AbstractClassifier.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib notebook\n",
    "import os\n",
    "import string\n",
    "import numpy as np\n",
    "import torch\n",
    "from torch import nn\n",
    "from torch.nn import functional as F\n",
    "import torch.utils.data as td\n",
    "import torchvision as tv\n",
    "import pandas as pd\n",
    "from PIL import Image\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda\n"
     ]
    }
   ],
   "source": [
    "device = 'cuda' if torch.cuda.is_available() else 'cpu'\n",
    "print(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_root_dir = '/home/stosifia/ML_Art_Proj/ml-art-final-shawheen-t-final/abstract_64x64'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ArtDataset(td.Dataset):\n",
    "    \n",
    "    def __init__(self, root_dir, mode=\"ABSTRACT_groundTruth\", image_size=(64, 64)):\n",
    "        super(ArtDataset, self).__init__()\n",
    "        self.image_size = image_size\n",
    "        self.mode = mode\n",
    "        self.dir = root_dir\n",
    "        self.data = pd.read_csv(os.path.join(root_dir, \"%s.csv\" % mode))\n",
    "        self.data[['Amusement','Anger','Awe','Content','Disgust','Excitement','Fear',\n",
    "                   'Sad']] = self.data[['Amusement','Anger','Awe','Content','Disgust','Excitement','Fear','Sad']].apply(pd.to_numeric)\n",
    "        self.data['class'] = self.data[['Amusement',\n",
    "                                           'Anger','Awe','Content','Disgust','Excitement','Fear','Sad']].apply(lambda x: x.argmax(), axis=1)\n",
    "        self.images_dir = os.path.join(root_dir, '/Images')\n",
    "        map_dict = {'Amusement': 0,'Anger': 1,'Awe': 2,'Content': 3,'Disgust': 4,'Excitement': 5,'Fear': 6,\n",
    "                   'Sad': 7}\n",
    "        self.data['class'] = self.data['class'].map(map_dict)\n",
    "        \n",
    "    def __len__(self):\n",
    "        return len(self.data)\n",
    "    \n",
    "    def __repr__(self):\n",
    "        return \"ArtDataset(mode={}, image_size={})\". \\\n",
    "            format(self.mode, self.image_size)\n",
    "    \n",
    "    def __getitem__(self, idx):\n",
    "        #img_path = os.path.join(self.images_dir, \\\n",
    "        #                        self.data.iloc[idx]['file_path'])\n",
    "        file_path = self.data.iloc[idx]['file_path'].replace(\"'\", \"\")\n",
    "        img_path = self.dir + self.images_dir + '/' + file_path\n",
    "        #bbox = self.data.iloc[idx][['x1', 'y1', 'x2', 'y2']]\n",
    "        img = Image.open(img_path).convert('RGB')\n",
    "        print(type(img))\n",
    "        #img = img.crop([bbox[0], bbox[1], bbox[2], bbox[3]])\n",
    "        transform = tv.transforms.Compose([\n",
    "            tv.transforms.Resize(self.image_size),\n",
    "            tv.transforms.ToTensor(),\n",
    "            tv.transforms.Normalize((.5,.5,.5),(.5,.5,.5),inplace=False)\n",
    "            ])\n",
    "        x = transform(img)\n",
    "        d = self.data.iloc[idx]['class']\n",
    "        return x, d\n",
    "    \n",
    "    def number_of_classes(self):\n",
    "        return self.data['class'].max() + 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def myimshow(image, ax=plt):\n",
    "    image = image.to('cpu').numpy()\n",
    "    image = np.moveaxis(image, [0, 1, 2], [2, 0, 1])\n",
    "    image = (image + 1) / 2\n",
    "    image[image < 0] = 0\n",
    "    image[image > 1] = 1 \n",
    "    h = ax.imshow(image)\n",
    "    ax.axis('off')\n",
    "    return h"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/ipykernel_launcher.py:12: FutureWarning: \n",
      "The current behaviour of 'Series.argmax' is deprecated, use 'idxmax'\n",
      "instead.\n",
      "The behavior of 'argmax' will be corrected to return the positional\n",
      "maximum in the future. For now, use 'series.values.argmax' or\n",
      "'np.argmax(np.array(values))' to get the position of the maximum\n",
      "row.\n",
      "  if sys.path[0] == '':\n"
     ]
    }
   ],
   "source": [
    "train_set = ArtDataset(dataset_root_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import nntools as nt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "class NNClassifier(nt.NeuralNetwork):\n",
    "    def __init__(self):\n",
    "        super(NNClassifier, self).__init__()\n",
    "        self.cross_entropy = nn.CrossEntropyLoss()\n",
    "        \n",
    "    def criterion(self, y, d):\n",
    "        return self.cross_entropy(y, d)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "vgg = tv.models.vgg16_bn(pretrained=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "class VGG16Transfer(NNClassifier):\n",
    "    def __init__(self, num_classes, fine_tuning=False):\n",
    "        super(VGG16Transfer, self).__init__()\n",
    "        vgg = tv.models.vgg16_bn(pretrained=True)\n",
    "        for param in vgg.parameters():\n",
    "            param.requires_grad = fine_tuning\n",
    "        self.features = vgg.features;\n",
    "        self.classifier = vgg.classifier\n",
    "        self.avgpool = vgg.avgpool\n",
    "       \n",
    "       \n",
    "\n",
    "        num_ftrs = vgg.classifier[6].in_features\n",
    "        self.classifier[6] = nn.Linear(num_ftrs, num_classes)\n",
    "        \n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = self.features(x)\n",
    "        x = self.avgpool(x)\n",
    "        f = torch.flatten(x, 1)\n",
    "        y = self.classifier(f)\n",
    "        return y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_classes = train_set.number_of_classes()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ClassificationStatsManager(nt.StatsManager):\n",
    "    def __init__(self):\n",
    "        super(ClassificationStatsManager, self).__init__()\n",
    "        \n",
    "    def init(self):\n",
    "        super(ClassificationStatsManager, self).init()\n",
