cnn_trainer.ipynb

{
 "nbformat": 4,
 "nbformat_minor": 0,
 "metadata": {
  "kernelspec": {
   "name": "python3",
   "display_name": "Python 3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  },
  "pycharm": {
   "stem_cell": {
    "cell_type": "raw",
    "source": [],
    "metadata": {
     "collapsed": false
    }
   }
  },
  "colab": {
   "name": "cnn_trainer.ipynb",
   "provenance": [],
   "collapsed_sections": []
  },
  "accelerator": "GPU",
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "7b5568bb46e145a2b42f43c655c40b2b": {
     "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_5ffb568834b94a7c92d97078f533a7a1",
      "_model_module": "@jupyter-widgets/controls",
      "children": [
       "IPY_MODEL_3fd5804812f943d3b21fb4593b1175da",
       "IPY_MODEL_cef010b5c083412abc55a06d51a14af4"
      ]
     }
    },
    "5ffb568834b94a7c92d97078f533a7a1": {
     "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
     }
    },
    "3fd5804812f943d3b21fb4593b1175da": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "IntProgressModel",
     "state": {
      "_view_name": "ProgressView",
      "style": "IPY_MODEL_bab504fcf3b84bf9908dfe747be07a7a",
      "_dom_classes": [],
      "description": "",
      "_model_name": "IntProgressModel",
      "bar_style": "success",
      "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_d87a83b4aca94dad835be8a9e5fcfc47"
     }
    },
    "cef010b5c083412abc55a06d51a14af4": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_view_name": "HTMLView",
      "style": "IPY_MODEL_ef052bef0abd419bae621f241c32dcae",
      "_dom_classes": [],
      "description": "",
      "_model_name": "HTMLModel",
      "placeholder": "​",
      "_view_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "value": "561758208it [00:07, 80082522.68it/s]",
      "_view_count": null,
      "_view_module_version": "1.5.0",
      "description_tooltip": null,
      "_model_module": "@jupyter-widgets/controls",
      "layout": "IPY_MODEL_80d966c01581431cbc965ce0a7b0e74a"
     }
    },
    "bab504fcf3b84bf9908dfe747be07a7a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_view_name": "StyleView",
      "_model_name": "ProgressStyleModel",
      "description_width": "",
      "_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"
     }
    },
    "d87a83b4aca94dad835be8a9e5fcfc47": {
     "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
     }
    },
    "ef052bef0abd419bae621f241c32dcae": {
     "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"
     }
    },
    "80d966c01581431cbc965ce0a7b0e74a": {
     "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": {
    "pycharm": {
     "is_executing": false
    },
    "id": "4nvYJvEdS-_K",
    "colab_type": "code",
    "colab": {}
   },
   "source": [
    "import numpy as np\n",
    "import torch\n",
    "\n",
    "from torch import nn, optim\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision import datasets, transforms\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "epochs = 5\n",
    "learning_rate= 0.1\n",
    "momentum=0.5"
   ],
   "execution_count": 0,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    },
    "id": "19l3XMfhS-_P",
    "colab_type": "code",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 236,
     "referenced_widgets": [
      "7b5568bb46e145a2b42f43c655c40b2b",
      "5ffb568834b94a7c92d97078f533a7a1",
      "3fd5804812f943d3b21fb4593b1175da",
      "cef010b5c083412abc55a06d51a14af4",
      "bab504fcf3b84bf9908dfe747be07a7a",
      "d87a83b4aca94dad835be8a9e5fcfc47",
      "ef052bef0abd419bae621f241c32dcae",
      "80d966c01581431cbc965ce0a7b0e74a"
     ]
    },
    "outputId": "62fd6e01-b293-40ee-91e4-68d420cbe850"
   },
   "source": [
    "# Values for Normalize() are the global mean and\n",
    "# standard deviation of the MNIST dataset\n",
    "transform = transforms.Compose([lambda img: transforms.functional.rotate(img, -90, fill=(0,)),\n",
    "                                lambda img: transforms.functional.hflip(img),\n",
    "                                transforms.ToTensor(),\n",
    "                                transforms.Normalize((0.5,), (0.5,))])\n",
    "\n",
    "train_data = datasets.EMNIST('dataset/', split='byclass', train=True,\n",
    "                             download=True, transform=transform)\n",
    "\n",
    "test_data = datasets.EMNIST('dataset/', split='byclass', train=False,\n",
    "                            download=True, transform=transform)"
   ],
   "execution_count": 2,
   "outputs": [
    {
     "output_type": "stream",
     "text": [
      "Downloading and extracting zip archive\n",
      "Downloading http://www.itl.nist.gov/iaui/vip/cs_links/EMNIST/gzip.zip to dataset/EMNIST/raw/emnist.zip\n"
     ],
     "name": "stdout"
    },
    {
     "output_type": "display_data",
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7b5568bb46e145a2b42f43c655c40b2b",
       "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": [
      "\n",
      "Extracting dataset/EMNIST/raw/emnist.zip to dataset/EMNIST/raw\n",
      "Processing byclass\n",
      "Processing bymerge\n",
      "Processing balanced\n",
      "Processing letters\n",
      "Processing digits\n",
      "Processing mnist\n",
      "Done!\n"
     ],
     "name": "stdout"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    },
    "id": "f7oa2xoqS-_T",
    "colab_type": "code",
    "colab": {}
   },
   "source": [
    "train_loader = DataLoader(train_data,\n",
    "                          batch_size=64,\n",
    "                          shuffle=True)\n",
    "\n",
    "test_loader = DataLoader(test_data,\n",
    "                         batch_size=1000,\n",
    "                         shuffle=True)"
   ],
   "execution_count": 0,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    },
