluate epoch: 71 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.148\n",
            "Average loss: 0.002396748846322298\n",
            "[ Evaluate epoch: 71 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.37\n",
            "Average loss: 0.005443435165286064\n",
            "Model Saved!\n",
            "[ Train epoch: 71 ]\n",
            "Train accuarcy: 90.518\n",
            "Train average loss: 0.0021245071169734\n",
            "[ Evaluate epoch: 72 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.686\n",
            "Average loss: 0.0020993077206611635\n",
            "[ Evaluate epoch: 72 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.27\n",
            "Average loss: 0.004786073262989521\n",
            "Model Saved!\n",
            "[ Train epoch: 72 ]\n",
            "Train accuarcy: 90.744\n",
            "Train average loss: 0.0020804139015078546\n",
            "[ Evaluate epoch: 73 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.922\n",
            "Average loss: 0.0020337397876381874\n",
            "[ Evaluate epoch: 73 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.51\n",
            "Average loss: 0.00460781568735838\n",
            "Model Saved!\n",
            "[ Train epoch: 73 ]\n",
            "Train accuarcy: 90.938\n",
            "Train average loss: 0.0020531281915307045\n",
            "[ Evaluate epoch: 74 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.856\n",
            "Average loss: 0.0024146295776963232\n",
            "[ Evaluate epoch: 74 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.75\n",
            "Average loss: 0.005013406307995319\n",
            "Model Saved!\n",
            "[ Train epoch: 74 ]\n",
            "Train accuarcy: 90.836\n",
            "Train average loss: 0.0020623920434713365\n",
            "[ Evaluate epoch: 75 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.43\n",
            "Average loss: 0.0023186046448349954\n",
            "[ Evaluate epoch: 75 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.54\n",
            "Average loss: 0.00533557496368885\n",
            "Model Saved!\n",
            "[ Train epoch: 75 ]\n",
            "Train accuarcy: 91.054\n",
            "Train average loss: 0.002029870225787163\n",
            "[ Evaluate epoch: 76 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.772\n",
            "Average loss: 0.002059174551218748\n",
            "[ Evaluate epoch: 76 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.59\n",
            "Average loss: 0.0049664821267127995\n",
            "Model Saved!\n",
            "[ Train epoch: 76 ]\n",
            "Train accuarcy: 90.938\n",
            "Train average loss: 0.0020093604089319707\n",
            "[ Evaluate epoch: 77 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.77\n",
            "Average loss: 0.002471777711212635\n",
            "[ Evaluate epoch: 77 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.64\n",
            "Average loss: 0.005461936186254025\n",
            "Model Saved!\n",
            "[ Train epoch: 77 ]\n",
            "Train accuarcy: 90.822\n",
            "Train average loss: 0.0020389106261730194\n",
            "[ Evaluate epoch: 78 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.422\n",
            "Average loss: 0.0020935496720671655\n",
            "[ Evaluate epoch: 78 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.8\n",
            "Average loss: 0.004899804717302322\n",
            "Model Saved!\n",
            "[ Train epoch: 78 ]\n",
            "Train accuarcy: 91.25\n",
            "Train average loss: 0.0019684065102040767\n",
            "[ Evaluate epoch: 79 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.792\n",
            "Average loss: 0.002240134288072586\n",
            "[ Evaluate epoch: 79 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.8\n",
            "Average loss: 0.005303021997213364\n",
            "Model Saved!\n",
            "[ Train epoch: 79 ]\n",
            "Train accuarcy: 91.332\n",
            "Train average loss: 0.001953127797245979\n",
            "[ Evaluate epoch: 80 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.964\n",
            "Average loss: 0.002422406205236912\n",
            "[ Evaluate epoch: 80 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.86\n",
            "Average loss: 0.005775544318556786\n",
            "Model Saved!\n",
            "[ Train epoch: 80 ]\n",
            "Train accuarcy: 91.434\n",
            "Train average loss: 0.0019519099861383439\n",
            "[ Evaluate epoch: 81 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.248\n",
            "Average loss: 0.002363971815109253\n",
            "[ Evaluate epoch: 81 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.09\n",
            "Average loss: 0.005234177964925766\n",
            "Model Saved!\n",
            "[ Train epoch: 81 ]\n",
            "Train accuarcy: 91.416\n",
            "Train average loss: 0.0019255355429649353\n",
            "[ Evaluate epoch: 82 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.68\n",
            "Average loss: 0.002768961534500122\n",
            "[ Evaluate epoch: 82 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.24\n",
            "Average loss: 0.006155692523717881\n",
            "Model Saved!\n",
            "[ Train epoch: 82 ]\n",
            "Train accuarcy: 91.34\n",
            "Train average loss: 0.0019215625077486037\n",
            "[ Evaluate epoch: 83 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.48\n",
            "Average loss: 0.0019253560151159764\n",
            "[ Evaluate epoch: 83 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.05\n",
            "Average loss: 0.004732358418405056\n",
            "Model Saved!\n",
            "[ Train epoch: 83 ]\n",
            "Train accuarcy: 91.448\n",
            "Train average loss: 0.0018948890744149684\n",
            "[ Evaluate epoch: 84 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.214\n",
            "Average loss: 0.001936251626163721\n",
            "[ Evaluate epoch: 84 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.23\n",
            "Average loss: 0.004842511874437332\n",
            "Model Saved!\n",
            "[ Train epoch: 84 ]\n",
            "Train accuarcy: 91.724\n",
            "Train average loss: 0.0018529511280357838\n",
            "[ Evaluate epoch: 85 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.844\n",
            "Average loss: 0.0018203527250885963\n",
            "[ Evaluate epoch: 85 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.89\n",
            "Average loss: 0.004499032941460609\n",
            "Model Saved!\n",
            "[ Train epoch: 85 ]\n",
            "Train accuarcy: 91.672\n",
            "Train average loss: 0.0018639163364470004\n",
            "[ Evaluate epoch: 86 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.854\n",
            "Average loss: 0.0024928308044373988\n",
            "[ Evaluate epoch: 86 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.31\n",
            "Average loss: 0.005696209442615509\n",
            "Model Saved!\n",
            "[ Train epoch: 86 ]\n",
            "Train accuarcy: 91.962\n",
            "Train average loss: 0.0018093635764718057\n",
            "[ Evaluate epoch: 87 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.27\n",
            "Average loss: 0.001950888827443123\n",
            "[ Evaluate epoch: 87 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.16\n",
            "Average loss: 0.0048444052964448925\n",
            "Model Saved!\n",
            "[ Train epoch: 87 ]\n",
            "Train accuarcy: 92.118\n",
            "Train average loss: 0.0017822439782321452\n",
            "[ Evaluate epoch: 88 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.142\n",
            "Average loss: 0.0023892678970098497\n",
            "[ Evaluate epoch: 88 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.87\n",
            "Average loss: 0.005840279106795788\n",
            "Model Saved!\n",
            "[ Train epoch: 88 ]\n",
            "Train accuarcy: 91.84\n",
            "Train average loss: 0.0018071828071773053\n",
            "[ Evaluate epoch: 89 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.258\n",
            "Average loss: 0.002348327297270298\n",
            "[ Evaluate epoch: 89 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.27\n",
            "Average loss: 0.005676394435763359\n",
            "Model Saved!\n",
            "[ Train epoch: 89 ]\n",
            "Train accuarcy: 92.12\n",
            "Train average loss: 0.001755658572167158\n",
            "[ Evaluate epoch: 90 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.804\n",
            "Average loss: 0.0018266481663286687\n",
            "[ Evaluate epoch: 90 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.91\n",
            "Average loss: 0.0050148952573537825\n",
            "Model Saved!\n",
            "[ Train epoch: 90 ]\n",
            "Train accuarcy: 92.14\n",
            "Train average loss: 0.001742959765344858\n",
            "[ Evaluate epoch: 91 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.342\n",
            "Average loss: 0.0017356676068902015\n",
            "[ Evaluate epoch: 91 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.74\n",
            "Average loss: 0.004635260562598705\n",
            "Model Saved!\n",
            "[ Train epoch: 91 ]\n",
            "Train accuarcy: 92.126\n",
            "Train average loss: 0.0017601321560144424\n",
            "[ Evaluate epoch: 92 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.286\n",
            "Average loss: 0.0019514402569830419\n",
            "[ Evaluate epoch: 92 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.28\n",
            "Average loss: 0.005052451376616955\n",
            "Model Saved!\n",
            "[ Train epoch: 92 ]\n",
            "Train accuarcy: 92.236\n",
            "Train average loss: 0.0017313011918962\n",
            "[ Evaluate epoch: 93 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.638\n",
            "Average loss: 0.0020957288151979446\n",
            "[ Evaluate epoch: 93 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.68\n",
            "Average loss: 0.005160995531082153\n",
            "Model Saved!\n",
            "[ Train epoch: 93 ]\n",
            "Train accuarcy: 92.278\n",
            "Train average loss: 0.0017043230345845223\n",
            "[ Evaluate epoch: 94 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.614\n",
            "Average loss: 0.0018389779411256313\n",
            "[ Evaluate epoch: 94 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.4\n",
            "Average loss: 0.0049468509018421175\n",
            "Model Saved!\n",
            "[ Train epoch: 94 ]\n",
            "Train accuarcy: 92.414\n",
            "Train average loss: 0.0017055284690856935\n",
            "[ Evaluate epoch: 95 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.506\n",
            "Average loss: 0.0020444219030439853\n",
            "[ Evaluate epoch: 95 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.12\n",
            "Average loss: 0.005264464074373245\n",
            "Model Saved!\n",
            "[ Train epoch: 95 ]\n",
            "Train accuarcy: 92.412\n",
            "Train average loss: 0.0016699421140551568\n",
            "[ Evaluate epoch: 96 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.422\n",
            "Average loss: 0.0018952744215726851\n",
            "[ Evaluate epoch: 96 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.36\n",
            "Average loss: 0.004934432782232761\n",
            "Model Saved!\n",
            "[ Train epoch: 96 ]\n",
            "Train accuarcy: 92.572\n",
            "Train average loss: 0.0016528896482288838\n",
            "[ Evaluate epoch: 97 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.372\n",
            "Average loss: 0.0016800634703040124\n",
            "[ Evaluate epoch: 97 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.57\n",
            "Average loss: 0.004745410427451134\n",
            "Model Saved!\n",
            "[ Train epoch: 97 ]\n",
            "Train accuarcy: 92.536\n",
            "Train average loss: 0.001656544391363859\n",
            "[ Evaluate epoch: 98 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.56\n",
            "Average loss: 0.00207331231161952\n",
            "[ Evaluate epoch: 98 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.44\n",
            "Average loss: 0.00521899276971817\n",
            "Model Saved!\n",
            "[ Train epoch: 98 ]\n",
            "Train accuarcy: 92.656\n",
            "Train average loss: 0.0016229471318423747\n",
            "[ Evaluate epoch: 99 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.152\n",
            "Average loss: 0.00217573885306716\n",
            "[ Evaluate epoch: 99 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.12\n",
            "Average loss: 0.005984907838702202\n",
            "Model Saved!\n",
            "[ Train epoch: 99 ]\n",
            "Train accuarcy: 92.852\n",
            "Train average loss: 0.0015971066327393054\n",
            "[ Evaluate epoch: 100 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.9\n",
            "Average loss: 0.001553055421113968\n",
            "[ Evaluate epoch: 100 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.01\n",
            "Average loss: 0.0046507990583777425\n",
            "Model Saved!\n",
            "[ Train epoch: 100 ]\n",
            "Train accuarcy: 92.602\n",
            "Train average loss: 0.0016159902708232402\n",
            "[ Evaluate epoch: 101 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.04\n",
            "Average loss: 0.0015276932252943516\n",
            "[ Evaluate epoch: 101 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.1\n",
            "Average loss: 0.004552506597340107\n",
            "Model Saved!\n",
            "[ Train epoch: 101 ]\n",
            "Train accuarcy: 92.908\n",
            "Train average loss: 0.001600219620168209\n",
            "[ Evaluate epoch: 102 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.436\n",
            "Average loss: 0.001872838778346777\n",
            "[ Evaluate epoch: 102 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.73\n",
            "Average loss: 0.005135749359428883\n",
            "Model Saved!\n",
            "[ Train epoch: 102 ]\n",
            "Train accuarcy: 92.938\n",
            "Train average loss: 0.0015582521551847457\n",
            "[ Evaluate epoch: 103 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.522\n",
            "Average loss: 0.002340422757267952\n",
            "[ Evaluate epoch: 103 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.65\n",
            "Average loss: 0.006105969968438148\n",
            "Model Saved!\n",
            "[ Train epoch: 103 ]\n",
            "Train accuarcy: 93.04\n",
            "Train average loss: 0.001558600062429905\n",
            "[ Evaluate epoch: 104 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.938\n",
            "Average loss: 0.0017862955155968667\n",
            "[ Evaluate epoch: 104 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.28\n",
            "Average loss: 0.005144470727443695\n",
            "Model Saved!\n",
            "[ Train epoch: 104 ]\n",
            "Train accuarcy: 93.006\n",
            "Train average loss: 0.0015643371304869653\n",
            "[ Evaluate epoch: 105 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.222\n",
            "Average loss: 0.001505751754641533\n",
            "[ Evaluate epoch: 105 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.17\n",
            "Average loss: 0.004600071506202221\n",
            "Model Saved!\n",
            "[ Train epoch: 105 ]\n",
            "Train accuarcy: 93.218\n",
            "Train average loss: 0.0015279770496487617\n",
            "[ Evaluate epoch: 106 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.964\n",
            "Average loss: 0.001749312388598919\n",
            "[ Evaluate epoch: 106 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.9\n",
            "Average loss: 0.0052036006957292555\n",
            "Model Saved!\n",
            "[ Train epoch: 106 ]\n",
            "Train accuarcy: 92.978\n",
            "Train average loss: 0.0015478423130512238\n",
            "[ Evaluate epoch: 107 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.48\n",
            "Average loss: 0.0014507379084080457\n",
            "[ Evaluate epoch: 107 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.5\n",
            "Average loss: 0.00466154665350914\n",
            "Model Saved!\n",
            "[ Train epoch: 107 ]\n",
            "Train accuarcy: 93.236\n",
            "Train average loss: 0.0015183063511550426\n",
            "[ Evaluate epoch: 108 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.016\n",
            "Average loss: 0.0015420014316588641\n",
            "[ Evaluate epoch: 108 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.84\n",
            "Average loss: 0.004678727668523788\n",
            "Model Saved!\n",
            "[ Train epoch: 108 ]\n",
            "Train accuarcy: 93.144\n",
            "Train average loss: 0.0015166910170018673\n",
            "[ Evaluate epoch: 109 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.358\n",
            "Average loss: 0.0016862829305231571\n",
            "[ Evaluate epoch: 109 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.94\n",
            "Average loss: 0.005220810636878013\n",
            "Model Saved!\n",
            "[ Train epoch: 109 ]\n",
            "Train accuarcy: 93.196\n",
            "Train average loss: 0.00151847204297781\n",
            "[ Evaluate epoch: 110 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.934\n",
            "Average loss: 0.0017583664245903491\n",
            "[ Evaluate epoch: 110 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.86\n",
            "Average loss: 0.0052591357797384265\n",
            "Model Saved!\n",
            "[ Train epoch: 110 ]\n",
            "Train accuarcy: 93.268\n",
            "Train average loss: 0.001494973704069853\n",
            "[ Evaluate epoch: 111 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.442\n",
            "Average loss: 0.0016438250024616717\n",
            "[ Evaluate epoch: 111 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.07\n",
            "Average loss: 0.005062827344238758\n",
            "Model Saved!\n",
            "[ Train epoch: 111 ]\n",
            "Train accuarcy: 93.332\n",
            "Train average loss: 0.001476973647773266\n",
            "[ Evaluate epoch: 112 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.62\n",
            "Average loss: 0.001818659664094448\n",
            "[ Evaluate epoch: 112 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.45\n",
            "Average loss: 0.005463538175821304\n",
            "Model Saved!\n",
            "[ Train epoch: 112 ]\n",
            "Train accuarcy: 93.396\n",
            "Train average loss: 0.0014593772959709168\n",
            "[ Evaluate epoch: 113 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.706\n",
            "Average loss: 0.0014242622885107994\n",
            "[ Evaluate epoch: 113 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.3\n",
            "Average loss: 0.0044618489861488345\n",
            "Model Saved!\n",
            "[ Train epoch: 113 ]\n",
            "Train accuarcy: 93.482\n",
            "Train average loss: 0.0014551735623180866\n",
            "[ Evaluate epoch: 114 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.834\n",
            "Average loss: 0.001566984634399414\n",
            "[ Evaluate epoch: 114 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.71\n",
            "Average loss: 0.005007794253528118\n",
            "Model Saved!\n",
            "[ Train epoch: 114 ]\n",
            "Train accuarcy: 93.536\n",
            "Train average loss: 0.0014450239312648774\n",
            "[ Evaluate epoch: 115 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.706\n",
            "Average loss: 0.001597405465990305\n",
            "[ Evaluate epoch: 115 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.51\n",
            "Average loss: 0.004911632156372071\n",
            "Model Saved!\n",
            "[ Train epoch: 115 ]\n",
            "Train accuarcy: 93.65\n",
