n-taku/DropoutSample.ipynb

## DropoutSample.ipynb
{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "DropoutSample.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "4ca9e7d2066647fb94cd239c10a28b82": {
          "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_645c6c6d069540659fc526c8eca021cd",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_03e7a8c15037407fa754e4ab9a843b40",
              "IPY_MODEL_6911709e987f431ab7540546177facae"
            ]
          }
        },
        "645c6c6d069540659fc526c8eca021cd": {
          "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
          }
        },
        "03e7a8c15037407fa754e4ab9a843b40": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "IntProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_566169d71db9434c898e3b478ddd7cb6",
            "_dom_classes": [],
            "description": "",
            "_model_name": "IntProgressModel",
            "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_52f4e79e40514e149516dbc2585433bf"
          }
        },
        "6911709e987f431ab7540546177facae": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_2e7dfe522c244d279fb6cb70f35aaa93",
            "_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, 55136303.11it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_9e022a3b468748c29da7b9d975d9fccd"
          }
        },
        "566169d71db9434c898e3b478ddd7cb6": {
          "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"
          }
        },
        "52f4e79e40514e149516dbc2585433bf": {
          "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
          }
        },
        "2e7dfe522c244d279fb6cb70f35aaa93": {
          "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"
          }
        },
        "9e022a3b468748c29da7b9d975d9fccd": {
          "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
          }
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "N0OQ3rVRINln",
        "colab_type": "code",
        "outputId": "53472c70-2ad9-4967-9a37-bcc3d2d70fea",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000,
          "referenced_widgets": [
            "4ca9e7d2066647fb94cd239c10a28b82",
            "645c6c6d069540659fc526c8eca021cd",
            "03e7a8c15037407fa754e4ab9a843b40",
            "6911709e987f431ab7540546177facae",
            "566169d71db9434c898e3b478ddd7cb6",
            "52f4e79e40514e149516dbc2585433bf",
            "2e7dfe522c244d279fb6cb70f35aaa93",
            "9e022a3b468748c29da7b9d975d9fccd"
          ]
        }
      },
      "source": [
        "import torch\n",
        "import torchvision\n",
        "import torch.nn as nn\n",
        "import torch.optim as optim\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import pickle\n",
        "from torchsummary import summary\n",
        "import torch.nn.functional as F\n",
        "\n",
        "BATCH_SIZE = 100\n",
        "EPOCH = 50\n",
        "PATH = \"Dataset\"\n",
        "\n",
        "transform = torchvision.transforms.Compose([torchvision.transforms.Resize((32,32)),torchvision.transforms.ToTensor(),torchvision.transforms.Normalize((0.5,), (0.5,))])\n",
        "\n",
        "trainset = torchvision.datasets.CIFAR10(root = PATH, train = True, download = True, transform = transform)\n",
        "trainloader = torch.utils.data.DataLoader(trainset, batch_size = BATCH_SIZE,\n",
        "                        shuffle = True, num_workers = 2)\n",
        "\n",
        "testset = torchvision.datasets.CIFAR10(root = PATH, train = False, download = True, transform = transform)\n",
        "testloader = torch.utils.data.DataLoader(testset, batch_size = BATCH_SIZE,\n",
        "                        shuffle = False, num_workers = 2)\n",
        "\n",
        "class Net(nn.Module):\n",
        "    def __init__(self):\n",
        "        super(Net, self).__init__()\n",