    "        self.running_accuracy = 0\n",
    "        \n",
    "    def accumulate(self, loss, x, y, d):\n",
    "        super(ClassificationStatsManager, self).accumulate(loss, x, y, d)\n",
    "        _, l = torch.max(y, 1)\n",
    "        self.l = l\n",
    "        #self.running_accuracy += torch.mean((l == d).float())\n",
    "        self.running_accuracy = torch.mean((l == d).float())\n",
    "        \n",
    "    def summarize(self):\n",
    "        loss = super(ClassificationStatsManager, self).summarize()\n",
    "        #accuracy = 100 * torch.mean((self.l == d).float())\n",
    "        # COMPLETE\n",
    "        accuracy = 100 * self.running_accuracy\n",
    "        return {'loss': loss, 'accuracy': accuracy}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "class GANAlbumCovers(td.Dataset):\n",
    "    \n",
    "    def __init__(self, root_dir, image_size=(64, 64)):\n",
    "        super(GANAlbumCovers, self).__init__()\n",
    "        self.image_size = image_size\n",
    "        self.dir = root_dir\n",
    "        self.dirs = os.listdir(root_dir)\n",
    "        self.data = self.dirs\n",
    "        for k in self.dirs:\n",
    "            if k.split('.')[-1] != 'png':\n",
    "                self.dirs.remove(k)\n",
    "                \n",
    "        \n",
    "    def __len__(self):\n",
    "        return len(self.data)\n",
    "    \n",
    "    def __repr__(self):\n",
    "        return \"GANAlbumCovers(image_size={})\". \\\n",
    "            format(self.image_size)\n",
    "    \n",
    "    \n",
    "    \n",
    "    def __getitem__(self, idx):\n",
    "        file_path = self.dirs[idx]\n",
    "        img_path = self.dir + '/' + file_path\n",
    "        img = Image.open(img_path).convert('RGB')\n",
    "        transform = tv.transforms.Compose([\n",
    "            tv.transforms.Resize(self.image_size),\n",
    "            tv.transforms.ToTensor(),\n",
    "            tv.transforms.Normalize((.5,.5,.5),(.5,.5,.5),inplace=False)\n",
    "            ])\n",
    "        x = transform(img)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_set = GANAlbumCovers('GANAlbumArt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "val_set = GANAlbumCovers('GANAlbumArt')\n",
    "pred_set = GANAlbumCovers('GANAlbumArt')\n",
    "lr = 1e-3\n",
    "net = VGG16Transfer(num_classes)\n",
    "net = net.to(device)\n",
    "adam = torch.optim.Adam(net.parameters(), lr=lr)\n",
    "stats_manager = ClassificationStatsManager()\n",
    "exp1 = nt.Experiment(net, train_set, val_set, pred_set, adam, stats_manager,\n",
    "               output_dir=\"abstractclass1\", perform_validation_during_training=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot(exp, fig, axes):\n",
    "    axes[0].clear()\n",
    "    axes[1].clear()\n",
    "    axes[0].set_title('Training Loss')\n",
    "    axes[1].set_title('Training Accuracy')\n",
    "    axes[0].plot([exp.history[k]['loss'] for k in range(exp.epoch)],\n",
    "                 label='training loss')\n",
    "    axes[1].plot([exp.history[k]['accuracy'] for k in range(exp.epoch)],\n",
    "                 label='training accuracy')\n",
    "    plt.tight_layout()\n",
    "    fig.canvas.draw()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "/* Put everything inside the global mpl namespace */\n",
       "window.mpl = {};\n",
       "\n",
       "\n",
       "mpl.get_websocket_type = function() {\n",
       "    if (typeof(WebSocket) !== 'undefined') {\n",
       "        return WebSocket;\n",
       "    } else if (typeof(MozWebSocket) !== 'undefined') {\n",
       "        return MozWebSocket;\n",
       "    } else {\n",
       "        alert('Your browser does not have WebSocket support. ' +\n",
       "              'Please try Chrome, Safari or Firefox ≥ 6. ' +\n",
       "              'Firefox 4 and 5 are also supported but you ' +\n",
       "              'have to enable WebSockets in about:config.');\n",
       "    };\n",
       "}\n",
       "\n",
       "mpl.figure = function(figure_id, websocket, ondownload, parent_element) {\n",
       "    this.id = figure_id;\n",
       "\n",
       "    this.ws = websocket;\n",
       "\n",
       "    this.supports_binary = (this.ws.binaryType != undefined);\n",
       "\n",
       "    if (!this.supports_binary) {\n",
       "        var warnings = document.getElementById(\"mpl-warnings\");\n",
       "        if (warnings) {\n",
       "            warnings.style.display = 'block';\n",
       "            warnings.textContent = (\n",
       "                \"This browser does not support binary websocket messages. \" +\n",
       "                    \"Performance may be slow.\");\n",
       "        }\n",
       "    }\n",
       "\n",
       "    this.imageObj = new Image();\n",
       "\n",
       "    this.context = undefined;\n",
       "    this.message = undefined;\n",
       "    this.canvas = undefined;\n",
       "    this.rubberband_canvas = undefined;\n",
       "    this.rubberband_context = undefined;\n",
       "    this.format_dropdown = undefined;\n",
       "\n",
       "    this.image_mode = 'full';\n",
       "\n",
       "    this.root = $('<div/>');\n",
       "    this._root_extra_style(this.root)\n",
       "    this.root.attr('style', 'display: inline-block');\n",
       "\n",
       "    $(parent_element).append(this.root);\n",
       "\n",
       "    this._init_header(this);\n",
       "    this._init_canvas(this);\n",
       "    this._init_toolbar(this);\n",
       "\n",
       "    var fig = this;\n",
       "\n",
       "    this.waiting = false;\n",
       "\n",
       "    this.ws.onopen =  function () {\n",
       "            fig.send_message(\"supports_binary\", {value: fig.supports_binary});\n",
       "            fig.send_message(\"send_image_mode\", {});\n",
       "            if (mpl.ratio != 1) {\n",