            "Train average loss: 0.00141018545627594\n",
            "[ Evaluate epoch: 116 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.2\n",
            "Average loss: 0.0015207276278734207\n",
            "[ Evaluate epoch: 116 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.37\n",
            "Average loss: 0.004763335031270981\n",
            "Model Saved!\n",
            "[ Train epoch: 116 ]\n",
            "Train accuarcy: 93.47\n",
            "Train average loss: 0.0014251949261128902\n",
            "[ Evaluate epoch: 117 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.102\n",
            "Average loss: 0.0017112621587514877\n",
            "[ Evaluate epoch: 117 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.55\n",
            "Average loss: 0.005405692195892334\n",
            "Model Saved!\n",
            "[ Train epoch: 117 ]\n",
            "Train accuarcy: 93.598\n",
            "Train average loss: 0.001420460220426321\n",
            "[ Evaluate epoch: 118 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.904\n",
            "Average loss: 0.0017633913104236126\n",
            "[ Evaluate epoch: 118 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.97\n",
            "Average loss: 0.005330607151985169\n",
            "Model Saved!\n",
            "[ Train epoch: 118 ]\n",
            "Train accuarcy: 93.876\n",
            "Train average loss: 0.0013616407236456871\n",
            "[ Evaluate epoch: 119 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.344\n",
            "Average loss: 0.0014516093024611472\n",
            "[ Evaluate epoch: 119 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.8\n",
            "Average loss: 0.00479470526278019\n",
            "Model Saved!\n",
            "[ Train epoch: 119 ]\n",
            "Train accuarcy: 93.584\n",
            "Train average loss: 0.0013890419847518206\n",
            "[ Evaluate epoch: 120 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.708\n",
            "Average loss: 0.001369913989752531\n",
            "[ Evaluate epoch: 120 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.34\n",
            "Average loss: 0.004658742812275887\n",
            "Model Saved!\n",
            "[ Train epoch: 120 ]\n",
            "Train accuarcy: 93.908\n",
            "Train average loss: 0.0013612946936488151\n",
            "[ Evaluate epoch: 121 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.206\n",
            "Average loss: 0.0016742620293796063\n",
            "[ Evaluate epoch: 121 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.14\n",
            "Average loss: 0.00526201740205288\n",
            "Model Saved!\n",
            "[ Train epoch: 121 ]\n",
            "Train accuarcy: 93.688\n",
            "Train average loss: 0.001385708252042532\n",
            "[ Evaluate epoch: 122 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.686\n",
            "Average loss: 0.0015610795575380326\n",
            "[ Evaluate epoch: 122 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.28\n",
            "Average loss: 0.005212627609074116\n",
            "Model Saved!\n",
            "[ Train epoch: 122 ]\n",
            "Train accuarcy: 93.932\n",
            "Train average loss: 0.0013399976379424334\n",
            "[ Evaluate epoch: 123 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.2\n",
            "Average loss: 0.0016853195518255234\n",
            "[ Evaluate epoch: 123 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.62\n",
            "Average loss: 0.005317183083295822\n",
            "Model Saved!\n",
            "[ Train epoch: 123 ]\n",
            "Train accuarcy: 93.946\n",
            "Train average loss: 0.0013340223255753516\n",
            "[ Evaluate epoch: 124 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.754\n",
            "Average loss: 0.0013702534230053424\n",
            "[ Evaluate epoch: 124 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.06\n",
            "Average loss: 0.00499342640042305\n",
            "Model Saved!\n",
            "[ Train epoch: 124 ]\n",
            "Train accuarcy: 94.058\n",
            "Train average loss: 0.0013216384200006722\n",
            "[ Evaluate epoch: 125 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.68\n",
            "Average loss: 0.001824860959202051\n",
            "[ Evaluate epoch: 125 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.05\n",
            "Average loss: 0.005779959234595298\n",
            "Model Saved!\n",
            "[ Train epoch: 125 ]\n",
            "Train accuarcy: 94.2\n",
            "Train average loss: 0.0012873859702795744\n",
            "[ Evaluate epoch: 126 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.046\n",
            "Average loss: 0.0014870516504347324\n",
            "[ Evaluate epoch: 126 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.37\n",
            "Average loss: 0.005281049685180187\n",
            "Model Saved!\n",
            "[ Train epoch: 126 ]\n",
            "Train accuarcy: 94.224\n",
            "Train average loss: 0.0012935139305889607\n",
            "[ Evaluate epoch: 127 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.648\n",
            "Average loss: 0.0014049320497363806\n",
            "[ Evaluate epoch: 127 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.2\n",
            "Average loss: 0.004981830479204655\n",
            "Model Saved!\n",
            "[ Train epoch: 127 ]\n",
            "Train accuarcy: 94.292\n",
            "Train average loss: 0.0012773258554190398\n",
            "[ Evaluate epoch: 128 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.648\n",
            "Average loss: 0.0011953418508172035\n",
            "[ Evaluate epoch: 128 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.9\n",
            "Average loss: 0.0046162235498428345\n",
            "Model Saved!\n",
            "[ Train epoch: 128 ]\n",
            "Train accuarcy: 94.072\n",
            "Train average loss: 0.001302171194255352\n",
            "[ Evaluate epoch: 129 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.002\n",
            "Average loss: 0.0013200770441442729\n",
            "[ Evaluate epoch: 129 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.5\n",
            "Average loss: 0.0048255420222878456\n",
            "Model Saved!\n",
            "[ Train epoch: 129 ]\n",
            "Train accuarcy: 94.376\n",
            "Train average loss: 0.0012496739134192467\n",
            "[ Evaluate epoch: 130 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.53\n",
            "Average loss: 0.0012390374666452408\n",
            "[ Evaluate epoch: 130 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.78\n",
            "Average loss: 0.004614832520484924\n",
            "Model Saved!\n",
            "[ Train epoch: 130 ]\n",
            "Train accuarcy: 94.114\n",
            "Train average loss: 0.0012784611286222934\n",
            "[ Evaluate epoch: 131 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.644\n",
            "Average loss: 0.0015526300357282161\n",
            "[ Evaluate epoch: 131 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.41\n",
            "Average loss: 0.00527374724894762\n",
            "Model Saved!\n",
            "[ Train epoch: 131 ]\n",
            "Train accuarcy: 94.1\n",
            "Train average loss: 0.0012948179645836354\n",
            "[ Evaluate epoch: 132 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.416\n",
            "Average loss: 0.0012596984423696995\n",
            "[ Evaluate epoch: 132 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.27\n",
            "Average loss: 0.004780171628296376\n",
            "Model Saved!\n",
            "[ Train epoch: 132 ]\n",
            "Train accuarcy: 94.396\n",
            "Train average loss: 0.0012318658327311277\n",
            "[ Evaluate epoch: 133 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.764\n",
            "Average loss: 0.0015266129000484944\n",
            "[ Evaluate epoch: 133 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.14\n",
            "Average loss: 0.0053344780027866365\n",
            "Model Saved!\n",
            "[ Train epoch: 133 ]\n",
            "Train accuarcy: 94.326\n",
            "Train average loss: 0.0012464214316010476\n",
            "[ Evaluate epoch: 134 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.444\n",
            "Average loss: 0.0012442403109371662\n",
            "[ Evaluate epoch: 134 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.09\n",
            "Average loss: 0.004898670417070389\n",
            "Model Saved!\n",
            "[ Train epoch: 134 ]\n",
            "Train accuarcy: 94.454\n",
            "Train average loss: 0.001235911298468709\n",
            "[ Evaluate epoch: 135 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.572\n",
            "Average loss: 0.0018121142971515656\n",
            "[ Evaluate epoch: 135 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.32\n",
            "Average loss: 0.006158099493384361\n",
            "Model Saved!\n",
            "[ Train epoch: 135 ]\n",
            "Train accuarcy: 94.63\n",
            "Train average loss: 0.0011881812311708927\n",
            "[ Evaluate epoch: 136 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.5\n",
            "Average loss: 0.0013950374042242765\n",
            "[ Evaluate epoch: 136 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.85\n",
            "Average loss: 0.00518513006567955\n",
            "Model Saved!\n",
            "[ Train epoch: 136 ]\n",
            "Train accuarcy: 94.73\n",
            "Train average loss: 0.001185340774729848\n",
            "[ Evaluate epoch: 137 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.562\n",
            "Average loss: 0.0014292687226831913\n",
            "[ Evaluate epoch: 137 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.36\n",
            "Average loss: 0.005284326091408729\n",
            "Model Saved!\n",
            "[ Train epoch: 137 ]\n",
            "Train accuarcy: 94.492\n",
            "Train average loss: 0.001196297579407692\n",
            "[ Evaluate epoch: 138 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.836\n",
            "Average loss: 0.0011457947609573602\n",
            "[ Evaluate epoch: 138 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.96\n",
            "Average loss: 0.004720726577937603\n",
            "Model Saved!\n",
            "[ Train epoch: 138 ]\n",
            "Train accuarcy: 94.614\n",
            "Train average loss: 0.0011880820931494235\n",
            "[ Evaluate epoch: 139 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.164\n",
            "Average loss: 0.0010970904817432165\n",
            "[ Evaluate epoch: 139 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.0\n",
            "Average loss: 0.004853061231970787\n",
            "Model Saved!\n",
            "[ Train epoch: 139 ]\n",
            "Train accuarcy: 94.724\n",
            "Train average loss: 0.0011843700546771288\n",
            "[ Evaluate epoch: 140 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.142\n",
            "Average loss: 0.0012940843567252158\n",
            "[ Evaluate epoch: 140 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.66\n",
            "Average loss: 0.005214060087502003\n",
            "Model Saved!\n",
            "[ Train epoch: 140 ]\n",
            "Train accuarcy: 94.662\n",
            "Train average loss: 0.0011718866154551506\n",
            "[ Evaluate epoch: 141 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.776\n",
            "Average loss: 0.0015980749866366386\n",
            "[ Evaluate epoch: 141 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.17\n",
            "Average loss: 0.005448312306404114\n",
            "Model Saved!\n",
            "[ Train epoch: 141 ]\n",
            "Train accuarcy: 94.814\n",
            "Train average loss: 0.001158011154383421\n",
            "[ Evaluate epoch: 142 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.168\n",
            "Average loss: 0.001266441128924489\n",
            "[ Evaluate epoch: 142 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.74\n",
            "Average loss: 0.005190663400292396\n",
            "Model Saved!\n",
            "[ Train epoch: 142 ]\n",
            "Train accuarcy: 94.992\n",
            "Train average loss: 0.0011064887572079898\n",
            "[ Evaluate epoch: 143 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.49\n",
            "Average loss: 0.0011861949461698532\n",
            "[ Evaluate epoch: 143 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.27\n",
            "Average loss: 0.005027725276350975\n",
            "Model Saved!\n",
            "[ Train epoch: 143 ]\n",
            "Train accuarcy: 94.866\n",
            "Train average loss: 0.0011484579116851092\n",
            "[ Evaluate epoch: 144 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.326\n",
            "Average loss: 0.0010300065395236016\n",
            "[ Evaluate epoch: 144 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.34\n",
            "Average loss: 0.004622347770631314\n",
            "Model Saved!\n",
            "[ Train epoch: 144 ]\n",
            "Train accuarcy: 95.056\n",
            "Train average loss: 0.001103136898353696\n",
            "[ Evaluate epoch: 145 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.168\n",
            "Average loss: 0.0012455630036443473\n",
            "[ Evaluate epoch: 145 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.49\n",
            "Average loss: 0.00512883141040802\n",
            "Model Saved!\n",
            "[ Train epoch: 145 ]\n",
            "Train accuarcy: 94.854\n",
            "Train average loss: 0.0011319215316325426\n",
            "[ Evaluate epoch: 146 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.442\n",
            "Average loss: 0.0012057607062160968\n",
            "[ Evaluate epoch: 146 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.9\n",
            "Average loss: 0.005135745891928673\n",
            "Model Saved!\n",
            "[ Train epoch: 146 ]\n",
            "Train accuarcy: 94.888\n",
            "Train average loss: 0.0011237548445910215\n",
            "[ Evaluate epoch: 147 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.164\n",
            "Average loss: 0.0010729436661303044\n",
            "[ Evaluate epoch: 147 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.65\n",
            "Average loss: 0.004715619780123234\n",
            "Model Saved!\n",
            "[ Train epoch: 147 ]\n",
            "Train accuarcy: 95.132\n",
            "Train average loss: 0.00109877969622612\n",
            "[ Evaluate epoch: 148 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.01\n",
            "Average loss: 0.00181718118339777\n",
            "[ Evaluate epoch: 148 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.92\n",
            "Average loss: 0.006300028735399246\n",
            "Model Saved!\n",
            "[ Train epoch: 148 ]\n",
            "Train accuarcy: 95.016\n",
            "Train average loss: 0.001106970606148243\n",
            "[ Evaluate epoch: 149 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.79\n",
            "Average loss: 0.001334890656620264\n",
            "[ Evaluate epoch: 149 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.17\n",
            "Average loss: 0.005305547529459\n",
            "Model Saved!\n",
            "[ Train epoch: 149 ]\n",
            "Train accuarcy: 95.206\n",
            "Train average loss: 0.001067991279810667\n",
            "[ Evaluate epoch: 150 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.482\n",
            "Average loss: 0.0012513006218522788\n",
            "[ Evaluate epoch: 150 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.49\n",
            "Average loss: 0.005262648868560791\n",
            "Model Saved!\n",
            "[ Train epoch: 150 ]\n",
            "Train accuarcy: 95.216\n",
            "Train average loss: 0.0010631465552002192\n",
            "[ Evaluate epoch: 151 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.748\n",
            "Average loss: 0.0011604523842036723\n",
            "[ Evaluate epoch: 151 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.53\n",
            "Average loss: 0.005011829289793968\n",
            "Model Saved!\n",
            "[ Train epoch: 151 ]\n",
            "Train accuarcy: 95.086\n",
            "Train average loss: 0.0010843187020346523\n",
            "[ Evaluate epoch: 152 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.974\n",
            "Average loss: 0.000926325793042779\n",
            "[ Evaluate epoch: 152 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.5\n",
            "Average loss: 0.0044556084334850315\n",
            "Model Saved!\n",
            "[ Train epoch: 152 ]\n",
            "Train accuarcy: 95.148\n",
            "Train average loss: 0.001073494775146246\n",
            "[ Evaluate epoch: 153 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.42\n",
            "Average loss: 0.0012350118534266948\n",
            "[ Evaluate epoch: 153 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.6\n",
            "Average loss: 0.005402468833327293\n",
            "Model Saved!\n",
            "[ Train epoch: 153 ]\n",
            "Train accuarcy: 95.432\n",
            "Train average loss: 0.0010256036406755447\n",
            "[ Evaluate epoch: 154 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.54\n",
            "Average loss: 0.00116081729978323\n",
            "[ Evaluate epoch: 154 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.89\n",
            "Average loss: 0.005177117440104484\n",
            "Model Saved!\n",
            "[ Train epoch: 154 ]\n",
            "Train accuarcy: 95.28\n",
            "Train average loss: 0.001056133826225996\n",
            "[ Evaluate epoch: 155 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.582\n",
            "Average loss: 0.0013821366861462593\n",
            "[ Evaluate epoch: 155 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.29\n",
            "Average loss: 0.005640987730026245\n",
            "Model Saved!\n",
            "[ Train epoch: 155 ]\n",
            "Train accuarcy: 95.258\n",
            "Train average loss: 0.0010566977542638778\n",
            "[ Evaluate epoch: 156 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.206\n",
            "Average loss: 0.0014553173510730266\n",
            "[ Evaluate epoch: 156 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.81\n",
            "Average loss: 0.005906212076544762\n",
            "Model Saved!\n",
            "[ Train epoch: 156 ]\n",
            "Train accuarcy: 95.214\n",
            "Train average loss: 0.0010531968251615762\n",
            "[ Evaluate epoch: 157 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.494\n",
            "Average loss: 0.0012103031408041715\n",
            "[ Evaluate epoch: 157 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.87\n",
            "Average loss: 0.005507680299878121\n",
            "Model Saved!\n",
            "[ Train epoch: 157 ]\n",
            "Train accuarcy: 95.39\n",
            "Train average loss: 0.001005502728894353\n",
            "[ Evaluate epoch: 158 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.832\n",
            "Average loss: 0.0011122146708518266\n",
            "[ Evaluate epoch: 158 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.3\n",
            "Average loss: 0.005087156218290329\n",
            "Model Saved!\n",
            "[ Train epoch: 158 ]\n",
            "Train accuarcy: 95.264\n",
            "Train average loss: 0.0010286715381592512\n",
            "[ Evaluate epoch: 159 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.098\n",
            "Average loss: 0.0012694057466834784\n",
            "[ Evaluate epoch: 159 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.78\n",
            "Average loss: 0.005117600034177303\n",
            "Model Saved!\n",
            "[ Train epoch: 159 ]\n",
            "Train accuarcy: 95.248\n",
            "Train average loss: 0.001020832963064313\n",
            "[ Evaluate epoch: 160 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.838\n",
            "Average loss: 0.0009100555117428303\n",
            "[ Evaluate epoch: 160 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.49\n",
            "Average loss: 0.004621302434802055\n",