        "        self.pool = nn.MaxPool2d(2, 2)\n",
        "        self.conv1 = nn.Conv2d(3, 16, 5)\n",
        "        self.conv2 = nn.Conv2d(16, 32, 5)\n",
        "        self.conv3 = nn.Conv2d(32, 32, 5)\n",
        "        self.fc1 = nn.Linear(32 * 6 * 6, 256)\n",
        "        self.fc2 = nn.Linear(256, 10)\n",
        "        self.dropout1 = torch.nn.Dropout2d(p=0.2)\n",
        "        self.dropout2 = torch.nn.Dropout2d(p=0.3)\n",
        "        self.dropout3 = torch.nn.Dropout(p=0.3)\n",
        "    def forward(self, x):\n",
        "        x = self.dropout1(x)\n",
        "        x = self.pool(F.relu(self.conv1(x)))\n",
        "        x = self.dropout2(x)\n",
        "        x = F.relu(self.conv2(x))\n",
        "        x = self.dropout2(x)\n",
        "        x = F.relu(self.conv3(x))\n",
        "        x = self.dropout2(x)\n",
        "        x = torch.flatten(x, 1)\n",
        "        x = F.relu(self.fc1(x))\n",
        "        x = self.dropout3(x)\n",
        "        x = self.fc2(x)\n",
        "        return x\n",
        "\n",
        "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
        "net = Net()\n",
        "net = net.to(device)\n",
        "criterion = nn.CrossEntropyLoss()\n",
        "optimizer = optim.SGD(net.parameters(), lr=0.01, momentum=0.9)\n",
        "\n",
        "train_loss_value=[]      #trainingのlossを保持するlist\n",
        "train_acc_value=[]       #trainingのaccuracyを保持するlist\n",
        "test_loss_value=[]       #testのlossを保持するlist\n",
        "test_acc_value=[]        #testのaccuracyを保持するlist \n",
        "\n",
        "summary(net, (3, 32, 32))\n",
        "\n",
        "for epoch in range(EPOCH):\n",
        "    net.train()\n",
        "    print('epoch', epoch+1)    #epoch数の出力\n",
        "    for (inputs, labels) in trainloader:\n",
        "        inputs, labels = inputs.to(device), labels.to(device)\n",
        "        optimizer.zero_grad()\n",
        "        outputs = net(inputs)\n",
        "        loss = criterion(outputs, labels)\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "\n",
        "    sum_loss = 0.0   #lossの合計\n",
        "    sum_correct = 0 #正解率の合計\n",
        "    sum_total = 0     #dataの数の合計\n",
        "\n",
        "    #train dataを使ってテストをする(パラメータ更新がないようになっている)\n",
        "    net.eval()\n",
        "    for (inputs, labels) in trainloader:\n",
        "        inputs, labels = inputs.to(device), labels.to(device)\n",
        "        optimizer.zero_grad()\n",
        "        outputs = net(inputs)\n",
        "        loss = criterion(outputs, labels)\n",
        "        #lossを足していく\n",
        "        sum_loss += loss.item()\n",
        "        #出力の最大値の添字(予想位置)を取得\n",
        "        _, predicted = outputs.max(1)\n",
        "        #labelの数を足していくことでデータの総和を取る                      \n",
        "        sum_total += labels.size(0)\n",
        "        #予想位置と実際の正解を比べ,正解している数だけ足す\n",
        "        sum_correct += (predicted == labels).sum().item()\n",
        "    \n",
        "    #lossとaccuracy出力\n",
        "    print(\"train mean loss={}, accuracy={}\"\n",
        "    .format(sum_loss*BATCH_SIZE/len(trainloader.dataset), float(sum_correct/sum_total)))\n",
        "    #traindataのlossをグラフ描画のためにlistに保持\n",
        "    train_loss_value.append(sum_loss*BATCH_SIZE/len(trainloader.dataset))\n",
        "    #traindataのaccuracyをグラフ描画のためにlistに保持\n",
        "    train_acc_value.append(float(sum_correct/sum_total))\n",
        "\n",
        "    sum_loss = 0.0\n",
        "    sum_correct = 0\n",
        "    sum_total = 0\n",
        "\n",
        "    #test dataを使ってテストをする\n",
        "    for (inputs, labels) in testloader:\n",
        "        inputs, labels = inputs.to(device), labels.to(device)\n",
        "        optimizer.zero_grad()\n",
        "        outputs = net(inputs)\n",
        "        loss = criterion(outputs, labels)\n",