       "                fig.send_message(\"set_dpi_ratio\", {'dpi_ratio': mpl.ratio});\n",
       "            }\n",
       "            fig.send_message(\"refresh\", {});\n",
       "        }\n",
       "\n",
       "    this.imageObj.onload = function() {\n",
       "            if (fig.image_mode == 'full') {\n",
       "                // Full images could contain transparency (where diff images\n",
       "                // almost always do), so we need to clear the canvas so that\n",
       "                // there is no ghosting.\n",
       "                fig.context.clearRect(0, 0, fig.canvas.width, fig.canvas.height);\n",
       "            }\n",
       "            fig.context.drawImage(fig.imageObj, 0, 0);\n",
       "        };\n",
       "\n",
       "    this.imageObj.onunload = function() {\n",
       "        fig.ws.close();\n",
       "    }\n",
       "\n",
       "    this.ws.onmessage = this._make_on_message_function(this);\n",
       "\n",
       "    this.ondownload = ondownload;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_header = function() {\n",
       "    var titlebar = $(\n",
       "        '<div class=\"ui-dialog-titlebar ui-widget-header ui-corner-all ' +\n",
       "        'ui-helper-clearfix\"/>');\n",
       "    var titletext = $(\n",
       "        '<div class=\"ui-dialog-title\" style=\"width: 100%; ' +\n",
       "        'text-align: center; padding: 3px;\"/>');\n",
       "    titlebar.append(titletext)\n",
       "    this.root.append(titlebar);\n",
       "    this.header = titletext[0];\n",
       "}\n",
       "\n",
       "\n",
       "\n",
       "mpl.figure.prototype._canvas_extra_style = function(canvas_div) {\n",
       "\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype._root_extra_style = function(canvas_div) {\n",
       "\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_canvas = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var canvas_div = $('<div/>');\n",
       "\n",
       "    canvas_div.attr('style', 'position: relative; clear: both; outline: 0');\n",
       "\n",
       "    function canvas_keyboard_event(event) {\n",
       "        return fig.key_event(event, event['data']);\n",
       "    }\n",
       "\n",
       "    canvas_div.keydown('key_press', canvas_keyboard_event);\n",
       "    canvas_div.keyup('key_release', canvas_keyboard_event);\n",
       "    this.canvas_div = canvas_div\n",
       "    this._canvas_extra_style(canvas_div)\n",
       "    this.root.append(canvas_div);\n",
       "\n",
       "    var canvas = $('<canvas/>');\n",
       "    canvas.addClass('mpl-canvas');\n",
       "    canvas.attr('style', \"left: 0; top: 0; z-index: 0; outline: 0\")\n",
       "\n",
       "    this.canvas = canvas[0];\n",
       "    this.context = canvas[0].getContext(\"2d\");\n",
       "\n",
       "    var backingStore = this.context.backingStorePixelRatio ||\n",
       "\tthis.context.webkitBackingStorePixelRatio ||\n",
       "\tthis.context.mozBackingStorePixelRatio ||\n",
       "\tthis.context.msBackingStorePixelRatio ||\n",
       "\tthis.context.oBackingStorePixelRatio ||\n",
       "\tthis.context.backingStorePixelRatio || 1;\n",
       "\n",
       "    mpl.ratio = (window.devicePixelRatio || 1) / backingStore;\n",
       "\n",
       "    var rubberband = $('<canvas/>');\n",
       "    rubberband.attr('style', \"position: absolute; left: 0; top: 0; z-index: 1;\")\n",
       "\n",
       "    var pass_mouse_events = true;\n",
       "\n",
       "    canvas_div.resizable({\n",
       "        start: function(event, ui) {\n",
       "            pass_mouse_events = false;\n",
       "        },\n",
       "        resize: function(event, ui) {\n",
       "            fig.request_resize(ui.size.width, ui.size.height);\n",
       "        },\n",
       "        stop: function(event, ui) {\n",
       "            pass_mouse_events = true;\n",
       "            fig.request_resize(ui.size.width, ui.size.height);\n",
       "        },\n",
       "    });\n",
       "\n",
       "    function mouse_event_fn(event) {\n",
       "        if (pass_mouse_events)\n",
       "            return fig.mouse_event(event, event['data']);\n",
       "    }\n",
       "\n",
       "    rubberband.mousedown('button_press', mouse_event_fn);\n",
       "    rubberband.mouseup('button_release', mouse_event_fn);\n",
       "    // Throttle sequential mouse events to 1 every 20ms.\n",
       "    rubberband.mousemove('motion_notify', mouse_event_fn);\n",
       "\n",
       "    rubberband.mouseenter('figure_enter', mouse_event_fn);\n",
       "    rubberband.mouseleave('figure_leave', mouse_event_fn);\n",
       "\n",
       "    canvas_div.on(\"wheel\", function (event) {\n",
       "        event = event.originalEvent;\n",
       "        event['data'] = 'scroll'\n",
       "        if (event.deltaY < 0) {\n",
       "            event.step = 1;\n",
       "        } else {\n",
       "            event.step = -1;\n",
       "        }\n",
       "        mouse_event_fn(event);\n",
       "    });\n",
       "\n",
       "    canvas_div.append(canvas);\n",
       "    canvas_div.append(rubberband);\n",
       "\n",
       "    this.rubberband = rubberband;\n",
       "    this.rubberband_canvas = rubberband[0];\n",
       "    this.rubberband_context = rubberband[0].getContext(\"2d\");\n",
       "    this.rubberband_context.strokeStyle = \"#000000\";\n",
       "\n",
       "    this._resize_canvas = function(width, height) {\n",
       "        // Keep the size of the canvas, canvas container, and rubber band\n",
       "        // canvas in synch.\n",
       "        canvas_div.css('width', width)\n",
       "        canvas_div.css('height', height)\n",
       "\n",
       "        canvas.attr('width', width * mpl.ratio);\n",
       "        canvas.attr('height', height * mpl.ratio);\n",
       "        canvas.attr('style', 'width: ' + width + 'px; height: ' + height + 'px;');\n",