            "Model Saved!\n",
            "[ Train epoch: 160 ]\n",
            "Train accuarcy: 95.456\n",
            "Train average loss: 0.0010052586156874895\n",
            "[ Evaluate epoch: 161 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.954\n",
            "Average loss: 0.0011208945009857417\n",
            "[ Evaluate epoch: 161 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.16\n",
            "Average loss: 0.0051752631157636645\n",
            "Model Saved!\n",
            "[ Train epoch: 161 ]\n",
            "Train accuarcy: 95.424\n",
            "Train average loss: 0.0010068212114274502\n",
            "[ Evaluate epoch: 162 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.388\n",
            "Average loss: 0.0012500098326802254\n",
            "[ Evaluate epoch: 162 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.55\n",
            "Average loss: 0.00530421893298626\n",
            "Model Saved!\n",
            "[ Train epoch: 162 ]\n",
            "Train accuarcy: 95.644\n",
            "Train average loss: 0.0009743446355313062\n",
            "[ Evaluate epoch: 163 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.202\n",
            "Average loss: 0.0008575506249815226\n",
            "[ Evaluate epoch: 163 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.4\n",
            "Average loss: 0.0046707042634487154\n",
            "Model Saved!\n",
            "[ Train epoch: 163 ]\n",
            "Train accuarcy: 95.416\n",
            "Train average loss: 0.0010026249862462283\n",
            "[ Evaluate epoch: 164 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.684\n",
            "Average loss: 0.0015704031893610953\n",
            "[ Evaluate epoch: 164 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.59\n",
            "Average loss: 0.005980801299214363\n",
            "Model Saved!\n",
            "[ Train epoch: 164 ]\n",
            "Train accuarcy: 95.652\n",
            "Train average loss: 0.000962966929525137\n",
            "[ Evaluate epoch: 165 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.876\n",
            "Average loss: 0.0009276958776265383\n",
            "[ Evaluate epoch: 165 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.02\n",
            "Average loss: 0.00477849697470665\n",
            "Model Saved!\n",
            "[ Train epoch: 165 ]\n",
            "Train accuarcy: 95.714\n",
            "Train average loss: 0.0009568340115249157\n",
            "[ Evaluate epoch: 166 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.36\n",
            "Average loss: 0.0012399638484418393\n",
            "[ Evaluate epoch: 166 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.53\n",
            "Average loss: 0.005519023081660271\n",
            "Model Saved!\n",
            "[ Train epoch: 166 ]\n",
            "Train accuarcy: 95.588\n",
            "Train average loss: 0.0009643837378174066\n",
            "[ Evaluate epoch: 167 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.42\n",
            "Average loss: 0.0010190542186051608\n",
            "[ Evaluate epoch: 167 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.41\n",
            "Average loss: 0.005239057153463364\n",
            "Model Saved!\n",
            "[ Train epoch: 167 ]\n",
            "Train accuarcy: 95.672\n",
            "Train average loss: 0.000957012479826808\n",
            "[ Evaluate epoch: 168 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.64\n",
            "Average loss: 0.0016105753932893277\n",
            "[ Evaluate epoch: 168 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.88\n",
            "Average loss: 0.006182407733798027\n",
            "Model Saved!\n",
            "[ Train epoch: 168 ]\n",
            "Train accuarcy: 95.674\n",
            "Train average loss: 0.0009667027206718921\n",
            "[ Evaluate epoch: 169 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.994\n",
            "Average loss: 0.0010988391206413508\n",
            "[ Evaluate epoch: 169 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.63\n",
            "Average loss: 0.0054343161806464195\n",
            "Model Saved!\n",
            "[ Train epoch: 169 ]\n",
            "Train accuarcy: 95.734\n",
            "Train average loss: 0.0009548396553471684\n",
            "[ Evaluate epoch: 170 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.884\n",
            "Average loss: 0.0013037056931853295\n",
            "[ Evaluate epoch: 170 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.94\n",
            "Average loss: 0.005755145946145058\n",
            "Model Saved!\n",
            "[ Train epoch: 170 ]\n",
            "Train accuarcy: 95.766\n",
            "Train average loss: 0.0009346784792840481\n",
            "[ Evaluate epoch: 171 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.736\n",
            "Average loss: 0.0011453513315320015\n",
            "[ Evaluate epoch: 171 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.75\n",
            "Average loss: 0.005553542158007622\n",
            "Model Saved!\n",
            "[ Train epoch: 171 ]\n",
            "Train accuarcy: 95.894\n",
            "Train average loss: 0.0009216444933414459\n",
            "[ Evaluate epoch: 172 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.482\n",
            "Average loss: 0.0010068383000791073\n",
            "[ Evaluate epoch: 172 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.98\n",
            "Average loss: 0.005117075027525425\n",
            "Model Saved!\n",
            "[ Train epoch: 172 ]\n",
            "Train accuarcy: 95.984\n",
            "Train average loss: 0.0009019873343408108\n",
            "[ Evaluate epoch: 173 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.146\n",
            "Average loss: 0.0010689633464813233\n",
            "[ Evaluate epoch: 173 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.09\n",
            "Average loss: 0.005402302649617195\n",
            "Model Saved!\n",
            "[ Train epoch: 173 ]\n",
            "Train accuarcy: 95.904\n",
            "Train average loss: 0.0009143470625579358\n",
            "[ Evaluate epoch: 174 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.772\n",
            "Average loss: 0.0009287503431737422\n",
            "[ Evaluate epoch: 174 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.27\n",
            "Average loss: 0.0051332277715206145\n",
            "Model Saved!\n",
            "[ Train epoch: 174 ]\n",
            "Train accuarcy: 95.812\n",
            "Train average loss: 0.0009247328148782253\n",
            "[ Evaluate epoch: 175 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.246\n",
            "Average loss: 0.0008438138277828694\n",
            "[ Evaluate epoch: 175 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.01\n",
            "Average loss: 0.004952797524631024\n",
            "Model Saved!\n",
            "[ Train epoch: 175 ]\n",
            "Train accuarcy: 95.882\n",
            "Train average loss: 0.0009133445844054222\n",
            "[ Evaluate epoch: 176 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.16\n",
            "Average loss: 0.0012325926905870437\n",
            "[ Evaluate epoch: 176 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.45\n",
            "Average loss: 0.005571892701089382\n",
            "Model Saved!\n",
            "[ Train epoch: 176 ]\n",
            "Train accuarcy: 95.93\n",
            "Train average loss: 0.0008986838594824076\n",
            "[ Evaluate epoch: 177 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.094\n",
            "Average loss: 0.0008539977966248989\n",
            "[ Evaluate epoch: 177 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.71\n",
            "Average loss: 0.0049470780238509175\n",
            "Model Saved!\n",
            "[ Train epoch: 177 ]\n",
            "Train accuarcy: 95.856\n",
            "Train average loss: 0.0009094946318864822\n",
            "[ Evaluate epoch: 178 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.91\n",
            "Average loss: 0.0008910634434223175\n",
            "[ Evaluate epoch: 178 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.05\n",
            "Average loss: 0.004866432398557663\n",
            "Model Saved!\n",
            "[ Train epoch: 178 ]\n",
            "Train accuarcy: 96.116\n",
            "Train average loss: 0.0008758975884318352\n",
            "[ Evaluate epoch: 179 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.878\n",
            "Average loss: 0.0009083799174427987\n",
            "[ Evaluate epoch: 179 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.73\n",
            "Average loss: 0.005137041570246219\n",
            "Model Saved!\n",
            "[ Train epoch: 179 ]\n",
            "Train accuarcy: 95.978\n",
            "Train average loss: 0.0008890655390173197\n",
            "[ Evaluate epoch: 180 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.798\n",
            "Average loss: 0.0009076251424849033\n",
            "[ Evaluate epoch: 180 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.51\n",
            "Average loss: 0.005062251898646355\n",
            "Model Saved!\n",
            "[ Train epoch: 180 ]\n",
            "Train accuarcy: 95.892\n",
            "Train average loss: 0.0009085815672576428\n",
            "[ Evaluate epoch: 181 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.22\n",
            "Average loss: 0.0008316566959768534\n",
            "[ Evaluate epoch: 181 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.87\n",
            "Average loss: 0.004910834318399429\n",
            "Model Saved!\n",
            "[ Train epoch: 181 ]\n",
            "Train accuarcy: 96.064\n",
            "Train average loss: 0.0008838576649874449\n",
            "[ Evaluate epoch: 182 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.512\n",
            "Average loss: 0.0007705732996016741\n",
            "[ Evaluate epoch: 182 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.24\n",
            "Average loss: 0.004899155183136463\n",
            "Model Saved!\n",
            "[ Train epoch: 182 ]\n",
            "Train accuarcy: 96.022\n",
            "Train average loss: 0.0008863645755499602\n",
            "[ Evaluate epoch: 183 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.0\n",
            "Average loss: 0.0010757137917727232\n",
            "[ Evaluate epoch: 183 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.05\n",
            "Average loss: 0.005601529079675674\n",
            "Model Saved!\n",
            "[ Train epoch: 183 ]\n",
            "Train accuarcy: 96.078\n",
            "Train average loss: 0.0008725066777691245\n",
            "[ Evaluate epoch: 184 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.844\n",
            "Average loss: 0.001103364553898573\n",
            "[ Evaluate epoch: 184 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.95\n",
            "Average loss: 0.005772867661714554\n",
            "Model Saved!\n",
            "[ Train epoch: 184 ]\n",
            "Train accuarcy: 96.218\n",
            "Train average loss: 0.000864395846016705\n",
            "[ Evaluate epoch: 185 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.972\n",
            "Average loss: 0.0008947306536138058\n",
            "[ Evaluate epoch: 185 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.24\n",
            "Average loss: 0.005076490986347199\n",
            "Model Saved!\n",
            "[ Train epoch: 185 ]\n",
            "Train accuarcy: 95.916\n",
            "Train average loss: 0.0008801327328756452\n",
            "[ Evaluate epoch: 186 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.488\n",
            "Average loss: 0.0011762519171088933\n",
            "[ Evaluate epoch: 186 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.11\n",
            "Average loss: 0.005488403987884522\n",
            "Model Saved!\n",
            "[ Train epoch: 186 ]\n",
            "Train accuarcy: 96.22\n",
            "Train average loss: 0.0008274661622941494\n",
            "[ Evaluate epoch: 187 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.512\n",
            "Average loss: 0.0009776083499193193\n",
            "[ Evaluate epoch: 187 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.23\n",
            "Average loss: 0.005435251641273498\n",
            "Model Saved!\n",
            "[ Train epoch: 187 ]\n",
            "Train accuarcy: 96.346\n",
            "Train average loss: 0.0008197418870031834\n",
            "[ Evaluate epoch: 188 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.944\n",
            "Average loss: 0.0008729425525665284\n",
            "[ Evaluate epoch: 188 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.58\n",
            "Average loss: 0.005394077463448048\n",
            "Model Saved!\n",
            "[ Train epoch: 188 ]\n",
            "Train accuarcy: 96.174\n",
            "Train average loss: 0.0008617914057523012\n",
            "[ Evaluate epoch: 189 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.712\n",
            "Average loss: 0.001159034144282341\n",
            "[ Evaluate epoch: 189 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.92\n",
            "Average loss: 0.00578606452792883\n",
            "Model Saved!\n",
            "[ Train epoch: 189 ]\n",
            "Train accuarcy: 96.322\n",
            "Train average loss: 0.0008159953295812011\n",
            "[ Evaluate epoch: 190 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.178\n",
            "Average loss: 0.0008359263784438372\n",
            "[ Evaluate epoch: 190 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.88\n",
            "Average loss: 0.005129341697692871\n",
            "Model Saved!\n",
            "[ Train epoch: 190 ]\n",
            "Train accuarcy: 96.2\n",
            "Train average loss: 0.0008429756799340248\n",
            "[ Evaluate epoch: 191 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.464\n",
            "Average loss: 0.0011724909677356481\n",
            "[ Evaluate epoch: 191 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.08\n",
            "Average loss: 0.005927043345570564\n",
            "Model Saved!\n",
            "[ Train epoch: 191 ]\n",
            "Train accuarcy: 96.33\n",
            "Train average loss: 0.0008241832600533963\n",
            "[ Evaluate epoch: 192 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.556\n",
            "Average loss: 0.0009413034410029649\n",
            "[ Evaluate epoch: 192 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.62\n",
            "Average loss: 0.005234596480429172\n",
            "Model Saved!\n",
            "[ Train epoch: 192 ]\n",
            "Train accuarcy: 96.272\n",
            "Train average loss: 0.0008231678784638643\n",
            "[ Evaluate epoch: 193 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.622\n",
            "Average loss: 0.0007567863643541933\n",
            "[ Evaluate epoch: 193 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.14\n",
            "Average loss: 0.004975267280638218\n",
            "Model Saved!\n",
            "[ Train epoch: 193 ]\n",
            "Train accuarcy: 96.36\n",
            "Train average loss: 0.0008071924554184079\n",
            "[ Evaluate epoch: 194 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.99\n",
            "Average loss: 0.0011066214552521705\n",
            "[ Evaluate epoch: 194 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.95\n",
            "Average loss: 0.005818372930586338\n",
            "Model Saved!\n",
            "[ Train epoch: 194 ]\n",
            "Train accuarcy: 96.396\n",
            "Train average loss: 0.0008005007077008485\n",
            "[ Evaluate epoch: 195 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.834\n",
            "Average loss: 0.0009010917633026838\n",
            "[ Evaluate epoch: 195 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.08\n",
            "Average loss: 0.005070348617434502\n",
            "Model Saved!\n",
            "[ Train epoch: 195 ]\n",
            "Train accuarcy: 96.436\n",
            "Train average loss: 0.000787624860405922\n",
            "[ Evaluate epoch: 196 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.798\n",
            "Average loss: 0.0007182506293058396\n",
            "[ Evaluate epoch: 196 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.02\n",
            "Average loss: 0.005063902421295643\n",
            "Model Saved!\n",
            "[ Train epoch: 196 ]\n",
            "Train accuarcy: 96.478\n",
            "Train average loss: 0.0007964004468545317\n",
            "[ Evaluate epoch: 197 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.394\n",
            "Average loss: 0.0007920889103040099\n",
            "[ Evaluate epoch: 197 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.98\n",
            "Average loss: 0.005088369265198707\n",
            "Model Saved!\n",
            "[ Train epoch: 197 ]\n",
            "Train accuarcy: 96.386\n",
            "Train average loss: 0.0008038885294646025\n",
            "[ Evaluate epoch: 198 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.418\n",
            "Average loss: 0.0010166950020566582\n",
            "[ Evaluate epoch: 198 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.24\n",
            "Average loss: 0.005643140822649002\n",
            "Model Saved!\n",
            "[ Train epoch: 198 ]\n",
            "Train accuarcy: 96.462\n",
            "Train average loss: 0.0007839756052196026\n",
            "[ Evaluate epoch: 199 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.1\n",
            "Average loss: 0.0006677962785959243\n",
            "[ Evaluate epoch: 199 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.61\n",
            "Average loss: 0.004854420457780361\n",
            "Model Saved!\n",
            "[ Train epoch: 199 ]\n",
            "Train accuarcy: 96.532\n",
            "Train average loss: 0.000774443809017539\n",
            "[ Evaluate epoch: 200 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.93\n",
            "Average loss: 0.0010657557057589292\n",
            "[ Evaluate epoch: 200 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.63\n",
            "Average loss: 0.005886369003355503\n",
            "Model Saved!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_ak5GxErzBJ3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 280
        },
        "outputId": "e776709d-71ad-4cb6-9249-a58d21d6bcf0"
      },
      "source": [
        "# 정확도(accuracy) 커브 시각화\n",
        "plt.plot([i for i in range(epoch + 1)], [i[0] for i in train_result])\n",
        "plt.plot([i for i in range(epoch + 1)], [i[0] for i in test_result])\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.legend([\"train\", \"test\"])\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6_Z98IiuzBcD",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "outputId": "6ec8a8ad-a918-4731-9cda-5db4dc9c5410"
      },
      "source": [
        "# 손실(loss) 커브 시각화\n",
        "plt.plot([i for i in range(epoch + 1)], [i[1] for i in train_result])\n",
        "plt.plot([i for i in range(epoch + 1)], [i[1] for i in test_result])\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"Loss\")\n",
        "plt.legend([\"train\", \"test\"])\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GrhGmgGNzCxC",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 568
        },
        "outputId": "507e24f8-20cd-48aa-81c2-9fe01de35abd"
      },
      "source": [
        "# 혼동 행렬(Confusion Matrix) 시각화 (학습 데이터셋)\n",
        "net = ResNet20()\n",
        "net = net.to(device)\n",
        "\n",
        "file_name = \"./checkpoint/ResNet20_CIFAR10.pt\"\n",
        "checkpoint = torch.load(file_name)\n",
        "net.load_state_dict(checkpoint['net'])\n",
        "\n",
        "confusion_matrix = get_confusion_matrix(net, 10, train_loader)\n",
        "print(\"[ 각 클래스당 데이터 개수 ]\")\n",
        "print(confusion_matrix.sum(1))\n",
        "\n",
        "print(\"[ 혼동 행렬(confusion matrix) 시각화 ]\")\n",
        "# 행(row)은 실제 레이블, 열(column)은 모델이 분류한 레이블\n",
        "res = pd.DataFrame(confusion_matrix.numpy(), index = [i for i in range(10)], columns = [i for i in range(10)])\n",
        "plt.figure(figsize = (10, 7))\n",
        "sn.heatmap(res, annot=True)\n",
        "plt.show()\n",
        "\n",
        "print(\"[ 각 클래스에 따른 정확도 ]\")\n",