    "id": "RDk3jVw9S-_V",
    "colab_type": "code",
    "colab": {}
   },
   "source": [
    "class Net(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Net, self).__init__()\n",
    "        self.layer1 = nn.Sequential(\n",
    "            nn.Conv2d(1, 128, kernel_size=5, stride=1, padding=2),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(kernel_size=2))\n",
    "        self.layer2 = nn.Sequential(\n",
    "            nn.Conv2d(128, 256, kernel_size=5, stride=1, padding=2),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(kernel_size=2))\n",
    "        self.dropout = nn.Dropout()  # Dropout to prevent overfitting\n",
    "        self.fc1 = nn.Linear(7 * 7 * 256, 512)\n",
    "        self.fc2 = nn.Linear(512, 62)\n",
    "        self.softmax = nn.LogSoftmax(dim=1)\n",
    "\n",
    "    def forward(self, x):\n",
    "        out = self.layer1(x)\n",
    "        out = self.layer2(out)\n",
    "        out = out.reshape(out.size(0), -1)\n",
    "        out = self.dropout(out)\n",
    "        out = self.fc1(out)\n",
    "        out = self.fc2(out)\n",
    "        out = self.softmax(out)\n",
    "\n",
    "        return out"
   ],
   "execution_count": 0,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    },
    "id": "a3JJiqA0S-_Y",
    "colab_type": "code",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 306
    },
    "outputId": "a479eeac-7ebf-42b9-f11c-97b4f08c964b"
   },
   "source": [
    "device = ('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "print(device)\n",
    "cnn = Net()\n",
    "cnn.to(device)"
   ],
   "execution_count": 5,
   "outputs": [
    {
     "output_type": "stream",
     "text": [
      "cuda\n"
     ],
     "name": "stdout"
    },
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "Net(\n",
       "  (layer1): Sequential(\n",
       "    (0): Conv2d(1, 128, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))\n",
       "    (1): ReLU()\n",
       "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "  )\n",
       "  (layer2): Sequential(\n",
       "    (0): Conv2d(128, 256, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))\n",
       "    (1): ReLU()\n",
       "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "  )\n",
       "  (dropout): Dropout(p=0.5, inplace=False)\n",
       "  (fc1): Linear(in_features=12544, out_features=512, bias=True)\n",
       "  (fc2): Linear(in_features=512, out_features=62, bias=True)\n",
       "  (softmax): LogSoftmax()\n",
       ")"
      ]
     },
     "metadata": {
      "tags": []
     },
     "execution_count": 5
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    },
    "id": "FSCVajh7S-_c",
    "colab_type": "code",
    "colab": {}
   },
   "source": [
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = optim.SGD(cnn.parameters(), lr=learning_rate, momentum=momentum)\n",
    "def train(epoch_n):\n",
    "    running_loss = 0\n",
    "    for idx, (images, labels) in enumerate(train_loader):\n",
    "        images = images.to(device)\n",
    "        labels = labels.to(device)\n",
    "        \n",
    "        optimizer.zero_grad()\n",
    "        \n",
    "        output = cnn(images)\n",
    "        loss = criterion(output, labels)\n",
    "        \n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        \n",
    "        running_loss += loss.item()\n",
    "        \n",
    "        if idx % 100 == 0:\n",
    "            print('Epoch {} -> [{}/{} ({:.0f}%)] Loss: {:.6}'.format(\n",
    "                epoch_n, idx * len(images), len(train_loader.dataset),\n",
    "                100. * idx / len(train_loader), running_loss/len(train_loader))\n",
    "            )\n",
    "        "
   ],
   "execution_count": 0,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    },
    "id": "JFoF6DYsS-_f",
    "colab_type": "code",
    "colab": {}
   },
   "source": [
    "def test():\n",
    "    correct, total, test_loss = 0, 0, 0\n",
    "    with torch.no_grad():\n",
    "        for images, labels in test_loader:\n",
    "            images = images.to(device)\n",
    "            labels = labels.to(device)\n",
    "\n",
    "            output = cnn(images)\n",
    "            \n",
    "            predicted_label = output.data.max(1, keepdim=True)[1]\n",
    "            correct += predicted_label.eq(labels.\n",
    "                                          data.view_as(predicted_label)).sum()\n",
    "        \n",
    "            loss = criterion(output, labels)\n",
    "            test_loss += loss.item()\n",
    "        \n",
    "        test_loss /= len(test_loader.dataset)"
   ],
   "execution_count": 0,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    },
    "id": "bTVMtm5nS-_i",
    "colab_type": "code",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "outputId": "678288a5-4154-44d6-aba9-fe24f20335f3"
   },
   "source": [
    "test()\n",
    "for e in range(1, epochs + 1):\n",
    "    train(e)\n",
    "    test()\n"
   ],
   "execution_count": 8,
   "outputs": [
    {
     "output_type": "stream",
     "text": [
      "Accuracy = 0.58%\n",
      "Epoch 1 -> [0/697932 (0%)] Loss: 0.00037627\n",
      "Epoch 1 -> [6400/697932 (1%)] Loss: 0.0182136\n",
      "Epoch 1 -> [12800/697932 (2%)] Loss: 0.0271421\n",
      "Epoch 1 -> [19200/697932 (3%)] Loss: 0.0346773\n",
      "Epoch 1 -> [25600/697932 (4%)] Loss: 0.0418186\n",
      "Epoch 1 -> [32000/697932 (5%)] Loss: 0.0484314\n",
      "Epoch 1 -> [38400/697932 (6%)] Loss: 0.0546063\n",
      "Epoch 1 -> [44800/697932 (6%)] Loss: 0.060403\n",
      "Epoch 1 -> [51200/697932 (7%)] Loss: 0.0663319\n",
      "Epoch 1 -> [57600/697932 (8%)] Loss: 0.0720033\n",
      "Epoch 1 -> [64000/697932 (9%)] Loss: 0.0774526\n",
      "Epoch 1 -> [70400/697932 (10%)] Loss: 0.0829084\n",
      "Epoch 1 -> [76800/697932 (11%)] Loss: 0.0882281\n",