       "\n",
       "        rubberband.attr('width', width);\n",
       "        rubberband.attr('height', height);\n",
       "    }\n",
       "\n",
       "    // Set the figure to an initial 600x600px, this will subsequently be updated\n",
       "    // upon first draw.\n",
       "    this._resize_canvas(600, 600);\n",
       "\n",
       "    // Disable right mouse context menu.\n",
       "    $(this.rubberband_canvas).bind(\"contextmenu\",function(e){\n",
       "        return false;\n",
       "    });\n",
       "\n",
       "    function set_focus () {\n",
       "        canvas.focus();\n",
       "        canvas_div.focus();\n",
       "    }\n",
       "\n",
       "    window.setTimeout(set_focus, 100);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_toolbar = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var nav_element = $('<div/>');\n",
       "    nav_element.attr('style', 'width: 100%');\n",
       "    this.root.append(nav_element);\n",
       "\n",
       "    // Define a callback function for later on.\n",
       "    function toolbar_event(event) {\n",
       "        return fig.toolbar_button_onclick(event['data']);\n",
       "    }\n",
       "    function toolbar_mouse_event(event) {\n",
       "        return fig.toolbar_button_onmouseover(event['data']);\n",
       "    }\n",
       "\n",
       "    for(var toolbar_ind in mpl.toolbar_items) {\n",
       "        var name = mpl.toolbar_items[toolbar_ind][0];\n",
       "        var tooltip = mpl.toolbar_items[toolbar_ind][1];\n",
       "        var image = mpl.toolbar_items[toolbar_ind][2];\n",
       "        var method_name = mpl.toolbar_items[toolbar_ind][3];\n",
       "\n",
       "        if (!name) {\n",
       "            // put a spacer in here.\n",
       "            continue;\n",
       "        }\n",
       "        var button = $('<button/>');\n",
       "        button.addClass('ui-button ui-widget ui-state-default ui-corner-all ' +\n",
       "                        'ui-button-icon-only');\n",
       "        button.attr('role', 'button');\n",
       "        button.attr('aria-disabled', 'false');\n",
       "        button.click(method_name, toolbar_event);\n",
       "        button.mouseover(tooltip, toolbar_mouse_event);\n",
       "\n",
       "        var icon_img = $('<span/>');\n",
       "        icon_img.addClass('ui-button-icon-primary ui-icon');\n",
       "        icon_img.addClass(image);\n",
       "        icon_img.addClass('ui-corner-all');\n",
       "\n",
       "        var tooltip_span = $('<span/>');\n",
       "        tooltip_span.addClass('ui-button-text');\n",
       "        tooltip_span.html(tooltip);\n",
       "\n",
       "        button.append(icon_img);\n",
       "        button.append(tooltip_span);\n",
       "\n",
       "        nav_element.append(button);\n",
       "    }\n",
       "\n",
       "    var fmt_picker_span = $('<span/>');\n",
       "\n",
       "    var fmt_picker = $('<select/>');\n",
       "    fmt_picker.addClass('mpl-toolbar-option ui-widget ui-widget-content');\n",
       "    fmt_picker_span.append(fmt_picker);\n",
       "    nav_element.append(fmt_picker_span);\n",
       "    this.format_dropdown = fmt_picker[0];\n",
       "\n",
       "    for (var ind in mpl.extensions) {\n",
       "        var fmt = mpl.extensions[ind];\n",
       "        var option = $(\n",
       "            '<option/>', {selected: fmt === mpl.default_extension}).html(fmt);\n",
       "        fmt_picker.append(option);\n",
       "    }\n",
       "\n",
       "    // Add hover states to the ui-buttons\n",
       "    $( \".ui-button\" ).hover(\n",
       "        function() { $(this).addClass(\"ui-state-hover\");},\n",
       "        function() { $(this).removeClass(\"ui-state-hover\");}\n",
       "    );\n",
       "\n",
       "    var status_bar = $('<span class=\"mpl-message\"/>');\n",
       "    nav_element.append(status_bar);\n",
       "    this.message = status_bar[0];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.request_resize = function(x_pixels, y_pixels) {\n",
       "    // Request matplotlib to resize the figure. Matplotlib will then trigger a resize in the client,\n",
       "    // which will in turn request a refresh of the image.\n",
       "    this.send_message('resize', {'width': x_pixels, 'height': y_pixels});\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.send_message = function(type, properties) {\n",
       "    properties['type'] = type;\n",
       "    properties['figure_id'] = this.id;\n",
       "    this.ws.send(JSON.stringify(properties));\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.send_draw_message = function() {\n",
       "    if (!this.waiting) {\n",
       "        this.waiting = true;\n",
       "        this.ws.send(JSON.stringify({type: \"draw\", figure_id: this.id}));\n",
       "    }\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype.handle_save = function(fig, msg) {\n",
       "    var format_dropdown = fig.format_dropdown;\n",
       "    var format = format_dropdown.options[format_dropdown.selectedIndex].value;\n",
       "    fig.ondownload(fig, format);\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype.handle_resize = function(fig, msg) {\n",
       "    var size = msg['size'];\n",
       "    if (size[0] != fig.canvas.width || size[1] != fig.canvas.height) {\n",
       "        fig._resize_canvas(size[0], size[1]);\n",
       "        fig.send_message(\"refresh\", {});\n",
       "    };\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_rubberband = function(fig, msg) {\n",
       "    var x0 = msg['x0'] / mpl.ratio;\n",
       "    var y0 = (fig.canvas.height - msg['y0']) / mpl.ratio;\n",
       "    var x1 = msg['x1'] / mpl.ratio;\n",
       "    var y1 = (fig.canvas.height - msg['y1']) / mpl.ratio;\n",
       "    x0 = Math.floor(x0) + 0.5;\n",