        "# (각 클래스마다 정답 개수 / 각 클래스마다 데이터의 개수)\n",
        "print(confusion_matrix.diag() / confusion_matrix.sum(1))\n",
        "\n",
        "print(\"[ 전체 평균 정확도 ]\")\n",
        "print(confusion_matrix.diag().sum() / confusion_matrix.sum())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[ 각 클래스당 데이터 개수 ]\n",
            "tensor([5000., 5000., 5000., 5000., 5000., 5000., 5000., 5000., 5000., 5000.])\n",
            "[ 혼동 행렬(confusion matrix) 시각화 ]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x504 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "[ 각 클래스에 따른 정확도 ]\n",
            "tensor([0.9526, 0.9776, 0.9840, 0.9672, 0.9618, 0.8174, 0.8890, 0.9922, 0.9910,\n",
            "        0.9686])\n",
            "[ 전체 평균 정확도 ]\n",
            "tensor(0.9501)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Xa5G-r4QzM9O",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 568
        },
        "outputId": "18b2d2c1-8905-407f-bfbc-7ef3e76f8b1a"
      },
      "source": [
        "# 혼동 행렬(Confusion Matrix) 시각화 (테스트 데이터셋)\n",
        "net = ResNet20()\n",
        "net = net.to(device)\n",
        "\n",
        "file_name = \"./checkpoint/ResNet20_CIFAR10.pt\"\n",
        "checkpoint = torch.load(file_name)\n",
        "net.load_state_dict(checkpoint['net'])\n",
        "\n",
        "confusion_matrix = get_confusion_matrix(net, 10, test_loader)\n",
        "print(\"[ 각 클래스당 데이터 개수 ]\")\n",
        "print(confusion_matrix.sum(1))\n",
        "\n",
        "print(\"[ 혼동 행렬(confusion matrix) 시각화 ]\")\n",
        "# 행(row)은 실제 레이블, 열(column)은 모델이 분류한 레이블\n",
        "res = pd.DataFrame(confusion_matrix.numpy(), index = [i for i in range(10)], columns = [i for i in range(10)])\n",
        "plt.figure(figsize = (10, 7))\n",
        "sn.heatmap(res, annot=True)\n",
        "plt.show()\n",
        "\n",
        "print(\"[ 각 클래스에 따른 정확도 ]\")\n",
        "# (각 클래스마다 정답 개수 / 각 클래스마다 데이터의 개수)\n",
        "print(confusion_matrix.diag() / confusion_matrix.sum(1))\n",
        "\n",
        "print(\"[ 전체 평균 정확도 ]\")\n",
        "print(confusion_matrix.diag().sum() / confusion_matrix.sum())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[ 각 클래스당 데이터 개수 ]\n",
            "tensor([1000., 1000., 1000., 1000., 1000., 1000., 1000., 1000., 1000., 1000.])\n",
            "[ 혼동 행렬(confusion matrix) 시각화 ]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x504 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "[ 각 클래스에 따른 정확도 ]\n",
            "tensor([0.8460, 0.9280, 0.8790, 0.8190, 0.8580, 0.6780, 0.7840, 0.9330, 0.9490,\n",
            "        0.8890])\n",
            "[ 전체 평균 정확도 ]\n",
            "tensor(0.8563)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-BB4Ldngz5J4"
      },
      "source": [
        "#### 결과 확인: ResNet44 결과 분석"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dYLyfXb8z5wN",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5f37fbd6-7a6e-4f31-fdf4-5c0b33ef02b0"
      },
      "source": [
        "# 학습 과정 중에 천천히 학습률(learning rate) 감소 시키기\n",
        "def adjust_learning_rate(optimizer, epoch):\n",
        "    lr = learning_rate\n",
        "    if epoch >= 100:\n",
        "        lr /= 10\n",
        "    if epoch >= 150:\n",
        "        lr /= 10\n",
        "    for param_group in optimizer.param_groups:\n",
        "        param_group['lr'] = lr\n",
        "\n",
        "\n",
        "net = ResNet44()\n",
        "net = net.to(device)\n",
        "\n",
        "epoch = 200\n",
        "learning_rate = 0.1\n",
        "file_name = \"ResNet44_CIFAR10.pt\"\n",
        "\n",
        "criterion = nn.CrossEntropyLoss()\n",
        "optimizer = optim.SGD(net.parameters(), lr=learning_rate, momentum=0.9, weight_decay=0.0002)\n",
        "\n",
        "train_result = []\n",
        "test_result = []\n",
        "\n",
        "train_result.append(evaluate(net, 0, file_name, train_loader, \"Train\"))\n",
        "test_result.append(evaluate(net, 0, file_name, test_loader, \"Test\"))\n",
        "\n",
        "for i in range(epoch):\n",
        "    adjust_learning_rate(optimizer, epoch)\n",
        "    train(net, i, optimizer, criterion, train_loader)\n",
        "    train_acc, train_loss = evaluate(net, i + 1, file_name, train_loader, \"Train\")\n",
        "    test_acc, test_loss = evaluate(net, i + 1, file_name, test_loader, \"Test\")\n",
        "\n",
        "    state = {\n",
        "        'net': net.state_dict()\n",
        "    }\n",
        "    if not os.path.isdir('checkpoint'):\n",
        "        os.mkdir('checkpoint')\n",
        "    torch.save(state, './checkpoint/' + file_name)\n",
        "    print('Model Saved!')\n",
        "\n",
        "    train_result.append((train_acc, train_loss))\n",
        "    test_result.append((test_acc, test_loss))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[ Evaluate epoch: 0 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 8.23\n",
            "Average loss: 0.01804466058731079\n",
            "[ Evaluate epoch: 0 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 7.53\n",
            "Average loss: 0.023054070711135865\n",
            "[ Train epoch: 0 ]\n",
            "Train accuarcy: 34.376\n",
            "Train average loss: 0.013844310224056244\n",
            "[ Evaluate epoch: 1 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 43.406\n",
            "Average loss: 0.012007368226051331\n",
            "[ Evaluate epoch: 1 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 45.17\n",
            "Average loss: 0.014778021264076234\n",
            "Model Saved!\n",
            "[ Train epoch: 1 ]\n",
            "Train accuarcy: 47.87\n",
            "Train average loss: 0.011068864789009094\n",
            "[ Evaluate epoch: 2 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 49.216\n",
            "Average loss: 0.010706596255302429\n",
            "[ Evaluate epoch: 2 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 51.34\n",
            "Average loss: 0.013130501890182495\n",
            "Model Saved!\n",
            "[ Train epoch: 2 ]\n",
            "Train accuarcy: 54.96\n",
            "Train average loss: 0.009680954512357712\n",
            "[ Evaluate epoch: 3 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 55.874\n",
            "Average loss: 0.009642336746454238\n",
            "[ Evaluate epoch: 3 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 57.04\n",
            "Average loss: 0.012065028142929077\n",
            "Model Saved!\n",
            "[ Train epoch: 3 ]\n",
            "Train accuarcy: 59.612\n",
            "Train average loss: 0.00872284465432167\n",
            "[ Evaluate epoch: 4 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 59.422\n",
            "Average loss: 0.008824412877559662\n",
            "[ Evaluate epoch: 4 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 60.14\n",
            "Average loss: 0.011278579026460648\n",
            "Model Saved!\n",
            "[ Train epoch: 4 ]\n",
            "Train accuarcy: 63.276\n",
            "Train average loss: 0.007972863472700119\n",
            "[ Evaluate epoch: 5 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 62.02\n",
            "Average loss: 0.008223439651727676\n",
            "[ Evaluate epoch: 5 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 61.79\n",
            "Average loss: 0.010675575828552246\n",
            "Model Saved!\n",
            "[ Train epoch: 5 ]\n",
            "Train accuarcy: 65.832\n",
            "Train average loss: 0.007447832851409912\n",
            "[ Evaluate epoch: 6 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 66.08\n",
            "Average loss: 0.007391515418291092\n",
            "[ Evaluate epoch: 6 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 65.53\n",
            "Average loss: 0.009607107192277909\n",
            "Model Saved!\n",
            "[ Train epoch: 6 ]\n",
            "Train accuarcy: 68.106\n",
            "Train average loss: 0.006989872361421585\n",
            "[ Evaluate epoch: 7 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 67.946\n",
            "Average loss: 0.007159084764719009\n",
            "[ Evaluate epoch: 7 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 67.35\n",
            "Average loss: 0.009321731293201447\n",
            "Model Saved!\n",
            "[ Train epoch: 7 ]\n",
            "Train accuarcy: 70.294\n",
            "Train average loss: 0.006577573331594467\n",
            "[ Evaluate epoch: 8 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 69.092\n",
            "Average loss: 0.006849699809551239\n",
            "[ Evaluate epoch: 8 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 68.5\n",
            "Average loss: 0.009159179067611694\n",
            "Model Saved!\n",
            "[ Train epoch: 8 ]\n",
            "Train accuarcy: 71.76\n",
            "Train average loss: 0.006236516612768173\n",
            "[ Evaluate epoch: 9 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 72.652\n",
            "Average loss: 0.0060601484811306\n",
            "[ Evaluate epoch: 9 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 71.43\n",
            "Average loss: 0.008192399597167969\n",
            "Model Saved!\n",
            "[ Train epoch: 9 ]\n",
            "Train accuarcy: 73.008\n",
            "Train average loss: 0.005949490607976914\n",
            "[ Evaluate epoch: 10 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 73.192\n",
            "Average loss: 0.005865816202163697\n",
            "[ Evaluate epoch: 10 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 71.55\n",
            "Average loss: 0.00819763902425766\n",
            "Model Saved!\n",
            "[ Train epoch: 10 ]\n",
            "Train accuarcy: 74.574\n",
            "Train average loss: 0.005674806401729584\n",
            "[ Evaluate epoch: 11 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 75.156\n",
            "Average loss: 0.00549000623703003\n",
            "[ Evaluate epoch: 11 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 73.58\n",
            "Average loss: 0.007640939337015152\n",
            "Model Saved!\n",
            "[ Train epoch: 11 ]\n",
            "Train accuarcy: 75.612\n",
            "Train average loss: 0.005373238849043846\n",
            "[ Evaluate epoch: 12 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 75.794\n",
            "Average loss: 0.005384901758432388\n",
            "[ Evaluate epoch: 12 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 75.14\n",
            "Average loss: 0.007497911894321442\n",
            "Model Saved!\n",
            "[ Train epoch: 12 ]\n",
            "Train accuarcy: 76.774\n",
            "Train average loss: 0.005188979137539864\n",
            "[ Evaluate epoch: 13 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 77.836\n",
            "Average loss: 0.0049360151779651645\n",
            "[ Evaluate epoch: 13 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 76.28\n",
            "Average loss: 0.006826666206121445\n",
            "Model Saved!\n",
            "[ Train epoch: 13 ]\n",
            "Train accuarcy: 77.916\n",
            "Train average loss: 0.00494793243587017\n",
            "[ Evaluate epoch: 14 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 74.986\n",
            "Average loss: 0.005514346307516098\n",
            "[ Evaluate epoch: 14 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 73.24\n",
            "Average loss: 0.00805829006433487\n",
            "Model Saved!\n",
            "[ Train epoch: 14 ]\n",
            "Train accuarcy: 78.78\n",
            "Train average loss: 0.0047443689739704135\n",
            "[ Evaluate epoch: 15 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 75.026\n",
            "Average loss: 0.005568043448925019\n",
            "[ Evaluate epoch: 15 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 73.56\n",
            "Average loss: 0.008040655273199082\n",
            "Model Saved!\n",
            "[ Train epoch: 15 ]\n",
            "Train accuarcy: 79.578\n",
            "Train average loss: 0.004546471909284592\n",
            "[ Evaluate epoch: 16 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 77.95\n",
            "Average loss: 0.0049271444135904315\n",
            "[ Evaluate epoch: 16 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 75.68\n",
            "Average loss: 0.007337997171282768\n",
            "Model Saved!\n",
            "[ Train epoch: 16 ]\n",
            "Train accuarcy: 80.094\n",
            "Train average loss: 0.004431230359077454\n",
            "[ Evaluate epoch: 17 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 80.732\n",
            "Average loss: 0.004300510830879211\n",
            "[ Evaluate epoch: 17 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 78.96\n",
            "Average loss: 0.006390996918082237\n",
            "Model Saved!\n",
            "[ Train epoch: 17 ]\n",
            "Train accuarcy: 80.7\n",
            "Train average loss: 0.004284186972975731\n",
            "[ Evaluate epoch: 18 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 80.484\n",
            "Average loss: 0.004362955002188682\n",
            "[ Evaluate epoch: 18 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 77.58\n",
            "Average loss: 0.006590753528475762\n",
            "Model Saved!\n",
            "[ Train epoch: 18 ]\n",
            "Train accuarcy: 81.594\n",
            "Train average loss: 0.004126073785424233\n",
            "[ Evaluate epoch: 19 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 79.862\n",
            "Average loss: 0.004491043961644172\n",
            "[ Evaluate epoch: 19 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 77.47\n",
            "Average loss: 0.006895955061912536\n",
            "Model Saved!\n",
            "[ Train epoch: 19 ]\n",
            "Train accuarcy: 82.074\n",
            "Train average loss: 0.004007879435420037\n",
            "[ Evaluate epoch: 20 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 81.276\n",
            "Average loss: 0.004157116215825081\n",
            "[ Evaluate epoch: 20 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 78.55\n",
            "Average loss: 0.00632230678498745\n",
            "Model Saved!\n",
            "[ Train epoch: 20 ]\n",
            "Train accuarcy: 82.692\n",
            "Train average loss: 0.00390907552421093\n",
            "[ Evaluate epoch: 21 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 81.75\n",
            "Average loss: 0.004052890666723251\n",
            "[ Evaluate epoch: 21 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 78.6\n",
            "Average loss: 0.006333705779910087\n",
            "Model Saved!\n",
            "[ Train epoch: 21 ]\n",
            "Train accuarcy: 83.074\n",
            "Train average loss: 0.003779558053016663\n",
            "[ Evaluate epoch: 22 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 84.126\n",
            "Average loss: 0.003618221559524536\n",
            "[ Evaluate epoch: 22 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 80.68\n",
            "Average loss: 0.0057089841336011886\n",
            "Model Saved!\n",
            "[ Train epoch: 22 ]\n",
            "Train accuarcy: 83.39\n",
            "Train average loss: 0.003701629781126976\n",
            "[ Evaluate epoch: 23 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 80.986\n",
            "Average loss: 0.0042530653470754624\n",
            "[ Evaluate epoch: 23 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 79.11\n",
            "Average loss: 0.006473427373170853\n",
            "Model Saved!\n",
            "[ Train epoch: 23 ]\n",
            "Train accuarcy: 83.974\n",
            "Train average loss: 0.003570171433985233\n",
            "[ Evaluate epoch: 24 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 77.2\n",
            "Average loss: 0.005240363914370537\n",
            "[ Evaluate epoch: 24 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 73.51\n",
            "Average loss: 0.008640158802270889\n",
            "Model Saved!\n",
            "[ Train epoch: 24 ]\n",
            "Train accuarcy: 84.298\n",
            "Train average loss: 0.0035046186330914496\n",
            "[ Evaluate epoch: 25 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 81.074\n",
            "Average loss: 0.004298709696531296\n",
            "[ Evaluate epoch: 25 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 77.96\n",
            "Average loss: 0.0068423412472009655\n",
            "Model Saved!\n",
            "[ Train epoch: 25 ]\n",
            "Train accuarcy: 84.748\n",
            "Train average loss: 0.0034271667033433915\n",
            "[ Evaluate epoch: 26 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 85.33\n",
            "Average loss: 0.0032785133814811707\n",
            "[ Evaluate epoch: 26 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 81.12\n",
            "Average loss: 0.0055242792189121245\n",
            "Model Saved!\n",
            "[ Train epoch: 26 ]\n",
            "Train accuarcy: 85.15\n",
            "Train average loss: 0.003330694009959698\n",
            "[ Evaluate epoch: 27 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 81.166\n",
            "Average loss: 0.004133918659687042\n",
            "[ Evaluate epoch: 27 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 78.59\n",
            "Average loss: 0.006802671140432358\n",
            "Model Saved!\n",
            "[ Train epoch: 27 ]\n",
            "Train accuarcy: 85.49\n",
            "Train average loss: 0.0032692204150557517\n",
            "[ Evaluate epoch: 28 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 83.154\n",
            "Average loss: 0.0038201028853654864\n",
            "[ Evaluate epoch: 28 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 79.8\n",
            "Average loss: 0.0063234081506729125\n",
            "Model Saved!\n",
            "[ Train epoch: 28 ]\n",
            "Train accuarcy: 85.656\n",
            "Train average loss: 0.0032064072865247728\n",
            "[ Evaluate epoch: 29 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 84.936\n",
            "Average loss: 0.0033529307237267496\n",
            "[ Evaluate epoch: 29 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 81.41\n",
            "Average loss: 0.005853905063867569\n",
            "Model Saved!\n",
            "[ Train epoch: 29 ]\n",
            "Train accuarcy: 86.126\n",
            "Train average loss: 0.0031062495067715645\n",
            "[ Evaluate epoch: 30 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.956\n",
            "Average loss: 0.002934105046093464\n",
            "[ Evaluate epoch: 30 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.75\n",
            "Average loss: 0.005297805017232895\n",
            "Model Saved!\n",
            "[ Train epoch: 30 ]\n",
            "Train accuarcy: 86.542\n",
            "Train average loss: 0.0030136006408929823\n",
            "[ Evaluate epoch: 31 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 85.776\n",
            "Average loss: 0.003162233197391033\n",
            "[ Evaluate epoch: 31 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.02\n",
            "Average loss: 0.005627511838078499\n",
            "Model Saved!\n",
            "[ Train epoch: 31 ]\n",
            "Train accuarcy: 86.696\n",
            "Train average loss: 0.0029977242866158485\n",
            "[ Evaluate epoch: 32 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 85.016\n",
            "Average loss: 0.003387478659451008\n",
            "[ Evaluate epoch: 32 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 80.99\n",
            "Average loss: 0.006129927095770836\n",
            "Model Saved!\n",
            "[ Train epoch: 32 ]\n",
            "Train accuarcy: 87.022\n",
            "Train average loss: 0.0029083969327807426\n",
            "[ Evaluate epoch: 33 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 83.414\n",
            "Average loss: 0.003824210302233696\n",
            "[ Evaluate epoch: 33 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 79.56\n",
            "Average loss: 0.006900449362397194\n",
            "Model Saved!\n",
            "[ Train epoch: 33 ]\n",
            "Train accuarcy: 87.386\n",
            "Train average loss: 0.0028106034380197525\n",
            "[ Evaluate epoch: 34 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.086\n",
            "Average loss: 0.0028574617579579355\n",