        "        sum_loss += loss.item()\n",
        "        _, predicted = outputs.max(1)\n",
        "        sum_total += labels.size(0)\n",
        "        sum_correct += (predicted == labels).sum().item()\n",
        "    print(\"test  mean loss={}, accuracy={}\"\n",
        "    .format(sum_loss*BATCH_SIZE/len(testloader.dataset), float(sum_correct/sum_total)))\n",
        "    test_loss_value.append(sum_loss*BATCH_SIZE/len(testloader.dataset))\n",
        "    test_acc_value.append(float(sum_correct/sum_total))\n",
        "\n",
        "#グラフ\n",
        "fig, (axL, axR) = plt.subplots(ncols=2, figsize=(12,6))\n",
        "\n",
        "#損失グラフ描画\n",
        "axL.plot(range(EPOCH), train_loss_value)\n",
        "axL.plot(range(EPOCH), test_loss_value, c='#00ff00')\n",
        "axL.set_xlabel('EPOCH')\n",
        "axL.set_ylabel('LOSS')\n",
        "axL.legend(['train loss', 'test loss'])\n",
        "axL.set_title('loss')\n",
        "\n",
        "#正答率グラフ描画\n",
        "axR.plot(range(EPOCH), train_acc_value)\n",
        "axR.plot(range(EPOCH), test_acc_value, c='#00ff00')\n",
        "axR.set_xlabel('EPOCH')\n",
        "axR.set_ylabel('ACCURACY')\n",
        "axR.legend(['train acc', 'test acc'])\n",
        "axR.set_title('accuracy')\n",
        "\n",
        "fig.savefig(\"loss_accuracy_image.png\")\n",
        "fig.show()"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to Dataset/cifar-10-python.tar.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "4ca9e7d2066647fb94cd239c10a28b82",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(IntProgress(value=1, bar_style='info', max=1), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting Dataset/cifar-10-python.tar.gz to Dataset\n",
            "Files already downloaded and verified\n",
            "----------------------------------------------------------------\n",
            "        Layer (type)               Output Shape         Param #\n",
            "================================================================\n",
            "         Dropout2d-1            [-1, 3, 32, 32]               0\n",
            "            Conv2d-2           [-1, 16, 28, 28]           1,216\n",
            "         MaxPool2d-3           [-1, 16, 14, 14]               0\n",
            "         Dropout2d-4           [-1, 16, 14, 14]               0\n",
            "            Conv2d-5           [-1, 32, 10, 10]          12,832\n",
            "         Dropout2d-6           [-1, 32, 10, 10]               0\n",
            "            Conv2d-7             [-1, 32, 6, 6]          25,632\n",
            "         Dropout2d-8             [-1, 32, 6, 6]               0\n",
            "            Linear-9                  [-1, 256]         295,168\n",
            "          Dropout-10                  [-1, 256]               0\n",
            "           Linear-11                   [-1, 10]           2,570\n",
            "================================================================\n",
            "Total params: 337,418\n",
            "Trainable params: 337,418\n",
            "Non-trainable params: 0\n",
            "----------------------------------------------------------------\n",
            "Input size (MB): 0.01\n",
            "Forward/backward pass size (MB): 0.24\n",
            "Params size (MB): 1.29\n",
            "Estimated Total Size (MB): 1.54\n",
            "----------------------------------------------------------------\n",
            "epoch 1\n",
            "train mean loss=1.8159868097305298, accuracy=0.34368\n",
            "test  mean loss=1.8120258855819702, accuracy=0.3463\n",
            "epoch 2\n",
            "train mean loss=1.6207420496940612, accuracy=0.42432\n",
            "test  mean loss=1.6235691463947297, accuracy=0.422\n",
            "epoch 3\n",
            "train mean loss=1.5814208030700683, accuracy=0.43602\n",
            "test  mean loss=1.5861244201660156, accuracy=0.4337\n",