       "    y0 = Math.floor(y0) + 0.5;\n",
       "    x1 = Math.floor(x1) + 0.5;\n",
       "    y1 = Math.floor(y1) + 0.5;\n",
       "    var min_x = Math.min(x0, x1);\n",
       "    var min_y = Math.min(y0, y1);\n",
       "    var width = Math.abs(x1 - x0);\n",
       "    var height = Math.abs(y1 - y0);\n",
       "\n",
       "    fig.rubberband_context.clearRect(\n",
       "        0, 0, fig.canvas.width / mpl.ratio, fig.canvas.height / mpl.ratio);\n",
       "\n",
       "    fig.rubberband_context.strokeRect(min_x, min_y, width, height);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_figure_label = function(fig, msg) {\n",
       "    // Updates the figure title.\n",
       "    fig.header.textContent = msg['label'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_cursor = function(fig, msg) {\n",
       "    var cursor = msg['cursor'];\n",
       "    switch(cursor)\n",
       "    {\n",
       "    case 0:\n",
       "        cursor = 'pointer';\n",
       "        break;\n",
       "    case 1:\n",
       "        cursor = 'default';\n",
       "        break;\n",
       "    case 2:\n",
       "        cursor = 'crosshair';\n",
       "        break;\n",
       "    case 3:\n",
       "        cursor = 'move';\n",
       "        break;\n",
       "    }\n",
       "    fig.rubberband_canvas.style.cursor = cursor;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_message = function(fig, msg) {\n",
       "    fig.message.textContent = msg['message'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_draw = function(fig, msg) {\n",
       "    // Request the server to send over a new figure.\n",
       "    fig.send_draw_message();\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_image_mode = function(fig, msg) {\n",
       "    fig.image_mode = msg['mode'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.updated_canvas_event = function() {\n",
       "    // Called whenever the canvas gets updated.\n",
       "    this.send_message(\"ack\", {});\n",
       "}\n",
       "\n",
       "// A function to construct a web socket function for onmessage handling.\n",
       "// Called in the figure constructor.\n",
       "mpl.figure.prototype._make_on_message_function = function(fig) {\n",
       "    return function socket_on_message(evt) {\n",
       "        if (evt.data instanceof Blob) {\n",
       "            /* FIXME: We get \"Resource interpreted as Image but\n",
       "             * transferred with MIME type text/plain:\" errors on\n",
       "             * Chrome.  But how to set the MIME type?  It doesn't seem\n",
       "             * to be part of the websocket stream */\n",
       "            evt.data.type = \"image/png\";\n",
       "\n",
       "            /* Free the memory for the previous frames */\n",
       "            if (fig.imageObj.src) {\n",
       "                (window.URL || window.webkitURL).revokeObjectURL(\n",
       "                    fig.imageObj.src);\n",
       "            }\n",
       "\n",
       "            fig.imageObj.src = (window.URL || window.webkitURL).createObjectURL(\n",
       "                evt.data);\n",
       "            fig.updated_canvas_event();\n",
       "            fig.waiting = false;\n",
       "            return;\n",
       "        }\n",
       "        else if (typeof evt.data === 'string' && evt.data.slice(0, 21) == \"data:image/png;base64\") {\n",
       "            fig.imageObj.src = evt.data;\n",
       "            fig.updated_canvas_event();\n",
       "            fig.waiting = false;\n",
       "            return;\n",
       "        }\n",
       "\n",
       "        var msg = JSON.parse(evt.data);\n",
       "        var msg_type = msg['type'];\n",
       "\n",
       "        // Call the  \"handle_{type}\" callback, which takes\n",
       "        // the figure and JSON message as its only arguments.\n",
       "        try {\n",
       "            var callback = fig[\"handle_\" + msg_type];\n",
       "        } catch (e) {\n",
       "            console.log(\"No handler for the '\" + msg_type + \"' message type: \", msg);\n",
       "            return;\n",
       "        }\n",
       "\n",
       "        if (callback) {\n",
       "            try {\n",
       "                // console.log(\"Handling '\" + msg_type + \"' message: \", msg);\n",
       "                callback(fig, msg);\n",
       "            } catch (e) {\n",
       "                console.log(\"Exception inside the 'handler_\" + msg_type + \"' callback:\", e, e.stack, msg);\n",
       "            }\n",
       "        }\n",
       "    };\n",
       "}\n",
       "\n",
       "// from http://stackoverflow.com/questions/1114465/getting-mouse-location-in-canvas\n",
       "mpl.findpos = function(e) {\n",
       "    //this section is from http://www.quirksmode.org/js/events_properties.html\n",
       "    var targ;\n",
       "    if (!e)\n",
       "        e = window.event;\n",
       "    if (e.target)\n",
       "        targ = e.target;\n",
       "    else if (e.srcElement)\n",
       "        targ = e.srcElement;\n",
       "    if (targ.nodeType == 3) // defeat Safari bug\n",
       "        targ = targ.parentNode;\n",
       "\n",
       "    // jQuery normalizes the pageX and pageY\n",
       "    // pageX,Y are the mouse positions relative to the document\n",
       "    // offset() returns the position of the element relative to the document\n",
       "    var x = e.pageX - $(targ).offset().left;\n",
       "    var y = e.pageY - $(targ).offset().top;\n",
       "\n",
       "    return {\"x\": x, \"y\": y};\n",
       "};\n",
       "\n",
       "/*\n",
       " * return a copy of an object with only non-object keys\n",
       " * we need this to avoid circular references\n",
       " * http://stackoverflow.com/a/24161582/3208463\n",
       " */\n",
       "function simpleKeys (original) {\n",