            "[ Evaluate epoch: 34 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.32\n",
            "Average loss: 0.005440272080898285\n",
            "Model Saved!\n",
            "[ Train epoch: 34 ]\n",
            "Train accuarcy: 87.568\n",
            "Train average loss: 0.0027895716997981073\n",
            "[ Evaluate epoch: 35 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 84.068\n",
            "Average loss: 0.0035477499574422837\n",
            "[ Evaluate epoch: 35 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 80.47\n",
            "Average loss: 0.006461804163455963\n",
            "Model Saved!\n",
            "[ Train epoch: 35 ]\n",
            "Train accuarcy: 87.472\n",
            "Train average loss: 0.00277077611297369\n",
            "[ Evaluate epoch: 36 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.93\n",
            "Average loss: 0.002961100376546383\n",
            "[ Evaluate epoch: 36 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.36\n",
            "Average loss: 0.005580607256293297\n",
            "Model Saved!\n",
            "[ Train epoch: 36 ]\n",
            "Train accuarcy: 87.794\n",
            "Train average loss: 0.0027129651746153833\n",
            "[ Evaluate epoch: 37 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.444\n",
            "Average loss: 0.0028238293623924255\n",
            "[ Evaluate epoch: 37 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.22\n",
            "Average loss: 0.005589551997184753\n",
            "Model Saved!\n",
            "[ Train epoch: 37 ]\n",
            "Train accuarcy: 88.254\n",
            "Train average loss: 0.0026372181883454323\n",
            "[ Evaluate epoch: 38 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.002\n",
            "Average loss: 0.0029185536152124404\n",
            "[ Evaluate epoch: 38 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.57\n",
            "Average loss: 0.005727735763788223\n",
            "Model Saved!\n",
            "[ Train epoch: 38 ]\n",
            "Train accuarcy: 88.588\n",
            "Train average loss: 0.0025574179300665856\n",
            "[ Evaluate epoch: 39 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.554\n",
            "Average loss: 0.0025274723947048185\n",
            "[ Evaluate epoch: 39 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.19\n",
            "Average loss: 0.005171592143177986\n",
            "Model Saved!\n",
            "[ Train epoch: 39 ]\n",
            "Train accuarcy: 88.594\n",
            "Train average loss: 0.0025450206834077833\n",
            "[ Evaluate epoch: 40 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 86.84\n",
            "Average loss: 0.0029069816088676454\n",
            "[ Evaluate epoch: 40 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.29\n",
            "Average loss: 0.005736093297600746\n",
            "Model Saved!\n",
            "[ Train epoch: 40 ]\n",
            "Train accuarcy: 88.848\n",
            "Train average loss: 0.002504960038661957\n",
            "[ Evaluate epoch: 41 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.17\n",
            "Average loss: 0.002484478429555893\n",
            "[ Evaluate epoch: 41 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.11\n",
            "Average loss: 0.00495948826521635\n",
            "Model Saved!\n",
            "[ Train epoch: 41 ]\n",
            "Train accuarcy: 89.18\n",
            "Train average loss: 0.002444927162826061\n",
            "[ Evaluate epoch: 42 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.116\n",
            "Average loss: 0.0024461315870285033\n",
            "[ Evaluate epoch: 42 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.63\n",
            "Average loss: 0.00514005831182003\n",
            "Model Saved!\n",
            "[ Train epoch: 42 ]\n",
            "Train accuarcy: 89.298\n",
            "Train average loss: 0.0023958534309267998\n",
            "[ Evaluate epoch: 43 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.918\n",
            "Average loss: 0.0024745049118995665\n",
            "[ Evaluate epoch: 43 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.46\n",
            "Average loss: 0.005420363393425941\n",
            "Model Saved!\n",
            "[ Train epoch: 43 ]\n",
            "Train accuarcy: 89.494\n",
            "Train average loss: 0.002338610978126526\n",
            "[ Evaluate epoch: 44 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.238\n",
            "Average loss: 0.00239520997017622\n",
            "[ Evaluate epoch: 44 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.78\n",
            "Average loss: 0.005239891415834427\n",
            "Model Saved!\n",
            "[ Train epoch: 44 ]\n",
            "Train accuarcy: 89.418\n",
            "Train average loss: 0.002335878254175186\n",
            "[ Evaluate epoch: 45 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.728\n",
            "Average loss: 0.0024902655392885207\n",
            "[ Evaluate epoch: 45 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.7\n",
            "Average loss: 0.005213559228181839\n",
            "Model Saved!\n",
            "[ Train epoch: 45 ]\n",
            "Train accuarcy: 89.936\n",
            "Train average loss: 0.0022251523214578627\n",
            "[ Evaluate epoch: 46 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.35\n",
            "Average loss: 0.002548605869114399\n",
            "[ Evaluate epoch: 46 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.71\n",
            "Average loss: 0.005594812086224556\n",
            "Model Saved!\n",
            "[ Train epoch: 46 ]\n",
            "Train accuarcy: 90.204\n",
            "Train average loss: 0.0022013918802142145\n",
            "[ Evaluate epoch: 47 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.04\n",
            "Average loss: 0.0024505753263831137\n",
            "[ Evaluate epoch: 47 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.09\n",
            "Average loss: 0.005560177725553513\n",
            "Model Saved!\n",
            "[ Train epoch: 47 ]\n",
            "Train accuarcy: 90.332\n",
            "Train average loss: 0.002164574733376503\n",
            "[ Evaluate epoch: 48 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.158\n",
            "Average loss: 0.0029248421216011046\n",
            "[ Evaluate epoch: 48 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 81.89\n",
            "Average loss: 0.00627820394039154\n",
            "Model Saved!\n",
            "[ Train epoch: 48 ]\n",
            "Train accuarcy: 90.416\n",
            "Train average loss: 0.0021498558327555655\n",
            "[ Evaluate epoch: 49 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 88.546\n",
            "Average loss: 0.002518880675435066\n",
            "[ Evaluate epoch: 49 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.94\n",
            "Average loss: 0.005762884190678596\n",
            "Model Saved!\n",
            "[ Train epoch: 49 ]\n",
            "Train accuarcy: 90.446\n",
            "Train average loss: 0.0021121452781558037\n",
            "[ Evaluate epoch: 50 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.498\n",
            "Average loss: 0.002080124515593052\n",
            "[ Evaluate epoch: 50 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.93\n",
            "Average loss: 0.004908264765143395\n",
            "Model Saved!\n",
            "[ Train epoch: 50 ]\n",
            "Train accuarcy: 90.544\n",
            "Train average loss: 0.0021097986751794813\n",
            "[ Evaluate epoch: 51 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.722\n",
            "Average loss: 0.0022349821528792383\n",
            "[ Evaluate epoch: 51 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.12\n",
            "Average loss: 0.005203643643856048\n",
            "Model Saved!\n",
            "[ Train epoch: 51 ]\n",
            "Train accuarcy: 90.826\n",
            "Train average loss: 0.002034930626153946\n",
            "[ Evaluate epoch: 52 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.686\n",
            "Average loss: 0.0020224190591275693\n",
            "[ Evaluate epoch: 52 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.0\n",
            "Average loss: 0.004761704675853252\n",
            "Model Saved!\n",
            "[ Train epoch: 52 ]\n",
            "Train accuarcy: 91.214\n",
            "Train average loss: 0.0019861243329942226\n",
            "[ Evaluate epoch: 53 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.122\n",
            "Average loss: 0.0021680285950005053\n",
            "[ Evaluate epoch: 53 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.16\n",
            "Average loss: 0.00541203243136406\n",
            "Model Saved!\n",
            "[ Train epoch: 53 ]\n",
            "Train accuarcy: 91.002\n",
            "Train average loss: 0.001990431120246649\n",
            "[ Evaluate epoch: 54 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.002\n",
            "Average loss: 0.0022015809170901775\n",
            "[ Evaluate epoch: 54 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.93\n",
            "Average loss: 0.005426722632348537\n",
            "Model Saved!\n",
            "[ Train epoch: 54 ]\n",
            "Train accuarcy: 91.538\n",
            "Train average loss: 0.0018826856657862662\n",
            "[ Evaluate epoch: 55 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.716\n",
            "Average loss: 0.0020381670549511908\n",
            "[ Evaluate epoch: 55 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.55\n",
            "Average loss: 0.005213813310861588\n",
            "Model Saved!\n",
            "[ Train epoch: 55 ]\n",
            "Train accuarcy: 91.546\n",
            "Train average loss: 0.001885304813235998\n",
            "[ Evaluate epoch: 56 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.45\n",
            "Average loss: 0.002119580723643303\n",
            "[ Evaluate epoch: 56 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.29\n",
            "Average loss: 0.005499856051802635\n",
            "Model Saved!\n",
            "[ Train epoch: 56 ]\n",
            "Train accuarcy: 91.676\n",
            "Train average loss: 0.0018537746070325374\n",
            "[ Evaluate epoch: 57 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.798\n",
            "Average loss: 0.002025262168645859\n",
            "[ Evaluate epoch: 57 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.02\n",
            "Average loss: 0.005214417040348053\n",
            "Model Saved!\n",
            "[ Train epoch: 57 ]\n",
            "Train accuarcy: 91.674\n",
            "Train average loss: 0.00184184207290411\n",
            "[ Evaluate epoch: 58 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 87.528\n",
            "Average loss: 0.0028372813925147057\n",
            "[ Evaluate epoch: 58 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 81.12\n",
            "Average loss: 0.006782781356573105\n",
            "Model Saved!\n",
            "[ Train epoch: 58 ]\n",
            "Train accuarcy: 91.856\n",
            "Train average loss: 0.0018003385865688324\n",
            "[ Evaluate epoch: 59 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.512\n",
            "Average loss: 0.002081736592501402\n",
            "[ Evaluate epoch: 59 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.19\n",
            "Average loss: 0.00536846081763506\n",
            "Model Saved!\n",
            "[ Train epoch: 59 ]\n",
            "Train accuarcy: 91.84\n",
            "Train average loss: 0.001788297730833292\n",
            "[ Evaluate epoch: 60 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.79\n",
            "Average loss: 0.0022792879623174667\n",
            "[ Evaluate epoch: 60 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.69\n",
            "Average loss: 0.005656598398089409\n",
            "Model Saved!\n",
            "[ Train epoch: 60 ]\n",
            "Train accuarcy: 92.264\n",
            "Train average loss: 0.001717493349313736\n",
            "[ Evaluate epoch: 61 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 89.312\n",
            "Average loss: 0.0023401007474958897\n",
            "[ Evaluate epoch: 61 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.63\n",
            "Average loss: 0.006203741389513016\n",
            "Model Saved!\n",
            "[ Train epoch: 61 ]\n",
            "Train accuarcy: 92.222\n",
            "Train average loss: 0.0017143058088421822\n",
            "[ Evaluate epoch: 62 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.684\n",
            "Average loss: 0.002049887252599001\n",
            "[ Evaluate epoch: 62 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.76\n",
            "Average loss: 0.005923941278457642\n",
            "Model Saved!\n",
            "[ Train epoch: 62 ]\n",
            "Train accuarcy: 92.22\n",
            "Train average loss: 0.0016922820909321308\n",
            "[ Evaluate epoch: 63 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.184\n",
            "Average loss: 0.0017025005556643008\n",
            "[ Evaluate epoch: 63 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.42\n",
            "Average loss: 0.00488553102016449\n",
            "Model Saved!\n",
            "[ Train epoch: 63 ]\n",
            "Train accuarcy: 92.42\n",
            "Train average loss: 0.0016512054325640202\n",
            "[ Evaluate epoch: 64 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.124\n",
            "Average loss: 0.0021848167580366135\n",
            "[ Evaluate epoch: 64 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.59\n",
            "Average loss: 0.005749639391899109\n",
            "Model Saved!\n",
            "[ Train epoch: 64 ]\n",
            "Train accuarcy: 92.458\n",
            "Train average loss: 0.0016431005641818046\n",
            "[ Evaluate epoch: 65 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.838\n",
            "Average loss: 0.001788975155800581\n",
            "[ Evaluate epoch: 65 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.27\n",
            "Average loss: 0.005260112681984902\n",
            "Model Saved!\n",
            "[ Train epoch: 65 ]\n",
            "Train accuarcy: 92.842\n",
            "Train average loss: 0.0015787538777291775\n",
            "[ Evaluate epoch: 66 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.636\n",
            "Average loss: 0.0016521554888784885\n",
            "[ Evaluate epoch: 66 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.39\n",
            "Average loss: 0.005133342500030995\n",
            "Model Saved!\n",
            "[ Train epoch: 66 ]\n",
            "Train accuarcy: 92.778\n",
            "Train average loss: 0.0015796556916832924\n",
            "[ Evaluate epoch: 67 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.48\n",
            "Average loss: 0.0016526510429382324\n",
            "[ Evaluate epoch: 67 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.08\n",
            "Average loss: 0.005258055791258812\n",
            "Model Saved!\n",
            "[ Train epoch: 67 ]\n",
            "Train accuarcy: 93.018\n",
            "Train average loss: 0.0015592386008799076\n",
            "[ Evaluate epoch: 68 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.586\n",
            "Average loss: 0.0018237745414674282\n",
            "[ Evaluate epoch: 68 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.48\n",
            "Average loss: 0.005592832140624523\n",
            "Model Saved!\n",
            "[ Train epoch: 68 ]\n",
            "Train accuarcy: 93.042\n",
            "Train average loss: 0.0015129969069361687\n",
            "[ Evaluate epoch: 69 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.924\n",
            "Average loss: 0.001748849087357521\n",
            "[ Evaluate epoch: 69 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.83\n",
            "Average loss: 0.005241120727360249\n",
            "Model Saved!\n",
            "[ Train epoch: 69 ]\n",
            "Train accuarcy: 93.118\n",
            "Train average loss: 0.001516791941076517\n",
            "[ Evaluate epoch: 70 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.352\n",
            "Average loss: 0.001691150660663843\n",
            "[ Evaluate epoch: 70 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.41\n",
            "Average loss: 0.0051459244936704635\n",
            "Model Saved!\n",
            "[ Train epoch: 70 ]\n",
            "Train accuarcy: 93.258\n",
            "Train average loss: 0.001477153984606266\n",
            "[ Evaluate epoch: 71 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.84\n",
            "Average loss: 0.0015717398315668106\n",
            "[ Evaluate epoch: 71 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.22\n",
            "Average loss: 0.005098667126893997\n",
            "Model Saved!\n",
            "[ Train epoch: 71 ]\n",
            "Train accuarcy: 93.472\n",
            "Train average loss: 0.001438194748610258\n",
            "[ Evaluate epoch: 72 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.944\n",
            "Average loss: 0.0015363681218028068\n",
            "[ Evaluate epoch: 72 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.47\n",
            "Average loss: 0.005055367256700993\n",
            "Model Saved!\n",
            "[ Train epoch: 72 ]\n",
            "Train accuarcy: 93.526\n",
            "Train average loss: 0.0014184688487648965\n",
            "[ Evaluate epoch: 73 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 90.338\n",
            "Average loss: 0.0021276057444512843\n",
            "[ Evaluate epoch: 73 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 82.92\n",
            "Average loss: 0.00650305295586586\n",
            "Model Saved!\n",
            "[ Train epoch: 73 ]\n",
            "Train accuarcy: 93.524\n",
            "Train average loss: 0.0014130667174607515\n",
            "[ Evaluate epoch: 74 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.34\n",
            "Average loss: 0.0016218524162471294\n",
            "[ Evaluate epoch: 74 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.49\n",
            "Average loss: 0.005270700418949127\n",
            "Model Saved!\n",
            "[ Train epoch: 74 ]\n",
            "Train accuarcy: 93.418\n",
            "Train average loss: 0.0014284930714964868\n",
            "[ Evaluate epoch: 75 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.422\n",
            "Average loss: 0.001646749137416482\n",
            "[ Evaluate epoch: 75 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.07\n",
            "Average loss: 0.005455764116346836\n",
            "Model Saved!\n",
            "[ Train epoch: 75 ]\n",
            "Train accuarcy: 93.772\n",
            "Train average loss: 0.0013725009699165822\n",
            "[ Evaluate epoch: 76 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.482\n",
            "Average loss: 0.001428646108880639\n",
            "[ Evaluate epoch: 76 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.79\n",
            "Average loss: 0.0049497868373990055\n",
            "Model Saved!\n",
            "[ Train epoch: 76 ]\n",
            "Train accuarcy: 93.63\n",
            "Train average loss: 0.0013716715733706952\n",
            "[ Evaluate epoch: 77 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.288\n",
            "Average loss: 0.0019287694236636162\n",
            "[ Evaluate epoch: 77 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.86\n",
            "Average loss: 0.005986008714139461\n",
            "Model Saved!\n",
            "[ Train epoch: 77 ]\n",
            "Train accuarcy: 93.84\n",
            "Train average loss: 0.001337160367295146\n",
            "[ Evaluate epoch: 78 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.176\n",
            "Average loss: 0.001488988307416439\n",
            "[ Evaluate epoch: 78 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.21\n",
            "Average loss: 0.00521336729824543\n",
            "Model Saved!\n",
            "[ Train epoch: 78 ]\n",
            "Train accuarcy: 94.104\n",
            "Train average loss: 0.0013153957277536393\n",
            "[ Evaluate epoch: 79 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.258\n",