      "Epoch 1 -> [83200/697932 (12%)] Loss: 0.0934637\n",
      "Epoch 1 -> [89600/697932 (13%)] Loss: 0.0987242\n",
      "Epoch 1 -> [96000/697932 (14%)] Loss: 0.103765\n",
      "Epoch 1 -> [102400/697932 (15%)] Loss: 0.108594\n",
      "Epoch 1 -> [108800/697932 (16%)] Loss: 0.113599\n",
      "Epoch 1 -> [115200/697932 (17%)] Loss: 0.118512\n",
      "Epoch 1 -> [121600/697932 (17%)] Loss: 0.123317\n",
      "Epoch 1 -> [128000/697932 (18%)] Loss: 0.128095\n",
      "Epoch 1 -> [134400/697932 (19%)] Loss: 0.132798\n",
      "Epoch 1 -> [140800/697932 (20%)] Loss: 0.137393\n",
      "Epoch 1 -> [147200/697932 (21%)] Loss: 0.141969\n",
      "Epoch 1 -> [153600/697932 (22%)] Loss: 0.14673\n",
      "Epoch 1 -> [160000/697932 (23%)] Loss: 0.151386\n",
      "Epoch 1 -> [166400/697932 (24%)] Loss: 0.156004\n",
      "Epoch 1 -> [172800/697932 (25%)] Loss: 0.160687\n",
      "Epoch 1 -> [179200/697932 (26%)] Loss: 0.165273\n",
      "Epoch 1 -> [185600/697932 (27%)] Loss: 0.169779\n",
      "Epoch 1 -> [192000/697932 (28%)] Loss: 0.174204\n",
      "Epoch 1 -> [198400/697932 (28%)] Loss: 0.178656\n",
      "Epoch 1 -> [204800/697932 (29%)] Loss: 0.183076\n",
      "Epoch 1 -> [211200/697932 (30%)] Loss: 0.187482\n",
      "Epoch 1 -> [217600/697932 (31%)] Loss: 0.191931\n",
      "Epoch 1 -> [224000/697932 (32%)] Loss: 0.19647\n",
      "Epoch 1 -> [230400/697932 (33%)] Loss: 0.200949\n",
      "Epoch 1 -> [236800/697932 (34%)] Loss: 0.205192\n",
      "Epoch 1 -> [243200/697932 (35%)] Loss: 0.209598\n",
      "Epoch 1 -> [249600/697932 (36%)] Loss: 0.213874\n",
      "Epoch 1 -> [256000/697932 (37%)] Loss: 0.218227\n",
      "Epoch 1 -> [262400/697932 (38%)] Loss: 0.222495\n",
      "Epoch 1 -> [268800/697932 (39%)] Loss: 0.226897\n",
      "Epoch 1 -> [275200/697932 (39%)] Loss: 0.231339\n",
      "Epoch 1 -> [281600/697932 (40%)] Loss: 0.235545\n",
      "Epoch 1 -> [288000/697932 (41%)] Loss: 0.239851\n",
      "Epoch 1 -> [294400/697932 (42%)] Loss: 0.244139\n",
      "Epoch 1 -> [300800/697932 (43%)] Loss: 0.248425\n",
      "Epoch 1 -> [307200/697932 (44%)] Loss: 0.252793\n",
      "Epoch 1 -> [313600/697932 (45%)] Loss: 0.257151\n",
      "Epoch 1 -> [320000/697932 (46%)] Loss: 0.261232\n",
      "Epoch 1 -> [326400/697932 (47%)] Loss: 0.265397\n",
      "Epoch 1 -> [332800/697932 (48%)] Loss: 0.269665\n",
      "Epoch 1 -> [339200/697932 (49%)] Loss: 0.27381\n",
      "Epoch 1 -> [345600/697932 (50%)] Loss: 0.278257\n",
      "Epoch 1 -> [352000/697932 (50%)] Loss: 0.282399\n",
      "Epoch 1 -> [358400/697932 (51%)] Loss: 0.286657\n",
      "Epoch 1 -> [364800/697932 (52%)] Loss: 0.291002\n",
      "Epoch 1 -> [371200/697932 (53%)] Loss: 0.295299\n",
      "Epoch 1 -> [377600/697932 (54%)] Loss: 0.299659\n",
      "Epoch 1 -> [384000/697932 (55%)] Loss: 0.303753\n",
      "Epoch 1 -> [390400/697932 (56%)] Loss: 0.307955\n",
      "Epoch 1 -> [396800/697932 (57%)] Loss: 0.312041\n",
      "Epoch 1 -> [403200/697932 (58%)] Loss: 0.316256\n",
      "Epoch 1 -> [409600/697932 (59%)] Loss: 0.320502\n",
      "Epoch 1 -> [416000/697932 (60%)] Loss: 0.324558\n",
      "Epoch 1 -> [422400/697932 (61%)] Loss: 0.32862\n",
      "Epoch 1 -> [428800/697932 (61%)] Loss: 0.332761\n",
      "Epoch 1 -> [435200/697932 (62%)] Loss: 0.336898\n",
      "Epoch 1 -> [441600/697932 (63%)] Loss: 0.34091\n",
      "Epoch 1 -> [448000/697932 (64%)] Loss: 0.345012\n",
      "Epoch 1 -> [454400/697932 (65%)] Loss: 0.349002\n",
      "Epoch 1 -> [460800/697932 (66%)] Loss: 0.353485\n",
      "Epoch 1 -> [467200/697932 (67%)] Loss: 0.357747\n",
      "Epoch 1 -> [473600/697932 (68%)] Loss: 0.361813\n",
      "Epoch 1 -> [480000/697932 (69%)] Loss: 0.366003\n",
      "Epoch 1 -> [486400/697932 (70%)] Loss: 0.370196\n",
      "Epoch 1 -> [492800/697932 (71%)] Loss: 0.374054\n",
      "Epoch 1 -> [499200/697932 (72%)] Loss: 0.378015\n",
      "Epoch 1 -> [505600/697932 (72%)] Loss: 0.38226\n",
      "Epoch 1 -> [512000/697932 (73%)] Loss: 0.386404\n",
      "Epoch 1 -> [518400/697932 (74%)] Loss: 0.39038\n",
      "Epoch 1 -> [524800/697932 (75%)] Loss: 0.394611\n",
      "Epoch 1 -> [531200/697932 (76%)] Loss: 0.398468\n",
      "Epoch 1 -> [537600/697932 (77%)] Loss: 0.402436\n",
      "Epoch 1 -> [544000/697932 (78%)] Loss: 0.406396\n",
      "Epoch 1 -> [550400/697932 (79%)] Loss: 0.410485\n",
      "Epoch 1 -> [556800/697932 (80%)] Loss: 0.414462\n",
      "Epoch 1 -> [563200/697932 (81%)] Loss: 0.418586\n",
      "Epoch 1 -> [569600/697932 (82%)] Loss: 0.422599\n",
      "Epoch 1 -> [576000/697932 (83%)] Loss: 0.426639\n",
      "Epoch 1 -> [582400/697932 (83%)] Loss: 0.430827\n",
      "Epoch 1 -> [588800/697932 (84%)] Loss: 0.434852\n",
      "Epoch 1 -> [595200/697932 (85%)] Loss: 0.438743\n",
      "Epoch 1 -> [601600/697932 (86%)] Loss: 0.442652\n",
      "Epoch 1 -> [608000/697932 (87%)] Loss: 0.446744\n",
      "Epoch 1 -> [614400/697932 (88%)] Loss: 0.450709\n",
      "Epoch 1 -> [620800/697932 (89%)] Loss: 0.454747\n",
      "Epoch 1 -> [627200/697932 (90%)] Loss: 0.458712\n",
      "Epoch 1 -> [633600/697932 (91%)] Loss: 0.462806\n",
      "Epoch 1 -> [640000/697932 (92%)] Loss: 0.46668\n",
      "Epoch 1 -> [646400/697932 (93%)] Loss: 0.470686\n",
      "Epoch 1 -> [652800/697932 (94%)] Loss: 0.47466\n",
      "Epoch 1 -> [659200/697932 (94%)] Loss: 0.47863\n",
      "Epoch 1 -> [665600/697932 (95%)] Loss: 0.482396\n",
      "Epoch 1 -> [672000/697932 (96%)] Loss: 0.486624\n",
      "Epoch 1 -> [678400/697932 (97%)] Loss: 0.490513\n",
      "Epoch 1 -> [684800/697932 (98%)] Loss: 0.494269\n",
      "Epoch 1 -> [691200/697932 (99%)] Loss: 0.498319\n",
      "Epoch 1 -> [697600/697932 (100%)] Loss: 0.502112\n",
      "Accuracy = 14.13%\n",
      "Epoch 2 -> [0/697932 (0%)] Loss: 4.46774e-05\n",
      "Epoch 2 -> [6400/697932 (1%)] Loss: 0.00380039\n",