       "  return Object.keys(original).reduce(function (obj, key) {\n",
       "    if (typeof original[key] !== 'object')\n",
       "        obj[key] = original[key]\n",
       "    return obj;\n",
       "  }, {});\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.mouse_event = function(event, name) {\n",
       "    var canvas_pos = mpl.findpos(event)\n",
       "\n",
       "    if (name === 'button_press')\n",
       "    {\n",
       "        this.canvas.focus();\n",
       "        this.canvas_div.focus();\n",
       "    }\n",
       "\n",
       "    var x = canvas_pos.x * mpl.ratio;\n",
       "    var y = canvas_pos.y * mpl.ratio;\n",
       "\n",
       "    this.send_message(name, {x: x, y: y, button: event.button,\n",
       "                             step: event.step,\n",
       "                             guiEvent: simpleKeys(event)});\n",
       "\n",
       "    /* This prevents the web browser from automatically changing to\n",
       "     * the text insertion cursor when the button is pressed.  We want\n",
       "     * to control all of the cursor setting manually through the\n",
       "     * 'cursor' event from matplotlib */\n",
       "    event.preventDefault();\n",
       "    return false;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._key_event_extra = function(event, name) {\n",
       "    // Handle any extra behaviour associated with a key event\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.key_event = function(event, name) {\n",
       "\n",
       "    // Prevent repeat events\n",
       "    if (name == 'key_press')\n",
       "    {\n",
       "        if (event.which === this._key)\n",
       "            return;\n",
       "        else\n",
       "            this._key = event.which;\n",
       "    }\n",
       "    if (name == 'key_release')\n",
       "        this._key = null;\n",
       "\n",
       "    var value = '';\n",
       "    if (event.ctrlKey && event.which != 17)\n",
       "        value += \"ctrl+\";\n",
       "    if (event.altKey && event.which != 18)\n",
       "        value += \"alt+\";\n",
       "    if (event.shiftKey && event.which != 16)\n",
       "        value += \"shift+\";\n",
       "\n",
       "    value += 'k';\n",
       "    value += event.which.toString();\n",
       "\n",
       "    this._key_event_extra(event, name);\n",
       "\n",
       "    this.send_message(name, {key: value,\n",
       "                             guiEvent: simpleKeys(event)});\n",
       "    return false;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.toolbar_button_onclick = function(name) {\n",
       "    if (name == 'download') {\n",
       "        this.handle_save(this, null);\n",
       "    } else {\n",
       "        this.send_message(\"toolbar_button\", {name: name});\n",
       "    }\n",
       "};\n",
       "\n",
       "mpl.figure.prototype.toolbar_button_onmouseover = function(tooltip) {\n",
       "    this.message.textContent = tooltip;\n",
       "};\n",
       "mpl.toolbar_items = [[\"Home\", \"Reset original view\", \"fa fa-home icon-home\", \"home\"], [\"Back\", \"Back to previous view\", \"fa fa-arrow-left icon-arrow-left\", \"back\"], [\"Forward\", \"Forward to next view\", \"fa fa-arrow-right icon-arrow-right\", \"forward\"], [\"\", \"\", \"\", \"\"], [\"Pan\", \"Pan axes with left mouse, zoom with right\", \"fa fa-arrows icon-move\", \"pan\"], [\"Zoom\", \"Zoom to rectangle\", \"fa fa-square-o icon-check-empty\", \"zoom\"], [\"\", \"\", \"\", \"\"], [\"Download\", \"Download plot\", \"fa fa-floppy-o icon-save\", \"download\"]];\n",
       "\n",
       "mpl.extensions = [\"eps\", \"jpeg\", \"pdf\", \"png\", \"ps\", \"raw\", \"svg\", \"tif\"];\n",
       "\n",
       "mpl.default_extension = \"png\";var comm_websocket_adapter = function(comm) {\n",
       "    // Create a \"websocket\"-like object which calls the given IPython comm\n",
       "    // object with the appropriate methods. Currently this is a non binary\n",
       "    // socket, so there is still some room for performance tuning.\n",
       "    var ws = {};\n",
       "\n",
       "    ws.close = function() {\n",
       "        comm.close()\n",
       "    };\n",
       "    ws.send = function(m) {\n",
       "        //console.log('sending', m);\n",
       "        comm.send(m);\n",
       "    };\n",
       "    // Register the callback with on_msg.\n",
       "    comm.on_msg(function(msg) {\n",
       "        //console.log('receiving', msg['content']['data'], msg);\n",
       "        // Pass the mpl event to the overridden (by mpl) onmessage function.\n",
       "        ws.onmessage(msg['content']['data'])\n",
       "    });\n",
       "    return ws;\n",
       "}\n",
       "\n",
       "mpl.mpl_figure_comm = function(comm, msg) {\n",
       "    // This is the function which gets called when the mpl process\n",
       "    // starts-up an IPython Comm through the \"matplotlib\" channel.\n",
       "\n",
       "    var id = msg.content.data.id;\n",
       "    // Get hold of the div created by the display call when the Comm\n",
       "    // socket was opened in Python.\n",
       "    var element = $(\"#\" + id);\n",
       "    var ws_proxy = comm_websocket_adapter(comm)\n",
       "\n",
       "    function ondownload(figure, format) {\n",
       "        window.open(figure.imageObj.src);\n",
       "    }\n",
       "\n",
       "    var fig = new mpl.figure(id, ws_proxy,\n",
       "                           ondownload,\n",
       "                           element.get(0));\n",
       "\n",
       "    // Call onopen now - mpl needs it, as it is assuming we've passed it a real\n",
       "    // web socket which is closed, not our websocket->open comm proxy.\n",
       "    ws_proxy.onopen();\n",
       "\n",
       "    fig.parent_element = element.get(0);\n",
       "    fig.cell_info = mpl.find_output_cell(\"<div id='\" + id + \"'></div>\");\n",
       "    if (!fig.cell_info) {\n",