            "Average loss: 0.0012564896123111247\n",
            "[ Evaluate epoch: 79 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.59\n",
            "Average loss: 0.004848410986363888\n",
            "Model Saved!\n",
            "[ Train epoch: 79 ]\n",
            "Train accuarcy: 94.176\n",
            "Train average loss: 0.0012855841106921434\n",
            "[ Evaluate epoch: 80 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.64\n",
            "Average loss: 0.0013934110264480113\n",
            "[ Evaluate epoch: 80 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.19\n",
            "Average loss: 0.005189351931214332\n",
            "Model Saved!\n",
            "[ Train epoch: 80 ]\n",
            "Train accuarcy: 94.366\n",
            "Train average loss: 0.0012399793373793363\n",
            "[ Evaluate epoch: 81 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.6\n",
            "Average loss: 0.00139430535197258\n",
            "[ Evaluate epoch: 81 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.74\n",
            "Average loss: 0.0049904406875371935\n",
            "Model Saved!\n",
            "[ Train epoch: 81 ]\n",
            "Train accuarcy: 94.152\n",
            "Train average loss: 0.0012834596599638462\n",
            "[ Evaluate epoch: 82 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.46\n",
            "Average loss: 0.0012186082573235036\n",
            "[ Evaluate epoch: 82 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.12\n",
            "Average loss: 0.0051017107099294665\n",
            "Model Saved!\n",
            "[ Train epoch: 82 ]\n",
            "Train accuarcy: 94.462\n",
            "Train average loss: 0.001235109964609146\n",
            "[ Evaluate epoch: 83 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.594\n",
            "Average loss: 0.001596508701890707\n",
            "[ Evaluate epoch: 83 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.82\n",
            "Average loss: 0.005747061444818974\n",
            "Model Saved!\n",
            "[ Train epoch: 83 ]\n",
            "Train accuarcy: 94.73\n",
            "Train average loss: 0.001171210616454482\n",
            "[ Evaluate epoch: 84 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.656\n",
            "Average loss: 0.0014137579915672542\n",
            "[ Evaluate epoch: 84 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.9\n",
            "Average loss: 0.005648056969046592\n",
            "Model Saved!\n",
            "[ Train epoch: 84 ]\n",
            "Train accuarcy: 94.516\n",
            "Train average loss: 0.001205733936727047\n",
            "[ Evaluate epoch: 85 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.948\n",
            "Average loss: 0.001540298744738102\n",
            "[ Evaluate epoch: 85 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.58\n",
            "Average loss: 0.005760502955317498\n",
            "Model Saved!\n",
            "[ Train epoch: 85 ]\n",
            "Train accuarcy: 94.638\n",
            "Train average loss: 0.0011720949570089578\n",
            "[ Evaluate epoch: 86 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 91.028\n",
            "Average loss: 0.0019755129630863667\n",
            "[ Evaluate epoch: 86 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.25\n",
            "Average loss: 0.006372438099980354\n",
            "Model Saved!\n",
            "[ Train epoch: 86 ]\n",
            "Train accuarcy: 94.75\n",
            "Train average loss: 0.0011569720275700092\n",
            "[ Evaluate epoch: 87 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.256\n",
            "Average loss: 0.00144476747892797\n",
            "[ Evaluate epoch: 87 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.04\n",
            "Average loss: 0.005628871729969978\n",
            "Model Saved!\n",
            "[ Train epoch: 87 ]\n",
            "Train accuarcy: 94.88\n",
            "Train average loss: 0.0011354441384226084\n",
            "[ Evaluate epoch: 88 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.918\n",
            "Average loss: 0.0015193199871480466\n",
            "[ Evaluate epoch: 88 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.6\n",
            "Average loss: 0.0058277149170637135\n",
            "Model Saved!\n",
            "[ Train epoch: 88 ]\n",
            "Train accuarcy: 94.876\n",
            "Train average loss: 0.001119600675702095\n",
            "[ Evaluate epoch: 89 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.05\n",
            "Average loss: 0.0017243827579915523\n",
            "[ Evaluate epoch: 89 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.44\n",
            "Average loss: 0.006084681895375252\n",
            "Model Saved!\n",
            "[ Train epoch: 89 ]\n",
            "Train accuarcy: 94.748\n",
            "Train average loss: 0.0011371107267588378\n",
            "[ Evaluate epoch: 90 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 92.16\n",
            "Average loss: 0.0018042908564209939\n",
            "[ Evaluate epoch: 90 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 83.87\n",
            "Average loss: 0.0065032521903514865\n",
            "Model Saved!\n",
            "[ Train epoch: 90 ]\n",
            "Train accuarcy: 94.932\n",
            "Train average loss: 0.001115916849449277\n",
            "[ Evaluate epoch: 91 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.58\n",
            "Average loss: 0.0013819556519389152\n",
            "[ Evaluate epoch: 91 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.98\n",
            "Average loss: 0.005836973237991333\n",
            "Model Saved!\n",
            "[ Train epoch: 91 ]\n",
            "Train accuarcy: 94.774\n",
            "Train average loss: 0.0011267067488282918\n",
            "[ Evaluate epoch: 92 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.614\n",
            "Average loss: 0.0009595707636326551\n",
            "[ Evaluate epoch: 92 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.75\n",
            "Average loss: 0.004946257221698761\n",
            "Model Saved!\n",
            "[ Train epoch: 92 ]\n",
            "Train accuarcy: 95.244\n",
            "Train average loss: 0.001044512087404728\n",
            "[ Evaluate epoch: 93 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.334\n",
            "Average loss: 0.0014369095236063003\n",
            "[ Evaluate epoch: 93 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.42\n",
            "Average loss: 0.0056332748472690585\n",
            "Model Saved!\n",
            "[ Train epoch: 93 ]\n",
            "Train accuarcy: 95.204\n",
            "Train average loss: 0.0010503529676049947\n",
            "[ Evaluate epoch: 94 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.514\n",
            "Average loss: 0.0013939222341775893\n",
            "[ Evaluate epoch: 94 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.04\n",
            "Average loss: 0.005899302694201469\n",
            "Model Saved!\n",
            "[ Train epoch: 94 ]\n",
            "Train accuarcy: 95.34\n",
            "Train average loss: 0.0010275380639731884\n",
            "[ Evaluate epoch: 95 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.74\n",
            "Average loss: 0.0013081385825574397\n",
            "[ Evaluate epoch: 95 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.33\n",
            "Average loss: 0.005767716464400292\n",
            "Model Saved!\n",
            "[ Train epoch: 95 ]\n",
            "Train accuarcy: 95.362\n",
            "Train average loss: 0.001008937105834484\n",
            "[ Evaluate epoch: 96 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.974\n",
            "Average loss: 0.0011059470029175281\n",
            "[ Evaluate epoch: 96 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.17\n",
            "Average loss: 0.005501060512661934\n",
            "Model Saved!\n",
            "[ Train epoch: 96 ]\n",
            "Train accuarcy: 95.602\n",
            "Train average loss: 0.0009841117317229509\n",
            "[ Evaluate epoch: 97 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.172\n",
            "Average loss: 0.0012494146256148814\n",
            "[ Evaluate epoch: 97 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.51\n",
            "Average loss: 0.005693880155682564\n",
            "Model Saved!\n",
            "[ Train epoch: 97 ]\n",
            "Train accuarcy: 95.496\n",
            "Train average loss: 0.0009839397632330656\n",
            "[ Evaluate epoch: 98 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.064\n",
            "Average loss: 0.001469404170513153\n",
            "[ Evaluate epoch: 98 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.08\n",
            "Average loss: 0.00591481132209301\n",
            "Model Saved!\n",
            "[ Train epoch: 98 ]\n",
            "Train accuarcy: 95.686\n",
            "Train average loss: 0.000940311985462904\n",
            "[ Evaluate epoch: 99 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.078\n",
            "Average loss: 0.0010634742978215218\n",
            "[ Evaluate epoch: 99 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.15\n",
            "Average loss: 0.005351937940716743\n",
            "Model Saved!\n",
            "[ Train epoch: 99 ]\n",
            "Train accuarcy: 95.738\n",
            "Train average loss: 0.000938284102678299\n",
            "[ Evaluate epoch: 100 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.78\n",
            "Average loss: 0.0011346867223083973\n",
            "[ Evaluate epoch: 100 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.54\n",
            "Average loss: 0.006029051844775677\n",
            "Model Saved!\n",
            "[ Train epoch: 100 ]\n",
            "Train accuarcy: 95.648\n",
            "Train average loss: 0.0009482035843282938\n",
            "[ Evaluate epoch: 101 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.628\n",
            "Average loss: 0.0011602404803782702\n",
            "[ Evaluate epoch: 101 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.35\n",
            "Average loss: 0.0057462224006652836\n",
            "Model Saved!\n",
            "[ Train epoch: 101 ]\n",
            "Train accuarcy: 95.874\n",
            "Train average loss: 0.0009178954271227122\n",
            "[ Evaluate epoch: 102 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.758\n",
            "Average loss: 0.0011490828823298217\n",
            "[ Evaluate epoch: 102 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.71\n",
            "Average loss: 0.005642840041220188\n",
            "Model Saved!\n",
            "[ Train epoch: 102 ]\n",
            "Train accuarcy: 95.822\n",
            "Train average loss: 0.000910372471883893\n",
            "[ Evaluate epoch: 103 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.302\n",
            "Average loss: 0.0010252394711971282\n",
            "[ Evaluate epoch: 103 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.3\n",
            "Average loss: 0.0055879000484943386\n",
            "Model Saved!\n",
            "[ Train epoch: 103 ]\n",
            "Train accuarcy: 95.982\n",
            "Train average loss: 0.0008982994614914059\n",
            "[ Evaluate epoch: 104 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.65\n",
            "Average loss: 0.0011782258185744285\n",
            "[ Evaluate epoch: 104 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.2\n",
            "Average loss: 0.00597061380147934\n",
            "Model Saved!\n",
            "[ Train epoch: 104 ]\n",
            "Train accuarcy: 96.066\n",
            "Train average loss: 0.0008632195469737053\n",
            "[ Evaluate epoch: 105 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.994\n",
            "Average loss: 0.0010718216688930988\n",
            "[ Evaluate epoch: 105 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.53\n",
            "Average loss: 0.00582713760137558\n",
            "Model Saved!\n",
            "[ Train epoch: 105 ]\n",
            "Train accuarcy: 95.818\n",
            "Train average loss: 0.0008955221135914325\n",
            "[ Evaluate epoch: 106 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.364\n",
            "Average loss: 0.000806650756187737\n",
            "[ Evaluate epoch: 106 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.13\n",
            "Average loss: 0.005114860171079635\n",
            "Model Saved!\n",
            "[ Train epoch: 106 ]\n",
            "Train accuarcy: 96.124\n",
            "Train average loss: 0.0008497137026488781\n",
            "[ Evaluate epoch: 107 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.59\n",
            "Average loss: 0.0009624328216910362\n",
            "[ Evaluate epoch: 107 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.41\n",
            "Average loss: 0.005323756138980389\n",
            "Model Saved!\n",
            "[ Train epoch: 107 ]\n",
            "Train accuarcy: 96.128\n",
            "Train average loss: 0.0008390653531998396\n",
            "[ Evaluate epoch: 108 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.942\n",
            "Average loss: 0.0008977019635587931\n",
            "[ Evaluate epoch: 108 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.64\n",
            "Average loss: 0.0054799766361713406\n",
            "Model Saved!\n",
            "[ Train epoch: 108 ]\n",
            "Train accuarcy: 96.144\n",
            "Train average loss: 0.0008382703277468682\n",
            "[ Evaluate epoch: 109 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.956\n",
            "Average loss: 0.0010676596372574568\n",
            "[ Evaluate epoch: 109 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.26\n",
            "Average loss: 0.005731083680689335\n",
            "Model Saved!\n",
            "[ Train epoch: 109 ]\n",
            "Train accuarcy: 96.23\n",
            "Train average loss: 0.000831655876673758\n",
            "[ Evaluate epoch: 110 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.206\n",
            "Average loss: 0.0012581640136241914\n",
            "[ Evaluate epoch: 110 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.45\n",
            "Average loss: 0.006223372483253479\n",
            "Model Saved!\n",
            "[ Train epoch: 110 ]\n",
            "Train accuarcy: 96.46\n",
            "Train average loss: 0.0007865648367628455\n",
            "[ Evaluate epoch: 111 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.186\n",
            "Average loss: 0.0008467200704663992\n",
            "[ Evaluate epoch: 111 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.59\n",
            "Average loss: 0.005545400449633598\n",
            "Model Saved!\n",
            "[ Train epoch: 111 ]\n",
            "Train accuarcy: 96.4\n",
            "Train average loss: 0.000810506005436182\n",
            "[ Evaluate epoch: 112 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.554\n",
            "Average loss: 0.000945382149219513\n",
            "[ Evaluate epoch: 112 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.43\n",
            "Average loss: 0.00551106132119894\n",
            "Model Saved!\n",
            "[ Train epoch: 112 ]\n",
            "Train accuarcy: 96.308\n",
            "Train average loss: 0.0007928130871802569\n",
            "[ Evaluate epoch: 113 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 94.648\n",
            "Average loss: 0.0011791195859760047\n",
            "[ Evaluate epoch: 113 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.26\n",
            "Average loss: 0.006159664475917816\n",
            "Model Saved!\n",
            "[ Train epoch: 113 ]\n",
            "Train accuarcy: 96.362\n",
            "Train average loss: 0.0007966032097488642\n",
            "[ Evaluate epoch: 114 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.112\n",
            "Average loss: 0.0008421407790482044\n",
            "[ Evaluate epoch: 114 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.89\n",
            "Average loss: 0.0054296237677335735\n",
            "Model Saved!\n",
            "[ Train epoch: 114 ]\n",
            "Train accuarcy: 96.416\n",
            "Train average loss: 0.0007891757584363222\n",
            "[ Evaluate epoch: 115 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.08\n",
            "Average loss: 0.0008613425156846642\n",
            "[ Evaluate epoch: 115 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.78\n",
            "Average loss: 0.005486521269381046\n",
            "Model Saved!\n",
            "[ Train epoch: 115 ]\n",
            "Train accuarcy: 96.43\n",
            "Train average loss: 0.0007711051720008254\n",
            "[ Evaluate epoch: 116 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.848\n",
            "Average loss: 0.0006956739592552185\n",
            "[ Evaluate epoch: 116 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.86\n",
            "Average loss: 0.005441032133996487\n",
            "Model Saved!\n",
            "[ Train epoch: 116 ]\n",
            "Train accuarcy: 96.426\n",
            "Train average loss: 0.0007788203894533216\n",
            "[ Evaluate epoch: 117 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.64\n",
            "Average loss: 0.0009410482879728079\n",
            "[ Evaluate epoch: 117 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.95\n",
            "Average loss: 0.005863416367769241\n",
            "Model Saved!\n",
            "[ Train epoch: 117 ]\n",
            "Train accuarcy: 96.722\n",
            "Train average loss: 0.0007385328582674265\n",
            "[ Evaluate epoch: 118 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.192\n",
            "Average loss: 0.0010069726401567458\n",
            "[ Evaluate epoch: 118 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.57\n",
            "Average loss: 0.006143499472737312\n",
            "Model Saved!\n",
            "[ Train epoch: 118 ]\n",
            "Train accuarcy: 96.79\n",
            "Train average loss: 0.0007084739200025797\n",
            "[ Evaluate epoch: 119 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.62\n",
            "Average loss: 0.000957219440639019\n",
            "[ Evaluate epoch: 119 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.14\n",
            "Average loss: 0.005987292736768722\n",
            "Model Saved!\n",
            "[ Train epoch: 119 ]\n",
            "Train accuarcy: 96.644\n",
            "Train average loss: 0.0007281481675431132\n",
            "[ Evaluate epoch: 120 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.028\n",
            "Average loss: 0.0008382503985241056\n",
            "[ Evaluate epoch: 120 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.27\n",
            "Average loss: 0.005831572160124779\n",
            "Model Saved!\n",
            "[ Train epoch: 120 ]\n",
            "Train accuarcy: 96.684\n",
            "Train average loss: 0.000730425311177969\n",
            "[ Evaluate epoch: 121 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.968\n",
            "Average loss: 0.0008683231054246426\n",
            "[ Evaluate epoch: 121 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.38\n",
            "Average loss: 0.0057247609436511995\n",
            "Model Saved!\n",
            "[ Train epoch: 121 ]\n",
            "Train accuarcy: 96.696\n",
            "Train average loss: 0.0007070171488076448\n",
            "[ Evaluate epoch: 122 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.338\n",
            "Average loss: 0.0010011236963421107\n",
            "[ Evaluate epoch: 122 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.31\n",
            "Average loss: 0.0063072766065597535\n",
            "Model Saved!\n",
            "[ Train epoch: 122 ]\n",
            "Train accuarcy: 96.974\n",
            "Train average loss: 0.0006778865579143167\n",
            "[ Evaluate epoch: 123 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.36\n",
            "Average loss: 0.0007916771875694394\n",
            "[ Evaluate epoch: 123 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.5\n",
            "Average loss: 0.005666989347338677\n",
            "Model Saved!\n",
            "[ Train epoch: 123 ]\n",
            "Train accuarcy: 96.906\n",
            "Train average loss: 0.0006806031867116689\n",