      "Epoch 2 -> [12800/697932 (2%)] Loss: 0.00777564\n",
      "Epoch 2 -> [19200/697932 (3%)] Loss: 0.0116853\n",
      "Epoch 2 -> [25600/697932 (4%)] Loss: 0.0155281\n",
      "Epoch 2 -> [32000/697932 (5%)] Loss: 0.0195244\n",
      "Epoch 2 -> [38400/697932 (6%)] Loss: 0.0233897\n",
      "Epoch 2 -> [44800/697932 (6%)] Loss: 0.0271763\n",
      "Epoch 2 -> [51200/697932 (7%)] Loss: 0.0312294\n",
      "Epoch 2 -> [57600/697932 (8%)] Loss: 0.0349418\n",
      "Epoch 2 -> [64000/697932 (9%)] Loss: 0.0386368\n",
      "Epoch 2 -> [70400/697932 (10%)] Loss: 0.0424859\n",
      "Epoch 2 -> [76800/697932 (11%)] Loss: 0.0462492\n",
      "Epoch 2 -> [83200/697932 (12%)] Loss: 0.0501684\n",
      "Epoch 2 -> [89600/697932 (13%)] Loss: 0.0539237\n",
      "Epoch 2 -> [96000/697932 (14%)] Loss: 0.057837\n",
      "Epoch 2 -> [102400/697932 (15%)] Loss: 0.0616857\n",
      "Epoch 2 -> [108800/697932 (16%)] Loss: 0.0657033\n",
      "Epoch 2 -> [115200/697932 (17%)] Loss: 0.069518\n",
      "Epoch 2 -> [121600/697932 (17%)] Loss: 0.0731245\n",
      "Epoch 2 -> [128000/697932 (18%)] Loss: 0.0768919\n",
      "Epoch 2 -> [134400/697932 (19%)] Loss: 0.0808197\n",
      "Epoch 2 -> [140800/697932 (20%)] Loss: 0.0847113\n",
      "Epoch 2 -> [147200/697932 (21%)] Loss: 0.0886481\n",
      "Epoch 2 -> [153600/697932 (22%)] Loss: 0.0924027\n",
      "Epoch 2 -> [160000/697932 (23%)] Loss: 0.0962677\n",
      "Epoch 2 -> [166400/697932 (24%)] Loss: 0.100045\n",
      "Epoch 2 -> [172800/697932 (25%)] Loss: 0.103732\n",
      "Epoch 2 -> [179200/697932 (26%)] Loss: 0.107829\n",
      "Epoch 2 -> [185600/697932 (27%)] Loss: 0.111676\n",
      "Epoch 2 -> [192000/697932 (28%)] Loss: 0.115571\n",
      "Epoch 2 -> [198400/697932 (28%)] Loss: 0.119356\n",
      "Epoch 2 -> [204800/697932 (29%)] Loss: 0.123124\n",
      "Epoch 2 -> [211200/697932 (30%)] Loss: 0.126964\n",
      "Epoch 2 -> [217600/697932 (31%)] Loss: 0.130593\n",
      "Epoch 2 -> [224000/697932 (32%)] Loss: 0.134586\n",
      "Epoch 2 -> [230400/697932 (33%)] Loss: 0.138228\n",
      "Epoch 2 -> [236800/697932 (34%)] Loss: 0.142111\n",
      "Epoch 2 -> [243200/697932 (35%)] Loss: 0.145952\n",
      "Epoch 2 -> [249600/697932 (36%)] Loss: 0.149675\n",
      "Epoch 2 -> [256000/697932 (37%)] Loss: 0.15344\n",
      "Epoch 2 -> [262400/697932 (38%)] Loss: 0.157271\n",
      "Epoch 2 -> [268800/697932 (39%)] Loss: 0.161041\n",
      "Epoch 2 -> [275200/697932 (39%)] Loss: 0.164729\n",
      "Epoch 2 -> [281600/697932 (40%)] Loss: 0.168559\n",
      "Epoch 2 -> [288000/697932 (41%)] Loss: 0.172394\n",
      "Epoch 2 -> [294400/697932 (42%)] Loss: 0.176191\n",
      "Epoch 2 -> [300800/697932 (43%)] Loss: 0.179906\n",
      "Epoch 2 -> [307200/697932 (44%)] Loss: 0.183652\n",
      "Epoch 2 -> [313600/697932 (45%)] Loss: 0.187559\n",
      "Epoch 2 -> [320000/697932 (46%)] Loss: 0.191401\n",
      "Epoch 2 -> [326400/697932 (47%)] Loss: 0.195198\n",
      "Epoch 2 -> [332800/697932 (48%)] Loss: 0.19918\n",
      "Epoch 2 -> [339200/697932 (49%)] Loss: 0.202883\n",
      "Epoch 2 -> [345600/697932 (50%)] Loss: 0.206677\n",
      "Epoch 2 -> [352000/697932 (50%)] Loss: 0.210437\n",
      "Epoch 2 -> [358400/697932 (51%)] Loss: 0.213981\n",
      "Epoch 2 -> [364800/697932 (52%)] Loss: 0.21781\n",
      "Epoch 2 -> [371200/697932 (53%)] Loss: 0.221645\n",
      "Epoch 2 -> [377600/697932 (54%)] Loss: 0.225491\n",
      "Epoch 2 -> [384000/697932 (55%)] Loss: 0.229299\n",
      "Epoch 2 -> [390400/697932 (56%)] Loss: 0.232934\n",
      "Epoch 2 -> [396800/697932 (57%)] Loss: 0.236739\n",
      "Epoch 2 -> [403200/697932 (58%)] Loss: 0.240545\n",
      "Epoch 2 -> [409600/697932 (59%)] Loss: 0.244269\n",
      "Epoch 2 -> [416000/697932 (60%)] Loss: 0.247744\n",
      "Epoch 2 -> [422400/697932 (61%)] Loss: 0.251519\n",
      "Epoch 2 -> [428800/697932 (61%)] Loss: 0.255194\n",
      "Epoch 2 -> [435200/697932 (62%)] Loss: 0.258986\n",
      "Epoch 2 -> [441600/697932 (63%)] Loss: 0.262739\n",
      "Epoch 2 -> [448000/697932 (64%)] Loss: 0.266475\n",
      "Epoch 2 -> [454400/697932 (65%)] Loss: 0.270425\n",
      "Epoch 2 -> [460800/697932 (66%)] Loss: 0.274132\n",
      "Epoch 2 -> [467200/697932 (67%)] Loss: 0.277713\n",
      "Epoch 2 -> [473600/697932 (68%)] Loss: 0.281472\n",
      "Epoch 2 -> [480000/697932 (69%)] Loss: 0.285189\n",
      "Epoch 2 -> [486400/697932 (70%)] Loss: 0.289057\n",
      "Epoch 2 -> [492800/697932 (71%)] Loss: 0.292756\n",
      "Epoch 2 -> [499200/697932 (72%)] Loss: 0.296469\n",
      "Epoch 2 -> [505600/697932 (72%)] Loss: 0.300302\n",
      "Epoch 2 -> [512000/697932 (73%)] Loss: 0.304072\n",
      "Epoch 2 -> [518400/697932 (74%)] Loss: 0.307742\n",
      "Epoch 2 -> [524800/697932 (75%)] Loss: 0.311527\n",
      "Epoch 2 -> [531200/697932 (76%)] Loss: 0.315209\n",
      "Epoch 2 -> [537600/697932 (77%)] Loss: 0.319214\n",
      "Epoch 2 -> [544000/697932 (78%)] Loss: 0.322993\n",
      "Epoch 2 -> [550400/697932 (79%)] Loss: 0.326736\n",
      "Epoch 2 -> [556800/697932 (80%)] Loss: 0.330558\n",
      "Epoch 2 -> [563200/697932 (81%)] Loss: 0.334318\n",
      "Epoch 2 -> [569600/697932 (82%)] Loss: 0.337964\n",
      "Epoch 2 -> [576000/697932 (83%)] Loss: 0.341899\n",
      "Epoch 2 -> [582400/697932 (83%)] Loss: 0.345619\n",
      "Epoch 2 -> [588800/697932 (84%)] Loss: 0.349354\n",
      "Epoch 2 -> [595200/697932 (85%)] Loss: 0.352981\n",
      "Epoch 2 -> [601600/697932 (86%)] Loss: 0.356545\n",
      "Epoch 2 -> [608000/697932 (87%)] Loss: 0.360408\n",
      "Epoch 2 -> [614400/697932 (88%)] Loss: 0.364188\n",
      "Epoch 2 -> [620800/697932 (89%)] Loss: 0.367835\n",
      "Epoch 2 -> [627200/697932 (90%)] Loss: 0.371614\n",
      "Epoch 2 -> [633600/697932 (91%)] Loss: 0.375501\n",
      "Epoch 2 -> [640000/697932 (92%)] Loss: 0.379235\n",
      "Epoch 2 -> [646400/697932 (93%)] Loss: 0.383231\n",