       "        console.error(\"Failed to find cell for figure\", id, fig);\n",
       "        return;\n",
       "    }\n",
       "\n",
       "    var output_index = fig.cell_info[2]\n",
       "    var cell = fig.cell_info[0];\n",
       "\n",
       "};\n",
       "\n",
       "mpl.figure.prototype.handle_close = function(fig, msg) {\n",
       "    var width = fig.canvas.width/mpl.ratio\n",
       "    fig.root.unbind('remove')\n",
       "\n",
       "    // Update the output cell to use the data from the current canvas.\n",
       "    fig.push_to_output();\n",
       "    var dataURL = fig.canvas.toDataURL();\n",
       "    // Re-enable the keyboard manager in IPython - without this line, in FF,\n",
       "    // the notebook keyboard shortcuts fail.\n",
       "    IPython.keyboard_manager.enable()\n",
       "    $(fig.parent_element).html('<img src=\"' + dataURL + '\" width=\"' + width + '\">');\n",
       "    fig.close_ws(fig, msg);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.close_ws = function(fig, msg){\n",
       "    fig.send_message('closing', msg);\n",
       "    // fig.ws.close()\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.push_to_output = function(remove_interactive) {\n",
       "    // Turn the data on the canvas into data in the output cell.\n",
       "    var width = this.canvas.width/mpl.ratio\n",
       "    var dataURL = this.canvas.toDataURL();\n",
       "    this.cell_info[1]['text/html'] = '<img src=\"' + dataURL + '\" width=\"' + width + '\">';\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.updated_canvas_event = function() {\n",
       "    // Tell IPython that the notebook contents must change.\n",
       "    IPython.notebook.set_dirty(true);\n",
       "    this.send_message(\"ack\", {});\n",
       "    var fig = this;\n",
       "    // Wait a second, then push the new image to the DOM so\n",
       "    // that it is saved nicely (might be nice to debounce this).\n",
       "    setTimeout(function () { fig.push_to_output() }, 1000);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_toolbar = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var nav_element = $('<div/>');\n",
       "    nav_element.attr('style', 'width: 100%');\n",
       "    this.root.append(nav_element);\n",
       "\n",
       "    // Define a callback function for later on.\n",
       "    function toolbar_event(event) {\n",
       "        return fig.toolbar_button_onclick(event['data']);\n",
       "    }\n",
       "    function toolbar_mouse_event(event) {\n",
       "        return fig.toolbar_button_onmouseover(event['data']);\n",
       "    }\n",
       "\n",
       "    for(var toolbar_ind in mpl.toolbar_items){\n",
       "        var name = mpl.toolbar_items[toolbar_ind][0];\n",
       "        var tooltip = mpl.toolbar_items[toolbar_ind][1];\n",
       "        var image = mpl.toolbar_items[toolbar_ind][2];\n",
       "        var method_name = mpl.toolbar_items[toolbar_ind][3];\n",
       "\n",
       "        if (!name) { continue; };\n",
       "\n",
       "        var button = $('<button class=\"btn btn-default\" href=\"#\" title=\"' + name + '\"><i class=\"fa ' + image + ' fa-lg\"></i></button>');\n",
       "        button.click(method_name, toolbar_event);\n",
       "        button.mouseover(tooltip, toolbar_mouse_event);\n",
       "        nav_element.append(button);\n",
       "    }\n",
       "\n",
       "    // Add the status bar.\n",
       "    var status_bar = $('<span class=\"mpl-message\" style=\"text-align:right; float: right;\"/>');\n",
       "    nav_element.append(status_bar);\n",
       "    this.message = status_bar[0];\n",
       "\n",
       "    // Add the close button to the window.\n",
       "    var buttongrp = $('<div class=\"btn-group inline pull-right\"></div>');\n",
       "    var button = $('<button class=\"btn btn-mini btn-primary\" href=\"#\" title=\"Stop Interaction\"><i class=\"fa fa-power-off icon-remove icon-large\"></i></button>');\n",
       "    button.click(function (evt) { fig.handle_close(fig, {}); } );\n",
       "    button.mouseover('Stop Interaction', toolbar_mouse_event);\n",
       "    buttongrp.append(button);\n",
       "    var titlebar = this.root.find($('.ui-dialog-titlebar'));\n",
       "    titlebar.prepend(buttongrp);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._root_extra_style = function(el){\n",
       "    var fig = this\n",
       "    el.on(\"remove\", function(){\n",
       "\tfig.close_ws(fig, {});\n",
       "    });\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._canvas_extra_style = function(el){\n",
       "    // this is important to make the div 'focusable\n",
       "    el.attr('tabindex', 0)\n",
       "    // reach out to IPython and tell the keyboard manager to turn it's self\n",
       "    // off when our div gets focus\n",
       "\n",
       "    // location in version 3\n",
       "    if (IPython.notebook.keyboard_manager) {\n",
       "        IPython.notebook.keyboard_manager.register_events(el);\n",
       "    }\n",
       "    else {\n",
       "        // location in version 2\n",
       "        IPython.keyboard_manager.register_events(el);\n",
       "    }\n",
       "\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._key_event_extra = function(event, name) {\n",
       "    var manager = IPython.notebook.keyboard_manager;\n",
       "    if (!manager)\n",
       "        manager = IPython.keyboard_manager;\n",
       "\n",
       "    // Check for shift+enter\n",
       "    if (event.shiftKey && event.which == 13) {\n",
       "        this.canvas_div.blur();\n",
       "        event.shiftKey = false;\n",
       "        // Send a \"J\" for go to next cell\n",
       "        event.which = 74;\n",
       "        event.keyCode = 74;\n",
       "        manager.command_mode();\n",
       "        manager.handle_keydown(event);\n",