            "epoch 4\n",
            "train mean loss=1.5673566706180573, accuracy=0.44768\n",
            "test  mean loss=1.5742740952968597, accuracy=0.443\n",
            "epoch 5\n",
            "train mean loss=1.4107752871513366, accuracy=0.51014\n",
            "test  mean loss=1.4336432147026061, accuracy=0.4979\n",
            "epoch 6\n",
            "train mean loss=1.425104397058487, accuracy=0.4981\n",
            "test  mean loss=1.4450274467468263, accuracy=0.4824\n",
            "epoch 7\n",
            "train mean loss=1.3653611505031586, accuracy=0.52516\n",
            "test  mean loss=1.398059984445572, accuracy=0.5021\n",
            "epoch 8\n",
            "train mean loss=1.3159274599552155, accuracy=0.54106\n",
            "test  mean loss=1.349722796678543, accuracy=0.5197\n",
            "epoch 9\n",
            "train mean loss=1.3272819707393646, accuracy=0.53824\n",
            "test  mean loss=1.3574199771881104, accuracy=0.5241\n",
            "epoch 10\n",
            "train mean loss=1.26170827794075, accuracy=0.55864\n",
            "test  mean loss=1.3057181930541992, accuracy=0.5409\n",
            "epoch 11\n",
            "train mean loss=1.2342692089080811, accuracy=0.577\n",
            "test  mean loss=1.278234715461731, accuracy=0.5582\n",
            "epoch 12\n",
            "train mean loss=1.2498343348503114, accuracy=0.57654\n",
            "test  mean loss=1.2962885987758637, accuracy=0.5574\n",
            "epoch 13\n",
            "train mean loss=1.23589399933815, accuracy=0.57406\n",
            "test  mean loss=1.2808181548118591, accuracy=0.553\n",
            "epoch 14\n",
            "train mean loss=1.221902673959732, accuracy=0.56866\n",
            "test  mean loss=1.2786028182506561, accuracy=0.5438\n",
            "epoch 15\n",
            "train mean loss=1.1803281693458556, accuracy=0.59204\n",
            "test  mean loss=1.2364378869533539, accuracy=0.5631\n",
            "epoch 16\n",
            "train mean loss=1.1954725732803344, accuracy=0.58648\n",
            "test  mean loss=1.2565995454788208, accuracy=0.558\n",
            "epoch 17\n",
            "train mean loss=1.1464856959581375, accuracy=0.6132\n",
            "test  mean loss=1.2132466542720795, accuracy=0.5842\n",
            "epoch 18\n",
            "train mean loss=1.176098375082016, accuracy=0.60276\n",
            "test  mean loss=1.2393678414821625, accuracy=0.5743\n",
            "epoch 19\n",
            "train mean loss=1.1301259944438935, accuracy=0.61618\n",
            "test  mean loss=1.2033439147472382, accuracy=0.5818\n",
            "epoch 20\n",
            "train mean loss=1.1357567695379258, accuracy=0.61362\n",
            "test  mean loss=1.214014275074005, accuracy=0.5727\n",
            "epoch 21\n",
            "train mean loss=1.1242799303531648, accuracy=0.63134\n",
            "test  mean loss=1.20374076128006, accuracy=0.5944\n",
            "epoch 22\n",
            "train mean loss=1.0910872992277145, accuracy=0.63728\n",
            "test  mean loss=1.1777085411548613, accuracy=0.598\n",
            "epoch 23\n",
            "train mean loss=1.0990907835960388, accuracy=0.62864\n",
            "test  mean loss=1.1902319759130477, accuracy=0.5849\n",
            "epoch 24\n",
            "train mean loss=1.09382466173172, accuracy=0.62944\n",
            "test  mean loss=1.1859383261203766, accuracy=0.5909\n",
            "epoch 25\n",
            "train mean loss=1.1111096861362457, accuracy=0.62746\n",
            "test  mean loss=1.2044089615345002, accuracy=0.5837\n",
            "epoch 26\n",
            "train mean loss=1.060552763223648, accuracy=0.64228\n",
            "test  mean loss=1.1628005802631378, accuracy=0.5966\n",
            "epoch 27\n",
            "train mean loss=1.1078788441419603, accuracy=0.62148\n",