      "Epoch 2 -> [652800/697932 (94%)] Loss: 0.38696\n",
      "Epoch 2 -> [659200/697932 (94%)] Loss: 0.3907\n",
      "Epoch 2 -> [665600/697932 (95%)] Loss: 0.394305\n",
      "Epoch 2 -> [672000/697932 (96%)] Loss: 0.398144\n",
      "Epoch 2 -> [678400/697932 (97%)] Loss: 0.401977\n",
      "Epoch 2 -> [684800/697932 (98%)] Loss: 0.405792\n",
      "Epoch 2 -> [691200/697932 (99%)] Loss: 0.409388\n",
      "Epoch 2 -> [697600/697932 (100%)] Loss: 0.412945\n",
      "Accuracy = 14.24%\n",
      "Epoch 3 -> [0/697932 (0%)] Loss: 3.36757e-05\n",
      "Epoch 3 -> [6400/697932 (1%)] Loss: 0.00373487\n",
      "Epoch 3 -> [12800/697932 (2%)] Loss: 0.00762101\n",
      "Epoch 3 -> [19200/697932 (3%)] Loss: 0.0113007\n",
      "Epoch 3 -> [25600/697932 (4%)] Loss: 0.014909\n",
      "Epoch 3 -> [32000/697932 (5%)] Loss: 0.0184652\n",
      "Epoch 3 -> [38400/697932 (6%)] Loss: 0.0218976\n",
      "Epoch 3 -> [44800/697932 (6%)] Loss: 0.0255173\n",
      "Epoch 3 -> [51200/697932 (7%)] Loss: 0.0290948\n",
      "Epoch 3 -> [57600/697932 (8%)] Loss: 0.0327469\n",
      "Epoch 3 -> [64000/697932 (9%)] Loss: 0.0363857\n",
      "Epoch 3 -> [70400/697932 (10%)] Loss: 0.0399901\n",
      "Epoch 3 -> [76800/697932 (11%)] Loss: 0.0436394\n",
      "Epoch 3 -> [83200/697932 (12%)] Loss: 0.0471455\n",
      "Epoch 3 -> [89600/697932 (13%)] Loss: 0.0507905\n",
      "Epoch 3 -> [96000/697932 (14%)] Loss: 0.0545718\n",
      "Epoch 3 -> [102400/697932 (15%)] Loss: 0.0581004\n",
      "Epoch 3 -> [108800/697932 (16%)] Loss: 0.0617787\n",
      "Epoch 3 -> [115200/697932 (17%)] Loss: 0.0655782\n",
      "Epoch 3 -> [121600/697932 (17%)] Loss: 0.069165\n",
      "Epoch 3 -> [128000/697932 (18%)] Loss: 0.072884\n",
      "Epoch 3 -> [134400/697932 (19%)] Loss: 0.0766227\n",
      "Epoch 3 -> [140800/697932 (20%)] Loss: 0.0802654\n",
      "Epoch 3 -> [147200/697932 (21%)] Loss: 0.0838395\n",
      "Epoch 3 -> [153600/697932 (22%)] Loss: 0.0874845\n",
      "Epoch 3 -> [160000/697932 (23%)] Loss: 0.0911082\n",
      "Epoch 3 -> [166400/697932 (24%)] Loss: 0.0946743\n",
      "Epoch 3 -> [172800/697932 (25%)] Loss: 0.0983718\n",
      "Epoch 3 -> [179200/697932 (26%)] Loss: 0.101931\n",
      "Epoch 3 -> [185600/697932 (27%)] Loss: 0.105525\n",
      "Epoch 3 -> [192000/697932 (28%)] Loss: 0.109087\n",
      "Epoch 3 -> [198400/697932 (28%)] Loss: 0.112753\n",
      "Epoch 3 -> [204800/697932 (29%)] Loss: 0.116576\n",
      "Epoch 3 -> [211200/697932 (30%)] Loss: 0.120167\n",
      "Epoch 3 -> [217600/697932 (31%)] Loss: 0.123721\n",
      "Epoch 3 -> [224000/697932 (32%)] Loss: 0.127385\n",
      "Epoch 3 -> [230400/697932 (33%)] Loss: 0.130921\n",
      "Epoch 3 -> [236800/697932 (34%)] Loss: 0.134517\n",
      "Epoch 3 -> [243200/697932 (35%)] Loss: 0.138157\n",
      "Epoch 3 -> [249600/697932 (36%)] Loss: 0.141742\n",
      "Epoch 3 -> [256000/697932 (37%)] Loss: 0.145262\n",
      "Epoch 3 -> [262400/697932 (38%)] Loss: 0.148737\n",
      "Epoch 3 -> [268800/697932 (39%)] Loss: 0.152437\n",
      "Epoch 3 -> [275200/697932 (39%)] Loss: 0.156124\n",
      "Epoch 3 -> [281600/697932 (40%)] Loss: 0.159945\n",
      "Epoch 3 -> [288000/697932 (41%)] Loss: 0.16358\n",
      "Epoch 3 -> [294400/697932 (42%)] Loss: 0.167147\n",
      "Epoch 3 -> [300800/697932 (43%)] Loss: 0.170775\n",
      "Epoch 3 -> [307200/697932 (44%)] Loss: 0.17444\n",
      "Epoch 3 -> [313600/697932 (45%)] Loss: 0.177962\n",
      "Epoch 3 -> [320000/697932 (46%)] Loss: 0.181682\n",
      "Epoch 3 -> [326400/697932 (47%)] Loss: 0.185311\n",
      "Epoch 3 -> [332800/697932 (48%)] Loss: 0.188883\n",
      "Epoch 3 -> [339200/697932 (49%)] Loss: 0.192378\n",
      "Epoch 3 -> [345600/697932 (50%)] Loss: 0.196065\n",
      "Epoch 3 -> [352000/697932 (50%)] Loss: 0.199588\n",
      "Epoch 3 -> [358400/697932 (51%)] Loss: 0.202904\n",
      "Epoch 3 -> [364800/697932 (52%)] Loss: 0.20657\n",
      "Epoch 3 -> [371200/697932 (53%)] Loss: 0.210362\n",
      "Epoch 3 -> [377600/697932 (54%)] Loss: 0.213923\n",
      "Epoch 3 -> [384000/697932 (55%)] Loss: 0.217559\n",
      "Epoch 3 -> [390400/697932 (56%)] Loss: 0.22109\n",
      "Epoch 3 -> [396800/697932 (57%)] Loss: 0.224712\n",
      "Epoch 3 -> [403200/697932 (58%)] Loss: 0.228365\n",
      "Epoch 3 -> [409600/697932 (59%)] Loss: 0.231895\n",
      "Epoch 3 -> [416000/697932 (60%)] Loss: 0.23547\n",
      "Epoch 3 -> [422400/697932 (61%)] Loss: 0.238815\n",
      "Epoch 3 -> [428800/697932 (61%)] Loss: 0.242485\n",
      "Epoch 3 -> [435200/697932 (62%)] Loss: 0.245984\n",
      "Epoch 3 -> [441600/697932 (63%)] Loss: 0.249624\n",
      "Epoch 3 -> [448000/697932 (64%)] Loss: 0.253036\n",
      "Epoch 3 -> [454400/697932 (65%)] Loss: 0.256766\n",
      "Epoch 3 -> [460800/697932 (66%)] Loss: 0.260341\n",
      "Epoch 3 -> [467200/697932 (67%)] Loss: 0.263927\n",
      "Epoch 3 -> [473600/697932 (68%)] Loss: 0.2675\n",
      "Epoch 3 -> [480000/697932 (69%)] Loss: 0.270803\n",
      "Epoch 3 -> [486400/697932 (70%)] Loss: 0.274509\n",
      "Epoch 3 -> [492800/697932 (71%)] Loss: 0.278079\n",
      "Epoch 3 -> [499200/697932 (72%)] Loss: 0.281663\n",
      "Epoch 3 -> [505600/697932 (72%)] Loss: 0.285311\n",
      "Epoch 3 -> [512000/697932 (73%)] Loss: 0.288896\n",
      "Epoch 3 -> [518400/697932 (74%)] Loss: 0.292457\n",
      "Epoch 3 -> [524800/697932 (75%)] Loss: 0.296175\n",
      "Epoch 3 -> [531200/697932 (76%)] Loss: 0.299804\n",
      "Epoch 3 -> [537600/697932 (77%)] Loss: 0.303327\n",
      "Epoch 3 -> [544000/697932 (78%)] Loss: 0.30685\n",
      "Epoch 3 -> [550400/697932 (79%)] Loss: 0.310571\n",
      "Epoch 3 -> [556800/697932 (80%)] Loss: 0.314075\n",
      "Epoch 3 -> [563200/697932 (81%)] Loss: 0.317884\n",
      "Epoch 3 -> [569600/697932 (82%)] Loss: 0.321488\n",
      "Epoch 3 -> [576000/697932 (83%)] Loss: 0.324963\n",