            "[ Evaluate epoch: 124 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.886\n",
            "Average loss: 0.0006948301212862134\n",
            "[ Evaluate epoch: 124 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.06\n",
            "Average loss: 0.005363665750622749\n",
            "Model Saved!\n",
            "[ Train epoch: 124 ]\n",
            "Train accuarcy: 96.98\n",
            "Train average loss: 0.0006764390096068383\n",
            "[ Evaluate epoch: 125 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.188\n",
            "Average loss: 0.0008074305886402726\n",
            "[ Evaluate epoch: 125 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.6\n",
            "Average loss: 0.005955830125510693\n",
            "Model Saved!\n",
            "[ Train epoch: 125 ]\n",
            "Train accuarcy: 97.046\n",
            "Train average loss: 0.0006603772382251918\n",
            "[ Evaluate epoch: 126 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 93.97\n",
            "Average loss: 0.0014070528087392448\n",
            "[ Evaluate epoch: 126 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.07\n",
            "Average loss: 0.006998639157414437\n",
            "Model Saved!\n",
            "[ Train epoch: 126 ]\n",
            "Train accuarcy: 96.928\n",
            "Train average loss: 0.0006768214470148086\n",
            "[ Evaluate epoch: 127 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.522\n",
            "Average loss: 0.0007559608330763877\n",
            "[ Evaluate epoch: 127 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.77\n",
            "Average loss: 0.005561041861772537\n",
            "Model Saved!\n",
            "[ Train epoch: 127 ]\n",
            "Train accuarcy: 96.962\n",
            "Train average loss: 0.0006714208780974149\n",
            "[ Evaluate epoch: 128 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.538\n",
            "Average loss: 0.000549374239332974\n",
            "[ Evaluate epoch: 128 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.4\n",
            "Average loss: 0.005295397208631039\n",
            "Model Saved!\n",
            "[ Train epoch: 128 ]\n",
            "Train accuarcy: 97.05\n",
            "Train average loss: 0.0006493313189968467\n",
            "[ Evaluate epoch: 129 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.156\n",
            "Average loss: 0.0010439447505027055\n",
            "[ Evaluate epoch: 129 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 84.94\n",
            "Average loss: 0.006638493499159813\n",
            "Model Saved!\n",
            "[ Train epoch: 129 ]\n",
            "Train accuarcy: 96.998\n",
            "Train average loss: 0.0006505800746381282\n",
            "[ Evaluate epoch: 130 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.312\n",
            "Average loss: 0.0006025701561383903\n",
            "[ Evaluate epoch: 130 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.28\n",
            "Average loss: 0.005527837370336056\n",
            "Model Saved!\n",
            "[ Train epoch: 130 ]\n",
            "Train accuarcy: 97.106\n",
            "Train average loss: 0.0006391004339233041\n",
            "[ Evaluate epoch: 131 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.594\n",
            "Average loss: 0.0009270183996111154\n",
            "[ Evaluate epoch: 131 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.96\n",
            "Average loss: 0.006306096521019935\n",
            "Model Saved!\n",
            "[ Train epoch: 131 ]\n",
            "Train accuarcy: 97.142\n",
            "Train average loss: 0.0006343139003589749\n",
            "[ Evaluate epoch: 132 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.526\n",
            "Average loss: 0.0007668824731186032\n",
            "[ Evaluate epoch: 132 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.97\n",
            "Average loss: 0.005758191893994808\n",
            "Model Saved!\n",
            "[ Train epoch: 132 ]\n",
            "Train accuarcy: 97.084\n",
            "Train average loss: 0.0006340351928770542\n",
            "[ Evaluate epoch: 133 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.984\n",
            "Average loss: 0.0008885273524001241\n",
            "[ Evaluate epoch: 133 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.83\n",
            "Average loss: 0.006319953757524491\n",
            "Model Saved!\n",
            "[ Train epoch: 133 ]\n",
            "Train accuarcy: 97.272\n",
            "Train average loss: 0.0006014353299327195\n",
            "[ Evaluate epoch: 134 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.796\n",
            "Average loss: 0.0008979545632749796\n",
            "[ Evaluate epoch: 134 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.97\n",
            "Average loss: 0.006468858841061592\n",
            "Model Saved!\n",
            "[ Train epoch: 134 ]\n",
            "Train accuarcy: 97.24\n",
            "Train average loss: 0.0006184178015775978\n",
            "[ Evaluate epoch: 135 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.272\n",
            "Average loss: 0.0007868140877038241\n",
            "[ Evaluate epoch: 135 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.93\n",
            "Average loss: 0.006166702412068844\n",
            "Model Saved!\n",
            "[ Train epoch: 135 ]\n",
            "Train accuarcy: 97.546\n",
            "Train average loss: 0.0005636500720307231\n",
            "[ Evaluate epoch: 136 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.54\n",
            "Average loss: 0.0009752213789150119\n",
            "[ Evaluate epoch: 136 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.46\n",
            "Average loss: 0.0066330703228712085\n",
            "Model Saved!\n",
            "[ Train epoch: 136 ]\n",
            "Train accuarcy: 97.276\n",
            "Train average loss: 0.0005901158619299531\n",
            "[ Evaluate epoch: 137 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 95.778\n",
            "Average loss: 0.0009064855194464325\n",
            "[ Evaluate epoch: 137 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.51\n",
            "Average loss: 0.006411244386434555\n",
            "Model Saved!\n",
            "[ Train epoch: 137 ]\n",
            "Train accuarcy: 97.334\n",
            "Train average loss: 0.0005758808297663927\n",
            "[ Evaluate epoch: 138 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.88\n",
            "Average loss: 0.0006720615584403277\n",
            "[ Evaluate epoch: 138 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.7\n",
            "Average loss: 0.0058579169005155566\n",
            "Model Saved!\n",
            "[ Train epoch: 138 ]\n",
            "Train accuarcy: 97.368\n",
            "Train average loss: 0.0005887621577084065\n",
            "[ Evaluate epoch: 139 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.592\n",
            "Average loss: 0.0005318045434169472\n",
            "[ Evaluate epoch: 139 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.82\n",
            "Average loss: 0.005464630416035652\n",
            "Model Saved!\n",
            "[ Train epoch: 139 ]\n",
            "Train accuarcy: 97.434\n",
            "Train average loss: 0.000572571397125721\n",
            "[ Evaluate epoch: 140 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.868\n",
            "Average loss: 0.0006779492324218154\n",
            "[ Evaluate epoch: 140 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.54\n",
            "Average loss: 0.005931519223749637\n",
            "Model Saved!\n",
            "[ Train epoch: 140 ]\n",
            "Train accuarcy: 97.534\n",
            "Train average loss: 0.0005463645709119738\n",
            "[ Evaluate epoch: 141 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.682\n",
            "Average loss: 0.0007160531600564718\n",
            "[ Evaluate epoch: 141 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.36\n",
            "Average loss: 0.006077237538993359\n",
            "Model Saved!\n",
            "[ Train epoch: 141 ]\n",
            "Train accuarcy: 97.556\n",
            "Train average loss: 0.0005436080637387931\n",
            "[ Evaluate epoch: 142 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.45\n",
            "Average loss: 0.0007674145157262683\n",
            "[ Evaluate epoch: 142 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.26\n",
            "Average loss: 0.006411433866620064\n",
            "Model Saved!\n",
            "[ Train epoch: 142 ]\n",
            "Train accuarcy: 97.302\n",
            "Train average loss: 0.0005951995620504021\n",
            "[ Evaluate epoch: 143 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.86\n",
            "Average loss: 0.0004769992837496102\n",
            "[ Evaluate epoch: 143 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.31\n",
            "Average loss: 0.005553242564201355\n",
            "Model Saved!\n",
            "[ Train epoch: 143 ]\n",
            "Train accuarcy: 97.506\n",
            "Train average loss: 0.0005401278430223465\n",
            "[ Evaluate epoch: 144 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.692\n",
            "Average loss: 0.0007017492616921663\n",
            "[ Evaluate epoch: 144 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.46\n",
            "Average loss: 0.006110509335994721\n",
            "Model Saved!\n",
            "[ Train epoch: 144 ]\n",
            "Train accuarcy: 97.388\n",
            "Train average loss: 0.0005672413031570613\n",
            "[ Evaluate epoch: 145 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.328\n",
            "Average loss: 0.0005920590873807668\n",
            "[ Evaluate epoch: 145 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.04\n",
            "Average loss: 0.005859510371088982\n",
            "Model Saved!\n",
            "[ Train epoch: 145 ]\n",
            "Train accuarcy: 97.576\n",
            "Train average loss: 0.0005372934284061194\n",
            "[ Evaluate epoch: 146 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.184\n",
            "Average loss: 0.0006062452374212444\n",
            "[ Evaluate epoch: 146 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.64\n",
            "Average loss: 0.006093193589150905\n",
            "Model Saved!\n",
            "[ Train epoch: 146 ]\n",
            "Train accuarcy: 97.54\n",
            "Train average loss: 0.0005380984432809055\n",
            "[ Evaluate epoch: 147 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.27\n",
            "Average loss: 0.0006097527317330241\n",
            "[ Evaluate epoch: 147 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.64\n",
            "Average loss: 0.005985092201828956\n",
            "Model Saved!\n",
            "[ Train epoch: 147 ]\n",
            "Train accuarcy: 97.56\n",
            "Train average loss: 0.0005329239517264068\n",
            "[ Evaluate epoch: 148 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.01\n",
            "Average loss: 0.0008428437936864793\n",
            "[ Evaluate epoch: 148 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 85.67\n",
            "Average loss: 0.006732973700761795\n",
            "Model Saved!\n",
            "[ Train epoch: 148 ]\n",
            "Train accuarcy: 97.612\n",
            "Train average loss: 0.0005319232991337777\n",
            "[ Evaluate epoch: 149 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.33\n",
            "Average loss: 0.0005777783841639757\n",
            "[ Evaluate epoch: 149 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.08\n",
            "Average loss: 0.005828060574829578\n",
            "Model Saved!\n",
            "[ Train epoch: 149 ]\n",
            "Train accuarcy: 97.666\n",
            "Train average loss: 0.000520798803716898\n",
            "[ Evaluate epoch: 150 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.708\n",
            "Average loss: 0.0005080302098207176\n",
            "[ Evaluate epoch: 150 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.43\n",
            "Average loss: 0.005628104841709137\n",
            "Model Saved!\n",
            "[ Train epoch: 150 ]\n",
            "Train accuarcy: 97.834\n",
            "Train average loss: 0.0004984406484663486\n",
            "[ Evaluate epoch: 151 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.878\n",
            "Average loss: 0.0006753422440215945\n",
            "[ Evaluate epoch: 151 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.91\n",
            "Average loss: 0.006265967418253421\n",
            "Model Saved!\n",
            "[ Train epoch: 151 ]\n",
            "Train accuarcy: 97.676\n",
            "Train average loss: 0.0005132595046237112\n",
            "[ Evaluate epoch: 152 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.2\n",
            "Average loss: 0.0004068324351869524\n",
            "[ Evaluate epoch: 152 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.93\n",
            "Average loss: 0.00552627979516983\n",
            "Model Saved!\n",
            "[ Train epoch: 152 ]\n",
            "Train accuarcy: 97.794\n",
            "Train average loss: 0.000493049136493355\n",
            "[ Evaluate epoch: 153 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.0\n",
            "Average loss: 0.0006395122933015228\n",
            "[ Evaluate epoch: 153 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.63\n",
            "Average loss: 0.006187140637636185\n",
            "Model Saved!\n",
            "[ Train epoch: 153 ]\n",
            "Train accuarcy: 97.866\n",
            "Train average loss: 0.0004819266748987138\n",
            "[ Evaluate epoch: 154 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.578\n",
            "Average loss: 0.0005354345125332475\n",
            "[ Evaluate epoch: 154 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.18\n",
            "Average loss: 0.005641478979587555\n",
            "Model Saved!\n",
            "[ Train epoch: 154 ]\n",
            "Train accuarcy: 97.766\n",
            "Train average loss: 0.0004924424568749964\n",
            "[ Evaluate epoch: 155 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.608\n",
            "Average loss: 0.0007330487101525068\n",
            "[ Evaluate epoch: 155 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.49\n",
            "Average loss: 0.006391701906919479\n",
            "Model Saved!\n",
            "[ Train epoch: 155 ]\n",
            "Train accuarcy: 97.84\n",
            "Train average loss: 0.00047968236058950427\n",
            "[ Evaluate epoch: 156 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.596\n",
            "Average loss: 0.000541824483666569\n",
            "[ Evaluate epoch: 156 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.85\n",
            "Average loss: 0.006039164793491363\n",
            "Model Saved!\n",
            "[ Train epoch: 156 ]\n",
            "Train accuarcy: 97.672\n",
            "Train average loss: 0.0005124400920607149\n",
            "[ Evaluate epoch: 157 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.57\n",
            "Average loss: 0.0005191707173548639\n",
            "[ Evaluate epoch: 157 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.81\n",
            "Average loss: 0.005945622655749321\n",
            "Model Saved!\n",
            "[ Train epoch: 157 ]\n",
            "Train accuarcy: 97.798\n",
            "Train average loss: 0.00047625433737412095\n",
            "[ Evaluate epoch: 158 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.006\n",
            "Average loss: 0.00044859709640964867\n",
            "[ Evaluate epoch: 158 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.42\n",
            "Average loss: 0.005798868246376514\n",
            "Model Saved!\n",
            "[ Train epoch: 158 ]\n",
            "Train accuarcy: 98.072\n",
            "Train average loss: 0.0004445687563344836\n",
            "[ Evaluate epoch: 159 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.32\n",
            "Average loss: 0.0005800497241690755\n",
            "[ Evaluate epoch: 159 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.15\n",
            "Average loss: 0.006471619218587876\n",
            "Model Saved!\n",
            "[ Train epoch: 159 ]\n",
            "Train accuarcy: 98.066\n",
            "Train average loss: 0.0004333729583397508\n",
            "[ Evaluate epoch: 160 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.928\n",
            "Average loss: 0.00047753520909696816\n",
            "[ Evaluate epoch: 160 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.31\n",
            "Average loss: 0.005691353790462017\n",
            "Model Saved!\n",
            "[ Train epoch: 160 ]\n",
            "Train accuarcy: 97.724\n",
            "Train average loss: 0.0005015881214477122\n",
            "[ Evaluate epoch: 161 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.198\n",
            "Average loss: 0.0004160804528929293\n",
            "[ Evaluate epoch: 161 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.26\n",
            "Average loss: 0.005850030155479908\n",
            "Model Saved!\n",
            "[ Train epoch: 161 ]\n",
            "Train accuarcy: 97.922\n",
            "Train average loss: 0.00045307264922186735\n",
            "[ Evaluate epoch: 162 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.44\n",
            "Average loss: 0.0003560870461072773\n",
            "[ Evaluate epoch: 162 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.85\n",
            "Average loss: 0.0057315062254667285\n",
            "Model Saved!\n",
            "[ Train epoch: 162 ]\n",
            "Train accuarcy: 97.952\n",
            "Train average loss: 0.0004556555575877428\n",
            "[ Evaluate epoch: 163 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.648\n",
            "Average loss: 0.0005162503281049431\n",
            "[ Evaluate epoch: 163 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.29\n",
            "Average loss: 0.005936167825758457\n",
            "Model Saved!\n",
            "[ Train epoch: 163 ]\n",
            "Train accuarcy: 97.976\n",
            "Train average loss: 0.00046063786486163734\n",
            "[ Evaluate epoch: 164 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.154\n",
            "Average loss: 0.0006358825175277889\n",
            "[ Evaluate epoch: 164 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.0\n",
            "Average loss: 0.006562225317955017\n",
            "Model Saved!\n",
            "[ Train epoch: 164 ]\n",
            "Train accuarcy: 98.092\n",
            "Train average loss: 0.00043669218172319235\n",
            "[ Evaluate epoch: 165 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.274\n",
            "Average loss: 0.0006004943595826626\n",
            "[ Evaluate epoch: 165 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.68\n",
            "Average loss: 0.00628475701957941\n",
            "Model Saved!\n",
            "[ Train epoch: 165 ]\n",
            "Train accuarcy: 98.124\n",
            "Train average loss: 0.00042319148724898693\n",
            "[ Evaluate epoch: 166 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.01\n",
            "Average loss: 0.0004387019842118025\n",
            "[ Evaluate epoch: 166 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.31\n",
            "Average loss: 0.005755605155229568\n",
            "Model Saved!\n",
            "[ Train epoch: 166 ]\n",
            "Train accuarcy: 98.042\n",
            "Train average loss: 0.0004332824080251157\n",
            "[ Evaluate epoch: 167 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.196\n",
            "Average loss: 0.00041190638210624456\n",
            "[ Evaluate epoch: 167 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.59\n",
            "Average loss: 0.005819363829493523\n",
            "Model Saved!\n",
            "[ Train epoch: 167 ]\n",
            "Train accuarcy: 98.0\n",
            "Train average loss: 0.00044194770502857864\n",
            "[ Evaluate epoch: 168 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.162\n",
            "Average loss: 0.00041243635567836466\n",
            "[ Evaluate epoch: 168 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.44\n",