            "test  mean loss=1.205456895828247, accuracy=0.5769\n",
            "epoch 28\n",
            "train mean loss=1.0619354034662247, accuracy=0.64332\n",
            "test  mean loss=1.1598036140203476, accuracy=0.6041\n",
            "epoch 29\n",
            "train mean loss=1.074964772105217, accuracy=0.64412\n",
            "test  mean loss=1.1790764093399049, accuracy=0.5966\n",
            "epoch 30\n",
            "train mean loss=1.022259269475937, accuracy=0.65998\n",
            "test  mean loss=1.136533917784691, accuracy=0.6116\n",
            "epoch 31\n",
            "train mean loss=1.0300654480457305, accuracy=0.65484\n",
            "test  mean loss=1.1454177016019822, accuracy=0.6076\n",
            "epoch 32\n",
            "train mean loss=1.0396746463775635, accuracy=0.65782\n",
            "test  mean loss=1.155140870809555, accuracy=0.612\n",
            "epoch 33\n",
            "train mean loss=1.009270124554634, accuracy=0.66294\n",
            "test  mean loss=1.1321960872411727, accuracy=0.6083\n",
            "epoch 34\n",
            "train mean loss=1.0156784734725952, accuracy=0.66126\n",
            "test  mean loss=1.1422304916381836, accuracy=0.6045\n",
            "epoch 35\n",
            "train mean loss=1.0245839591026307, accuracy=0.66564\n",
            "test  mean loss=1.1361274689435958, accuracy=0.6156\n",
            "epoch 36\n",
            "train mean loss=1.0340029946565628, accuracy=0.64826\n",
            "test  mean loss=1.1540372443199158, accuracy=0.5921\n",
            "epoch 37\n",
            "train mean loss=0.9913648378849029, accuracy=0.67228\n",
            "test  mean loss=1.12300263941288, accuracy=0.617\n",
            "epoch 38\n",
            "train mean loss=1.0042559345960618, accuracy=0.66048\n",
            "test  mean loss=1.1351888394355774, accuracy=0.606\n",
            "epoch 39\n",
            "train mean loss=0.984825072646141, accuracy=0.67082\n",
            "test  mean loss=1.1175998830795288, accuracy=0.616\n",
            "epoch 40\n",
            "train mean loss=0.9869006873369217, accuracy=0.67178\n",
            "test  mean loss=1.13115698158741, accuracy=0.6082\n",
            "epoch 41\n",
            "train mean loss=0.9745359154939651, accuracy=0.67576\n",
            "test  mean loss=1.120110120177269, accuracy=0.6123\n",
            "epoch 42\n",
            "train mean loss=1.0099540387392043, accuracy=0.66648\n",
            "test  mean loss=1.1493632823228837, accuracy=0.6036\n",
            "epoch 43\n",
            "train mean loss=1.0214283186197282, accuracy=0.65448\n",
            "test  mean loss=1.160828878879547, accuracy=0.5974\n",
            "epoch 44\n",
            "train mean loss=0.983869577050209, accuracy=0.66994\n",
            "test  mean loss=1.1340904533863068, accuracy=0.6042\n",
            "epoch 45\n",
            "train mean loss=0.9820229386091233, accuracy=0.67648\n",
            "test  mean loss=1.1275955641269684, accuracy=0.6114\n",
            "epoch 46\n",
            "train mean loss=0.9592411957979202, accuracy=0.6862\n",
            "test  mean loss=1.1144607275724412, accuracy=0.6173\n",
            "epoch 47\n",
            "train mean loss=0.953042807340622, accuracy=0.6901\n",
            "test  mean loss=1.1144620078802108, accuracy=0.6213\n",
            "epoch 48\n",
            "train mean loss=0.9597269052267074, accuracy=0.68054\n",
            "test  mean loss=1.1189067500829697, accuracy=0.6127\n",
            "epoch 49\n",
            "train mean loss=0.9670628876686096, accuracy=0.68064\n",
            "test  mean loss=1.1297900193929673, accuracy=0.609\n",
            "epoch 50\n",
            "train mean loss=0.929839742898941, accuracy=0.69798\n",
            "test  mean loss=1.0998518043756484, accuracy=0.6246\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x432 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}