      "Epoch 3 -> [582400/697932 (83%)] Loss: 0.328795\n",
      "Epoch 3 -> [588800/697932 (84%)] Loss: 0.332631\n",
      "Epoch 3 -> [595200/697932 (85%)] Loss: 0.336179\n",
      "Epoch 3 -> [601600/697932 (86%)] Loss: 0.339909\n",
      "Epoch 3 -> [608000/697932 (87%)] Loss: 0.343457\n",
      "Epoch 3 -> [614400/697932 (88%)] Loss: 0.347021\n",
      "Epoch 3 -> [620800/697932 (89%)] Loss: 0.350668\n",
      "Epoch 3 -> [627200/697932 (90%)] Loss: 0.354368\n",
      "Epoch 3 -> [633600/697932 (91%)] Loss: 0.357841\n",
      "Epoch 3 -> [640000/697932 (92%)] Loss: 0.361376\n",
      "Epoch 3 -> [646400/697932 (93%)] Loss: 0.365054\n",
      "Epoch 3 -> [652800/697932 (94%)] Loss: 0.368577\n",
      "Epoch 3 -> [659200/697932 (94%)] Loss: 0.372093\n",
      "Epoch 3 -> [665600/697932 (95%)] Loss: 0.375624\n",
      "Epoch 3 -> [672000/697932 (96%)] Loss: 0.379399\n",
      "Epoch 3 -> [678400/697932 (97%)] Loss: 0.383149\n",
      "Epoch 3 -> [684800/697932 (98%)] Loss: 0.387026\n",
      "Epoch 3 -> [691200/697932 (99%)] Loss: 0.390562\n",
      "Epoch 3 -> [697600/697932 (100%)] Loss: 0.394143\n",
      "Accuracy = 14.28%\n",
      "Epoch 4 -> [0/697932 (0%)] Loss: 2.17954e-05\n",
      "Epoch 4 -> [6400/697932 (1%)] Loss: 0.00355181\n",
      "Epoch 4 -> [12800/697932 (2%)] Loss: 0.00702947\n",
      "Epoch 4 -> [19200/697932 (3%)] Loss: 0.0104674\n",
      "Epoch 4 -> [25600/697932 (4%)] Loss: 0.0139766\n",
      "Epoch 4 -> [32000/697932 (5%)] Loss: 0.0174851\n",
      "Epoch 4 -> [38400/697932 (6%)] Loss: 0.0209994\n",
      "Epoch 4 -> [44800/697932 (6%)] Loss: 0.0243677\n",
      "Epoch 4 -> [51200/697932 (7%)] Loss: 0.027741\n",
      "Epoch 4 -> [57600/697932 (8%)] Loss: 0.0311697\n",
      "Epoch 4 -> [64000/697932 (9%)] Loss: 0.0345459\n",
      "Epoch 4 -> [70400/697932 (10%)] Loss: 0.0380745\n",
      "Epoch 4 -> [76800/697932 (11%)] Loss: 0.0415154\n",
      "Epoch 4 -> [83200/697932 (12%)] Loss: 0.0451318\n",
      "Epoch 4 -> [89600/697932 (13%)] Loss: 0.0484092\n",
      "Epoch 4 -> [96000/697932 (14%)] Loss: 0.0519796\n",
      "Epoch 4 -> [102400/697932 (15%)] Loss: 0.0555228\n",
      "Epoch 4 -> [108800/697932 (16%)] Loss: 0.0588919\n",
      "Epoch 4 -> [115200/697932 (17%)] Loss: 0.0625209\n",
      "Epoch 4 -> [121600/697932 (17%)] Loss: 0.0659739\n",
      "Epoch 4 -> [128000/697932 (18%)] Loss: 0.0694268\n",
      "Epoch 4 -> [134400/697932 (19%)] Loss: 0.0728982\n",
      "Epoch 4 -> [140800/697932 (20%)] Loss: 0.0763014\n",
      "Epoch 4 -> [147200/697932 (21%)] Loss: 0.079808\n",
      "Epoch 4 -> [153600/697932 (22%)] Loss: 0.0831335\n",
      "Epoch 4 -> [160000/697932 (23%)] Loss: 0.0865745\n",
      "Epoch 4 -> [166400/697932 (24%)] Loss: 0.0899939\n",
      "Epoch 4 -> [172800/697932 (25%)] Loss: 0.0933783\n",
      "Epoch 4 -> [179200/697932 (26%)] Loss: 0.0970162\n",
      "Epoch 4 -> [185600/697932 (27%)] Loss: 0.100658\n",
      "Epoch 4 -> [192000/697932 (28%)] Loss: 0.104256\n",
      "Epoch 4 -> [198400/697932 (28%)] Loss: 0.107922\n",
      "Epoch 4 -> [204800/697932 (29%)] Loss: 0.111284\n",
      "Epoch 4 -> [211200/697932 (30%)] Loss: 0.114713\n",
      "Epoch 4 -> [217600/697932 (31%)] Loss: 0.118103\n",
      "Epoch 4 -> [224000/697932 (32%)] Loss: 0.121681\n",
      "Epoch 4 -> [230400/697932 (33%)] Loss: 0.125023\n",
      "Epoch 4 -> [236800/697932 (34%)] Loss: 0.128335\n",
      "Epoch 4 -> [243200/697932 (35%)] Loss: 0.131853\n",
      "Epoch 4 -> [249600/697932 (36%)] Loss: 0.135376\n",
      "Epoch 4 -> [256000/697932 (37%)] Loss: 0.138929\n",
      "Epoch 4 -> [262400/697932 (38%)] Loss: 0.142467\n",
      "Epoch 4 -> [268800/697932 (39%)] Loss: 0.145954\n",
      "Epoch 4 -> [275200/697932 (39%)] Loss: 0.149449\n",
      "Epoch 4 -> [281600/697932 (40%)] Loss: 0.152971\n",
      "Epoch 4 -> [288000/697932 (41%)] Loss: 0.156463\n",
      "Epoch 4 -> [294400/697932 (42%)] Loss: 0.16012\n",
      "Epoch 4 -> [300800/697932 (43%)] Loss: 0.163445\n",
      "Epoch 4 -> [307200/697932 (44%)] Loss: 0.166905\n",
      "Epoch 4 -> [313600/697932 (45%)] Loss: 0.17034\n",
      "Epoch 4 -> [320000/697932 (46%)] Loss: 0.173764\n",
      "Epoch 4 -> [326400/697932 (47%)] Loss: 0.17733\n",
      "Epoch 4 -> [332800/697932 (48%)] Loss: 0.18102\n",
      "Epoch 4 -> [339200/697932 (49%)] Loss: 0.184312\n",
      "Epoch 4 -> [345600/697932 (50%)] Loss: 0.1879\n",
      "Epoch 4 -> [352000/697932 (50%)] Loss: 0.191539\n",
      "Epoch 4 -> [358400/697932 (51%)] Loss: 0.194959\n",
      "Epoch 4 -> [364800/697932 (52%)] Loss: 0.198682\n",
      "Epoch 4 -> [371200/697932 (53%)] Loss: 0.202151\n",
      "Epoch 4 -> [377600/697932 (54%)] Loss: 0.205759\n",
      "Epoch 4 -> [384000/697932 (55%)] Loss: 0.209108\n",
      "Epoch 4 -> [390400/697932 (56%)] Loss: 0.212618\n",
      "Epoch 4 -> [396800/697932 (57%)] Loss: 0.216218\n",
      "Epoch 4 -> [403200/697932 (58%)] Loss: 0.21977\n",
      "Epoch 4 -> [409600/697932 (59%)] Loss: 0.22351\n",
      "Epoch 4 -> [416000/697932 (60%)] Loss: 0.227051\n",
      "Epoch 4 -> [422400/697932 (61%)] Loss: 0.230692\n",
      "Epoch 4 -> [428800/697932 (61%)] Loss: 0.234155\n",
      "Epoch 4 -> [435200/697932 (62%)] Loss: 0.23755\n",
      "Epoch 4 -> [441600/697932 (63%)] Loss: 0.241129\n",
      "Epoch 4 -> [448000/697932 (64%)] Loss: 0.244702\n",
      "Epoch 4 -> [454400/697932 (65%)] Loss: 0.248334\n",
      "Epoch 4 -> [460800/697932 (66%)] Loss: 0.251881\n",
      "Epoch 4 -> [467200/697932 (67%)] Loss: 0.255635\n",
      "Epoch 4 -> [473600/697932 (68%)] Loss: 0.259125\n",
      "Epoch 4 -> [480000/697932 (69%)] Loss: 0.262758\n",
      "Epoch 4 -> [486400/697932 (70%)] Loss: 0.266199\n",
      "Epoch 4 -> [492800/697932 (71%)] Loss: 0.269938\n",
      "Epoch 4 -> [499200/697932 (72%)] Loss: 0.273517\n",