            "Average loss: 0.005939891695976258\n",
            "Model Saved!\n",
            "[ Train epoch: 168 ]\n",
            "Train accuarcy: 98.028\n",
            "Train average loss: 0.0004482599516957998\n",
            "[ Evaluate epoch: 169 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.398\n",
            "Average loss: 0.0005746898874640465\n",
            "[ Evaluate epoch: 169 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.75\n",
            "Average loss: 0.006666691909730435\n",
            "Model Saved!\n",
            "[ Train epoch: 169 ]\n",
            "Train accuarcy: 97.958\n",
            "Train average loss: 0.00044358643198385834\n",
            "[ Evaluate epoch: 170 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.89\n",
            "Average loss: 0.00046845802560448646\n",
            "[ Evaluate epoch: 170 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.92\n",
            "Average loss: 0.006415561065077782\n",
            "Model Saved!\n",
            "[ Train epoch: 170 ]\n",
            "Train accuarcy: 98.1\n",
            "Train average loss: 0.00041888088734820484\n",
            "[ Evaluate epoch: 171 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.838\n",
            "Average loss: 0.0004727420164272189\n",
            "[ Evaluate epoch: 171 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.3\n",
            "Average loss: 0.006055699795484543\n",
            "Model Saved!\n",
            "[ Train epoch: 171 ]\n",
            "Train accuarcy: 98.066\n",
            "Train average loss: 0.00043804697938263417\n",
            "[ Evaluate epoch: 172 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.338\n",
            "Average loss: 0.00037152255663648246\n",
            "[ Evaluate epoch: 172 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.78\n",
            "Average loss: 0.0057523394122719765\n",
            "Model Saved!\n",
            "[ Train epoch: 172 ]\n",
            "Train accuarcy: 98.252\n",
            "Train average loss: 0.00039000610956922173\n",
            "[ Evaluate epoch: 173 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.916\n",
            "Average loss: 0.0006779682497866452\n",
            "[ Evaluate epoch: 173 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.05\n",
            "Average loss: 0.0072108932018280026\n",
            "Model Saved!\n",
            "[ Train epoch: 173 ]\n",
            "Train accuarcy: 98.182\n",
            "Train average loss: 0.00040813595060259106\n",
            "[ Evaluate epoch: 174 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.44\n",
            "Average loss: 0.00035870483489707113\n",
            "[ Evaluate epoch: 174 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.66\n",
            "Average loss: 0.005884061455726623\n",
            "Model Saved!\n",
            "[ Train epoch: 174 ]\n",
            "Train accuarcy: 98.238\n",
            "Train average loss: 0.0003882586173154414\n",
            "[ Evaluate epoch: 175 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.26\n",
            "Average loss: 0.00039143483130261303\n",
            "[ Evaluate epoch: 175 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.28\n",
            "Average loss: 0.005898004360496998\n",
            "Model Saved!\n",
            "[ Train epoch: 175 ]\n",
            "Train accuarcy: 98.078\n",
            "Train average loss: 0.0004291195241361856\n",
            "[ Evaluate epoch: 176 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.906\n",
            "Average loss: 0.00046577920597046615\n",
            "[ Evaluate epoch: 176 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.31\n",
            "Average loss: 0.00637530668079853\n",
            "Model Saved!\n",
            "[ Train epoch: 176 ]\n",
            "Train accuarcy: 98.322\n",
            "Train average loss: 0.00037797272093594074\n",
            "[ Evaluate epoch: 177 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.136\n",
            "Average loss: 0.0006154919642210007\n",
            "[ Evaluate epoch: 177 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.59\n",
            "Average loss: 0.006469667226076126\n",
            "Model Saved!\n",
            "[ Train epoch: 177 ]\n",
            "Train accuarcy: 98.33\n",
            "Train average loss: 0.00037690838717855513\n",
            "[ Evaluate epoch: 178 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.614\n",
            "Average loss: 0.0003260784289613366\n",
            "[ Evaluate epoch: 178 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.63\n",
            "Average loss: 0.005666797874867916\n",
            "Model Saved!\n",
            "[ Train epoch: 178 ]\n",
            "Train accuarcy: 98.302\n",
            "Train average loss: 0.0003758429559133947\n",
            "[ Evaluate epoch: 179 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.71\n",
            "Average loss: 0.0005090822747815401\n",
            "[ Evaluate epoch: 179 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.04\n",
            "Average loss: 0.006555013839900494\n",
            "Model Saved!\n",
            "[ Train epoch: 179 ]\n",
            "Train accuarcy: 98.31\n",
            "Train average loss: 0.0003830062926374376\n",
            "[ Evaluate epoch: 180 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.296\n",
            "Average loss: 0.000591578925922513\n",
            "[ Evaluate epoch: 180 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.67\n",
            "Average loss: 0.00644140782058239\n",
            "Model Saved!\n",
            "[ Train epoch: 180 ]\n",
            "Train accuarcy: 98.234\n",
            "Train average loss: 0.00039020568776875733\n",
            "[ Evaluate epoch: 181 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.704\n",
            "Average loss: 0.0005121048786398023\n",
            "[ Evaluate epoch: 181 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.91\n",
            "Average loss: 0.006894730500876903\n",
            "Model Saved!\n",
            "[ Train epoch: 181 ]\n",
            "Train accuarcy: 98.356\n",
            "Train average loss: 0.00037058182451874016\n",
            "[ Evaluate epoch: 182 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.586\n",
            "Average loss: 0.00031359646663069727\n",
            "[ Evaluate epoch: 182 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.78\n",
            "Average loss: 0.006008354713022709\n",
            "Model Saved!\n",
            "[ Train epoch: 182 ]\n",
            "Train accuarcy: 98.384\n",
            "Train average loss: 0.0003541458744835109\n",
            "[ Evaluate epoch: 183 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.176\n",
            "Average loss: 0.00040781675485894084\n",
            "[ Evaluate epoch: 183 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.27\n",
            "Average loss: 0.006153648421168327\n",
            "Model Saved!\n",
            "[ Train epoch: 183 ]\n",
            "Train accuarcy: 98.308\n",
            "Train average loss: 0.0003860942975431681\n",
            "[ Evaluate epoch: 184 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.852\n",
            "Average loss: 0.0004784136290475726\n",
            "[ Evaluate epoch: 184 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.26\n",
            "Average loss: 0.006307877014577389\n",
            "Model Saved!\n",
            "[ Train epoch: 184 ]\n",
            "Train accuarcy: 98.274\n",
            "Train average loss: 0.0003845799693092704\n",
            "[ Evaluate epoch: 185 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.564\n",
            "Average loss: 0.00033388714861124755\n",
            "[ Evaluate epoch: 185 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.36\n",
            "Average loss: 0.0061711129084229465\n",
            "Model Saved!\n",
            "[ Train epoch: 185 ]\n",
            "Train accuarcy: 98.522\n",
            "Train average loss: 0.0003330060228332877\n",
            "[ Evaluate epoch: 186 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.944\n",
            "Average loss: 0.0004518892215471715\n",
            "[ Evaluate epoch: 186 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.25\n",
            "Average loss: 0.006064902596175671\n",
            "Model Saved!\n",
            "[ Train epoch: 186 ]\n",
            "Train accuarcy: 98.356\n",
            "Train average loss: 0.00037093552322126923\n",
            "[ Evaluate epoch: 187 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.428\n",
            "Average loss: 0.00035335532578639685\n",
            "[ Evaluate epoch: 187 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.87\n",
            "Average loss: 0.006065136897563934\n",
            "Model Saved!\n",
            "[ Train epoch: 187 ]\n",
            "Train accuarcy: 98.326\n",
            "Train average loss: 0.00038591235392726957\n",
            "[ Evaluate epoch: 188 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.984\n",
            "Average loss: 0.0004587698055244982\n",
            "[ Evaluate epoch: 188 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.36\n",
            "Average loss: 0.006465470142662525\n",
            "Model Saved!\n",
            "[ Train epoch: 188 ]\n",
            "Train accuarcy: 98.446\n",
            "Train average loss: 0.0003490367225650698\n",
            "[ Evaluate epoch: 189 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.48\n",
            "Average loss: 0.0005461241776868701\n",
            "[ Evaluate epoch: 189 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.55\n",
            "Average loss: 0.006767413955926895\n",
            "Model Saved!\n",
            "[ Train epoch: 189 ]\n",
            "Train accuarcy: 98.508\n",
            "Train average loss: 0.00033972649834118786\n",
            "[ Evaluate epoch: 190 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.16\n",
            "Average loss: 0.00041445073055103424\n",
            "[ Evaluate epoch: 190 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.92\n",
            "Average loss: 0.0064653852701187136\n",
            "Model Saved!\n",
            "[ Train epoch: 190 ]\n",
            "Train accuarcy: 98.472\n",
            "Train average loss: 0.00035107911890372635\n",
            "[ Evaluate epoch: 191 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.088\n",
            "Average loss: 0.0004105204921681434\n",
            "[ Evaluate epoch: 191 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.04\n",
            "Average loss: 0.0063040911197662355\n",
            "Model Saved!\n",
            "[ Train epoch: 191 ]\n",
            "Train accuarcy: 98.584\n",
            "Train average loss: 0.0003285239020828158\n",
            "[ Evaluate epoch: 192 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 96.632\n",
            "Average loss: 0.0007323157300241291\n",
            "[ Evaluate epoch: 192 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.36\n",
            "Average loss: 0.0070616171590983864\n",
            "Model Saved!\n",
            "[ Train epoch: 192 ]\n",
            "Train accuarcy: 98.422\n",
            "Train average loss: 0.00034997489230707287\n",
            "[ Evaluate epoch: 193 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.084\n",
            "Average loss: 0.0004240612926520407\n",
            "[ Evaluate epoch: 193 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.09\n",
            "Average loss: 0.006717846632003784\n",
            "Model Saved!\n",
            "[ Train epoch: 193 ]\n",
            "Train accuarcy: 98.558\n",
            "Train average loss: 0.0003358206014148891\n",
            "[ Evaluate epoch: 194 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.772\n",
            "Average loss: 0.000285071373777464\n",
            "[ Evaluate epoch: 194 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.67\n",
            "Average loss: 0.006085864207148552\n",
            "Model Saved!\n",
            "[ Train epoch: 194 ]\n",
            "Train accuarcy: 98.492\n",
            "Train average loss: 0.00034328153774142265\n",
            "[ Evaluate epoch: 195 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.888\n",
            "Average loss: 0.00045459046740084887\n",
            "[ Evaluate epoch: 195 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.1\n",
            "Average loss: 0.0063730937376618385\n",
            "Model Saved!\n",
            "[ Train epoch: 195 ]\n",
            "Train accuarcy: 98.704\n",
            "Train average loss: 0.00030058331951498987\n",
            "[ Evaluate epoch: 196 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.696\n",
            "Average loss: 0.0003055676651280373\n",
            "[ Evaluate epoch: 196 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.42\n",
            "Average loss: 0.006257934395968914\n",
            "Model Saved!\n",
            "[ Train epoch: 196 ]\n",
            "Train accuarcy: 98.518\n",
            "Train average loss: 0.00032928000303916635\n",
            "[ Evaluate epoch: 197 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.184\n",
            "Average loss: 0.000410848491974175\n",
            "[ Evaluate epoch: 197 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.59\n",
            "Average loss: 0.006490178798139096\n",
            "Model Saved!\n",
            "[ Train epoch: 197 ]\n",
            "Train accuarcy: 98.644\n",
            "Train average loss: 0.00031577832408249376\n",
            "[ Evaluate epoch: 198 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.094\n",
            "Average loss: 0.0004047103773243725\n",
            "[ Evaluate epoch: 198 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.29\n",
            "Average loss: 0.006602465744316578\n",
            "Model Saved!\n",
            "[ Train epoch: 198 ]\n",
            "Train accuarcy: 98.528\n",
            "Train average loss: 0.00033109492556657644\n",
            "[ Evaluate epoch: 199 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 98.832\n",
            "Average loss: 0.000275290095936507\n",
            "[ Evaluate epoch: 199 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 87.98\n",
            "Average loss: 0.0059389298439025875\n",
            "Model Saved!\n",
            "[ Train epoch: 199 ]\n",
            "Train accuarcy: 98.626\n",
            "Train average loss: 0.0003108837185613811\n",
            "[ Evaluate epoch: 200 ]\n",
            "Dataset: Train\n",
            "Accuarcy: 97.664\n",
            "Average loss: 0.0005187867158465087\n",
            "[ Evaluate epoch: 200 ]\n",
            "Dataset: Test\n",
            "Accuarcy: 86.65\n",
            "Average loss: 0.007070227617025375\n",
            "Model Saved!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jrVpbv1X0Aau",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "outputId": "ff0c657b-c0b4-4a86-9637-1e5a65f6eae2"
      },
      "source": [
        "# 정확도(accuracy) 커브 시각화\n",
        "plt.plot([i for i in range(epoch + 1)], [i[0] for i in train_result])\n",
        "plt.plot([i for i in range(epoch + 1)], [i[0] for i in test_result])\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.legend([\"train\", \"test\"])\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sdlhEdZW0C1x",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "outputId": "cdff8226-bfee-4e3e-9c61-5d9f61c0f1c3"
      },
      "source": [
        "# 손실(loss) 커브 시각화\n",
        "plt.plot([i for i in range(epoch + 1)], [i[1] for i in train_result])\n",
        "plt.plot([i for i in range(epoch + 1)], [i[1] for i in test_result])\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"Loss\")\n",
        "plt.legend([\"train\", \"test\"])\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZSoPuG4G0EWr",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 569
        },
        "outputId": "f8ede5fc-6434-41d5-c734-adf53ee8eb68"
      },
      "source": [
        "# 혼동 행렬(Confusion Matrix) 시각화 (학습 데이터셋)\n",
        "net = ResNet44()\n",
        "net = net.to(device)\n",
        "\n",
        "file_name = \"./checkpoint/ResNet44_CIFAR10.pt\"\n",
        "checkpoint = torch.load(file_name)\n",
        "net.load_state_dict(checkpoint['net'])\n",
        "\n",
        "confusion_matrix = get_confusion_matrix(net, 10, train_loader)\n",
        "print(\"[ 각 클래스당 데이터 개수 ]\")\n",
        "print(confusion_matrix.sum(1))\n",
        "\n",
        "print(\"[ 혼동 행렬(confusion matrix) 시각화 ]\")\n",
        "# 행(row)은 실제 레이블, 열(column)은 모델이 분류한 레이블\n",
        "res = pd.DataFrame(confusion_matrix.numpy(), index = [i for i in range(10)], columns = [i for i in range(10)])\n",
        "plt.figure(figsize = (10, 7))\n",
        "sn.heatmap(res, annot=True)\n",
        "plt.show()\n",
        "\n",
        "print(\"[ 각 클래스에 따른 정확도 ]\")\n",
        "# (각 클래스마다 정답 개수 / 각 클래스마다 데이터의 개수)\n",
        "print(confusion_matrix.diag() / confusion_matrix.sum(1))\n",
        "\n",
        "print(\"[ 전체 평균 정확도 ]\")\n",
        "print(confusion_matrix.diag().sum() / confusion_matrix.sum())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[ 각 클래스당 데이터 개수 ]\n",
            "tensor([5000., 5000., 5000., 5000., 5000., 5000., 5000., 5000., 5000., 5000.])\n",
            "[ 혼동 행렬(confusion matrix) 시각화 ]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x504 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "[ 각 클래스에 따른 정확도 ]\n",
            "tensor([0.9886, 0.9928, 0.9484, 0.9978, 0.9674, 0.9208, 0.9752, 0.9870, 0.9938,\n",
            "        0.9862])\n",
            "[ 전체 평균 정확도 ]\n",
            "tensor(0.9758)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "10Txa8ve0FxO",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 569
        },
        "outputId": "f1e790f6-4e8a-471e-fb7e-0e5f7e46b5c7"
      },
      "source": [
        "# 혼동 행렬(Confusion Matrix) 시각화 (테스트 데이터셋)\n",
        "net = ResNet44()\n",
        "net = net.to(device)\n",
        "\n",
        "file_name = \"./checkpoint/ResNet44_CIFAR10.pt\"\n",
        "checkpoint = torch.load(file_name)\n",
        "net.load_state_dict(checkpoint['net'])\n",
        "\n",
        "confusion_matrix = get_confusion_matrix(net, 10, test_loader)\n",
        "print(\"[ 각 클래스당 데이터 개수 ]\")\n",
        "print(confusion_matrix.sum(1))\n",
        "\n",
        "print(\"[ 혼동 행렬(confusion matrix) 시각화 ]\")\n",
        "# 행(row)은 실제 레이블, 열(column)은 모델이 분류한 레이블\n",
        "res = pd.DataFrame(confusion_matrix.numpy(), index = [i for i in range(10)], columns = [i for i in range(10)])\n",
        "plt.figure(figsize = (10, 7))\n",
        "sn.heatmap(res, annot=True)\n",
        "plt.show()\n",
        "\n",
        "print(\"[ 각 클래스에 따른 정확도 ]\")\n",
        "# (각 클래스마다 정답 개수 / 각 클래스마다 데이터의 개수)\n",
        "print(confusion_matrix.diag() / confusion_matrix.sum(1))\n",
        "\n",
        "print(\"[ 전체 평균 정확도 ]\")\n",
        "print(confusion_matrix.diag().sum() / confusion_matrix.sum())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[ 각 클래스당 데이터 개수 ]\n",
            "tensor([1000., 1000., 1000., 1000., 1000., 1000., 1000., 1000., 1000., 1000.])\n",
            "[ 혼동 행렬(confusion matrix) 시각화 ]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x504 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "[ 각 클래스에 따른 정확도 ]\n",
            "tensor([0.8940, 0.9470, 0.7370, 0.8900, 0.8510, 0.7110, 0.8670, 0.9080, 0.9470,\n",
            "        0.9130])\n",
            "[ 전체 평균 정확도 ]\n",
            "tensor(0.8665)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u0glICpm5lgV"
      },
      "source": [
        "from google.colab import files"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cU0vQP-c5m2K",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 17
        },
        "outputId": "1614f5e5-a7d6-49c2-88c8-9e017334b11d"
      },
      "source": [
        "files.download('./checkpoint/ResNet20_CIFAR10.pt')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_74cb0830-2f5e-4733-b1c9-27e2cf020f76\", \"ResNet20_CIFAR10.pt\", 1123441)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K9dVdkTw0K45",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 17
        },
        "outputId": "38fe3a71-bee9-4563-8ab1-0d7e27e6ab38"
      },
      "source": [
        "files.download('./checkpoint/ResNet44_CIFAR10.pt')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_d8ab353b-9067-4845-a8ab-e8cf5b8220a3\", \"ResNet44_CIFAR10.pt\", 2734801)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}