      "Epoch 4 -> [505600/697932 (72%)] Loss: 0.276934\n",
      "Epoch 4 -> [512000/697932 (73%)] Loss: 0.280349\n",
      "Epoch 4 -> [518400/697932 (74%)] Loss: 0.283821\n",
      "Epoch 4 -> [524800/697932 (75%)] Loss: 0.287337\n",
      "Epoch 4 -> [531200/697932 (76%)] Loss: 0.290874\n",
      "Epoch 4 -> [537600/697932 (77%)] Loss: 0.294421\n",
      "Epoch 4 -> [544000/697932 (78%)] Loss: 0.297864\n",
      "Epoch 4 -> [550400/697932 (79%)] Loss: 0.301462\n",
      "Epoch 4 -> [556800/697932 (80%)] Loss: 0.305111\n",
      "Epoch 4 -> [563200/697932 (81%)] Loss: 0.308559\n",
      "Epoch 4 -> [569600/697932 (82%)] Loss: 0.311946\n",
      "Epoch 4 -> [576000/697932 (83%)] Loss: 0.315547\n",
      "Epoch 4 -> [582400/697932 (83%)] Loss: 0.319035\n",
      "Epoch 4 -> [588800/697932 (84%)] Loss: 0.322422\n",
      "Epoch 4 -> [595200/697932 (85%)] Loss: 0.325894\n",
      "Epoch 4 -> [601600/697932 (86%)] Loss: 0.329326\n",
      "Epoch 4 -> [608000/697932 (87%)] Loss: 0.333062\n",
      "Epoch 4 -> [614400/697932 (88%)] Loss: 0.336587\n",
      "Epoch 4 -> [620800/697932 (89%)] Loss: 0.340008\n",
      "Epoch 4 -> [627200/697932 (90%)] Loss: 0.343634\n",
      "Epoch 4 -> [633600/697932 (91%)] Loss: 0.347112\n",
      "Epoch 4 -> [640000/697932 (92%)] Loss: 0.350581\n",
      "Epoch 4 -> [646400/697932 (93%)] Loss: 0.354328\n",
      "Epoch 4 -> [652800/697932 (94%)] Loss: 0.357811\n",
      "Epoch 4 -> [659200/697932 (94%)] Loss: 0.361196\n",
      "Epoch 4 -> [665600/697932 (95%)] Loss: 0.364655\n",
      "Epoch 4 -> [672000/697932 (96%)] Loss: 0.368155\n",
      "Epoch 4 -> [678400/697932 (97%)] Loss: 0.371701\n",
      "Epoch 4 -> [684800/697932 (98%)] Loss: 0.375265\n",
      "Epoch 4 -> [691200/697932 (99%)] Loss: 0.378743\n",
      "Epoch 4 -> [697600/697932 (100%)] Loss: 0.382294\n",
      "Accuracy = 14.25%\n",
      "Epoch 5 -> [0/697932 (0%)] Loss: 3.51913e-05\n",
      "Epoch 5 -> [6400/697932 (1%)] Loss: 0.00333927\n",
      "Epoch 5 -> [12800/697932 (2%)] Loss: 0.00688629\n",
      "Epoch 5 -> [19200/697932 (3%)] Loss: 0.010202\n",
      "Epoch 5 -> [25600/697932 (4%)] Loss: 0.0135894\n",
      "Epoch 5 -> [32000/697932 (5%)] Loss: 0.0170285\n",
      "Epoch 5 -> [38400/697932 (6%)] Loss: 0.0203835\n",
      "Epoch 5 -> [44800/697932 (6%)] Loss: 0.0235687\n",
      "Epoch 5 -> [51200/697932 (7%)] Loss: 0.0268973\n",
      "Epoch 5 -> [57600/697932 (8%)] Loss: 0.0304369\n",
      "Epoch 5 -> [64000/697932 (9%)] Loss: 0.0338533\n",
      "Epoch 5 -> [70400/697932 (10%)] Loss: 0.0372032\n",
      "Epoch 5 -> [76800/697932 (11%)] Loss: 0.0405358\n",
      "Epoch 5 -> [83200/697932 (12%)] Loss: 0.0440124\n",
      "Epoch 5 -> [89600/697932 (13%)] Loss: 0.0474151\n",
      "Epoch 5 -> [96000/697932 (14%)] Loss: 0.0508112\n",
      "Epoch 5 -> [102400/697932 (15%)] Loss: 0.0540678\n",
      "Epoch 5 -> [108800/697932 (16%)] Loss: 0.0575852\n",
      "Epoch 5 -> [115200/697932 (17%)] Loss: 0.0610861\n",
      "Epoch 5 -> [121600/697932 (17%)] Loss: 0.0642296\n",
      "Epoch 5 -> [128000/697932 (18%)] Loss: 0.0676591\n",
      "Epoch 5 -> [134400/697932 (19%)] Loss: 0.0710291\n",
      "Epoch 5 -> [140800/697932 (20%)] Loss: 0.0744292\n",
      "Epoch 5 -> [147200/697932 (21%)] Loss: 0.0779103\n",
      "Epoch 5 -> [153600/697932 (22%)] Loss: 0.0813416\n",
      "Epoch 5 -> [160000/697932 (23%)] Loss: 0.0847365\n",
      "Epoch 5 -> [166400/697932 (24%)] Loss: 0.0881198\n",
      "Epoch 5 -> [172800/697932 (25%)] Loss: 0.0913908\n",
      "Epoch 5 -> [179200/697932 (26%)] Loss: 0.0949743\n",
      "Epoch 5 -> [185600/697932 (27%)] Loss: 0.0983091\n",
      "Epoch 5 -> [192000/697932 (28%)] Loss: 0.101662\n",
      "Epoch 5 -> [198400/697932 (28%)] Loss: 0.105089\n",
      "Epoch 5 -> [204800/697932 (29%)] Loss: 0.108531\n",
      "Epoch 5 -> [211200/697932 (30%)] Loss: 0.112079\n",
      "Epoch 5 -> [217600/697932 (31%)] Loss: 0.115556\n",
      "Epoch 5 -> [224000/697932 (32%)] Loss: 0.119044\n",
      "Epoch 5 -> [230400/697932 (33%)] Loss: 0.122417\n",
      "Epoch 5 -> [236800/697932 (34%)] Loss: 0.125836\n",
      "Epoch 5 -> [243200/697932 (35%)] Loss: 0.129316\n",
      "Epoch 5 -> [249600/697932 (36%)] Loss: 0.132891\n",
      "Epoch 5 -> [256000/697932 (37%)] Loss: 0.136249\n",
      "Epoch 5 -> [262400/697932 (38%)] Loss: 0.139645\n",
      "Epoch 5 -> [268800/697932 (39%)] Loss: 0.143139\n",
      "Epoch 5 -> [275200/697932 (39%)] Loss: 0.14658\n",
      "Epoch 5 -> [281600/697932 (40%)] Loss: 0.150177\n",
      "Epoch 5 -> [288000/697932 (41%)] Loss: 0.153592\n",
      "Epoch 5 -> [294400/697932 (42%)] Loss: 0.156981\n",
      "Epoch 5 -> [300800/697932 (43%)] Loss: 0.160486\n",
      "Epoch 5 -> [307200/697932 (44%)] Loss: 0.163902\n",
      "Epoch 5 -> [313600/697932 (45%)] Loss: 0.167241\n",
      "Epoch 5 -> [320000/697932 (46%)] Loss: 0.170699\n",
      "Epoch 5 -> [326400/697932 (47%)] Loss: 0.17415\n",
      "Epoch 5 -> [332800/697932 (48%)] Loss: 0.177502\n",
      "Epoch 5 -> [339200/697932 (49%)] Loss: 0.180944\n",
      "Epoch 5 -> [345600/697932 (50%)] Loss: 0.184334\n",
      "Epoch 5 -> [352000/697932 (50%)] Loss: 0.187643\n",
      "Epoch 5 -> [358400/697932 (51%)] Loss: 0.191055\n",
      "Epoch 5 -> [364800/697932 (52%)] Loss: 0.194432\n",
      "Epoch 5 -> [371200/697932 (53%)] Loss: 0.197814\n",
      "Epoch 5 -> [377600/697932 (54%)] Loss: 0.201183\n",
      "Epoch 5 -> [384000/697932 (55%)] Loss: 0.204651\n",
      "Epoch 5 -> [390400/697932 (56%)] Loss: 0.20814\n",
      "Epoch 5 -> [396800/697932 (57%)] Loss: 0.21148\n",
      "Epoch 5 -> [403200/697932 (58%)] Loss: 0.215014\n",
      "Epoch 5 -> [409600/697932 (59%)] Loss: 0.218602\n",
      "Epoch 5 -> [416000/697932 (60%)] Loss: 0.222181\n",
      "Epoch 5 -> [422400/697932 (61%)] Loss: 0.225561\n",
      "Epoch 5 -> [428800/697932 (61%)] Loss: 0.228932\n",
      "Epoch 5 -> [435200/697932 (62%)] Loss: 0.23239\n",
      "Epoch 5 -> [441600/697932 (63%)] Loss: 0.235858\n",
      "Epoch 5 -> [448000/697932 (64%)] Loss: 0.239296\n",
      "Epoch 5 -> [454400/697932 (65%)] Loss: 0.242732\n",
      "Epoch 5 -> [460800/697932 (66%)] Loss: 0.246195\n",
      "Epoch 5 -> [467200/697932 (67%)] Loss: 0.249742\n",
      "Epoch 5 -> [473600/697932 (68%)] Loss: 0.253273\n",
      "Epoch 5 -> [480000/697932 (69%)] Loss: 0.256678\n",
      "Epoch 5 -> [486400/697932 (70%)] Loss: 0.26018\n",
      "Epoch 5 -> [492800/697932 (71%)] Loss: 0.263663\n",
      "Epoch 5 -> [499200/697932 (72%)] Loss: 0.267271\n",
      "Epoch 5 -> [505600/697932 (72%)] Loss: 0.270892\n",
      "Epoch 5 -> [512000/697932 (73%)] Loss: 0.274331\n",
      "Epoch 5 -> [518400/697932 (74%)] Loss: 0.277844\n",
      "Epoch 5 -> [524800/697932 (75%)] Loss: 0.281352\n",
      "Epoch 5 -> [531200/697932 (76%)] Loss: 0.284783\n",
      "Epoch 5 -> [537600/697932 (77%)] Loss: 0.288184\n",
      "Epoch 5 -> [544000/697932 (78%)] Loss: 0.291762\n",
      "Epoch 5 -> [550400/697932 (79%)] Loss: 0.295259\n",
      "Epoch 5 -> [556800/697932 (80%)] Loss: 0.298754\n",
      "Epoch 5 -> [563200/697932 (81%)] Loss: 0.302071\n",
      "Epoch 5 -> [569600/697932 (82%)] Loss: 0.305536\n",
      "Epoch 5 -> [576000/697932 (83%)] Loss: 0.308837\n",
      "Epoch 5 -> [582400/697932 (83%)] Loss: 0.31227\n",
      "Epoch 5 -> [588800/697932 (84%)] Loss: 0.315672\n",
      "Epoch 5 -> [595200/697932 (85%)] Loss: 0.319117\n",
      "Epoch 5 -> [601600/697932 (86%)] Loss: 0.322706\n",
      "Epoch 5 -> [608000/697932 (87%)] Loss: 0.326045\n",
      "Epoch 5 -> [614400/697932 (88%)] Loss: 0.329352\n",
      "Epoch 5 -> [620800/697932 (89%)] Loss: 0.33278\n",
      "Epoch 5 -> [627200/697932 (90%)] Loss: 0.336342\n",
      "Epoch 5 -> [633600/697932 (91%)] Loss: 0.339838\n",
      "Epoch 5 -> [640000/697932 (92%)] Loss: 0.343295\n",
      "Epoch 5 -> [646400/697932 (93%)] Loss: 0.346674\n",
      "Epoch 5 -> [652800/697932 (94%)] Loss: 0.350125\n",
      "Epoch 5 -> [659200/697932 (94%)] Loss: 0.353541\n",
      "Epoch 5 -> [665600/697932 (95%)] Loss: 0.356867\n",
      "Epoch 5 -> [672000/697932 (96%)] Loss: 0.360174\n",
      "Epoch 5 -> [678400/697932 (97%)] Loss: 0.363441\n",
      "Epoch 5 -> [684800/697932 (98%)] Loss: 0.367059\n",
      "Epoch 5 -> [691200/697932 (99%)] Loss: 0.370392\n",
      "Epoch 5 -> [697600/697932 (100%)] Loss: 0.373864\n",
      "Accuracy = 14.27%\n"
     ],
     "name": "stdout"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "9d3f-Ta8bIls",
    "colab_type": "code",
    "colab": {}
   },
   "source": [
    "def prediction_figure(img, prob, pred):\n",
    "    prob = prob.cpu().data.numpy().squeeze()\n",
    "    fig, ax1 = plt.subplots(figsize=(3, 3), ncols=1)\n",
    "    ax1.set_title('Predicted Digit = {}'.format(str(pred)))\n",
    "    ax1.imshow(img.resize_(1, 28, 28).numpy().squeeze(), cmap='gray_r')\n",
    "    ax1.axis('off')\n",
    "    plt.tight_layout()"
   ],
   "execution_count": 0,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "Uxomzd1YbMVZ",
    "colab_type": "code",
    "colab": {}
   },
   "source": [
    "def view_prediction():\n",
    "    images, labels = next(iter(test_loader))\n",
    "    # Turn off gradients to speed up this part\n",
    "    with torch.no_grad():\n",
    "        output = cnn(images.to(device))\n",
    "    # Convert the log-probability for checking probability of each class\n",
    "    probability = torch.exp(output)\n",
    "    prediction = list(output.cpu().numpy()[0])\n",
    "    classes = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz'\n",
    "    max_pred = prediction.index(max(prediction))\n",
    "    real_pred = classes[max_pred]\n",
    "\n",
    "    prediction_figure(images, probability, real_pred)"
   ],
   "execution_count": 0,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "V0YdbTopbrhk",
    "colab_type": "code",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 849
    },
    "outputId": "536fe9cd-e336-4eae-912d-7a97f8c73016"
   },
   "source": [
    "for x in range(4):\n",
    "  view_prediction()"
   ],
   "execution_count": 20,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMAAAADQCAYAAABRCm6cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     }
    },
    {
     "output_type": "display_data",
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMAAAADQCAYAAABRCm6cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     }
    },
    {
     "output_type": "display_data",
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMAAAADQCAYAAABRCm6cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMKaFAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     }
    },
    {
     "output_type": "display_data",
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMAAAADQCAYAAABRCm6cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAOKaAiCuKQDimgIgrikA4tp/AElMa5WG57foAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     }
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "PPDy_Xtuhwe3",
    "colab_type": "code",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "outputId": "3d618cde-fd6a-4666-a876-0617a8511942"
   },
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/gdrive')"
   ],
   "execution_count": 12,
   "outputs": [
    {
     "output_type": "stream",
     "text": [
      "Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount(\"/content/gdrive\", force_remount=True).\n"
     ],
     "name": "stdout"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "FPYi6pbGh-k-",
    "colab_type": "code",
    "colab": {}
   },
   "source": [
    "model_save_name = 'letter_model.pt'\n",
    "path = F\"/content/gdrive/My Drive/{model_save_name}\" \n",
    "torch.save(cnn.state_dict(), path)"
   ],
   "execution_count": 0,
   "outputs": []
  }
 ]
}