       "    }\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_save = function(fig, msg) {\n",
       "    fig.ondownload(fig, null);\n",
       "}\n",
       "\n",
       "\n",
       "mpl.find_output_cell = function(html_output) {\n",
       "    // Return the cell and output element which can be found *uniquely* in the notebook.\n",
       "    // Note - this is a bit hacky, but it is done because the \"notebook_saving.Notebook\"\n",
       "    // IPython event is triggered only after the cells have been serialised, which for\n",
       "    // our purposes (turning an active figure into a static one), is too late.\n",
       "    var cells = IPython.notebook.get_cells();\n",
       "    var ncells = cells.length;\n",
       "    for (var i=0; i<ncells; i++) {\n",
       "        var cell = cells[i];\n",
       "        if (cell.cell_type === 'code'){\n",
       "            for (var j=0; j<cell.output_area.outputs.length; j++) {\n",
       "                var data = cell.output_area.outputs[j];\n",
       "                if (data.data) {\n",
       "                    // IPython >= 3 moved mimebundle to data attribute of output\n",
       "                    data = data.data;\n",
       "                }\n",
       "                if (data['text/html'] == html_output) {\n",
       "                    return [cell, data, j];\n",
       "                }\n",
       "            }\n",
       "        }\n",
       "    }\n",
       "}\n",
       "\n",
       "// Register the function which deals with the matplotlib target/channel.\n",
       "// The kernel may be null if the page has been refreshed.\n",
       "if (IPython.notebook.kernel != null) {\n",
       "    IPython.notebook.kernel.comm_manager.register_target('matplotlib', mpl.mpl_figure_comm);\n",
       "}\n"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" width=\"699.444462973395\">"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start/Continue training from epoch 0\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 7, 5, 3, 7, 5, 2, 6, 2, 5, 5, 2, 1, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 0, 7, 2, 7, 0, 7, 7, 7, 3, 7, 7, 7, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 3, 3, 7, 2, 7, 7, 3, 5, 7, 3, 7, 7, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 7, 5, 3, 3, 3, 3, 7, 6, 7, 7, 3, 3, 3, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 7, 2, 7, 7, 7, 3, 3, 3, 7, 3, 3, 3, 7, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 7, 7, 3, 3, 6, 7, 3, 3, 3, 2, 3, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 3, 3, 3, 5, 3, 3, 7, 3, 6, 6, 3, 3, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 3, 3, 0, 3, 3, 3, 3, 6, 6, 3, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 3, 6, 3, 2, 6, 3, 3, 6, 3, 6, 6, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 3, 3, 6, 3, 3, 5, 6, 5, 0, 6, 6, 4, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 3, 3, 3, 3, 6, 3, 3, 3, 6, 5, 5, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 0, 0, 3, 3, 6, 6, 3, 2, 3, 0, 3, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 3, 0, 6, 6, 5, 3, 3, 3, 6, 6, 3, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 5, 6, 6, 0, 6, 3, 6, 2, 6, 6, 5, 0, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 0, 6, 5, 5, 3, 4, 6, 5, 3, 3, 0, 5, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 5, 6, 3, 2, 3, 0, 0, 3, 5, 3, 0, 5, 0, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 6, 0, 6, 3, 5, 2, 3, 7, 6, 0, 3, 5, 3, 4], device='cuda:0')\n",
      "Epoch 1 (Time: 1.93s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 3, 0, 3, 0, 6, 0, 3, 6, 0, 3, 3, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 4, 7, 3, 0, 3, 3, 6, 3, 5, 0, 5, 2, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 3, 3, 7, 3, 3, 3, 3, 3, 3, 3, 3, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 4, 3, 3, 3, 3, 4, 3, 3, 3, 5, 3, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 3, 3, 3, 6, 4, 4, 0, 7, 3, 5, 3, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 3, 4, 7, 5, 4, 7, 3, 0, 4, 3, 3, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 3, 7, 3, 3, 4, 3, 5, 3, 3, 0, 4, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 4, 7, 5, 3, 4, 3, 0, 4, 3, 3, 3, 4, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 2, 4, 4, 0, 4, 0, 7, 7, 0, 5, 7, 5, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 3, 5, 4, 0, 7, 2, 2, 3, 6, 5, 3, 4, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 5, 6, 7, 7, 3, 0, 4, 0, 3, 7, 7, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 7, 6, 5, 3, 4, 0, 6, 0, 5, 3, 2, 4, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 3, 2, 2, 7, 3, 3, 7, 7, 5, 2, 2, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 2, 3, 5, 7, 6, 2, 6, 7, 4, 5, 0, 5, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 7, 0, 7, 6, 5, 3, 2, 2, 3, 5, 2, 2, 0, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 5, 3, 5, 6, 3, 5, 5, 3, 0, 0, 2, 3, 7, 0, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 3, 3, 6, 3, 7, 3, 0, 6, 3, 5, 0, 7, 5], device='cuda:0')\n",
      "Epoch 2 (Time: 1.28s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 6, 3, 2, 6, 3, 3, 3, 0, 2, 5, 0, 7, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 5, 6, 0, 3, 6, 6, 3, 3, 7, 6, 5, 4, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 5, 5, 0, 3, 0, 3, 3, 0, 5, 3, 2, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 0, 7, 3, 3, 5, 6, 7, 2, 0, 0, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 6, 5, 3, 6, 3, 2, 3, 6, 3, 3, 0, 5, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 0, 3, 0, 3, 5, 5, 5, 3, 3, 3, 6, 6, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 6, 3, 3, 0, 3, 3, 0, 6, 6, 6, 4, 3, 0, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 3, 0, 0, 6, 6, 0, 3, 3, 0, 1, 6, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 0, 0, 3, 5, 3, 5, 4, 3, 7, 2, 2, 3, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 4, 0, 3, 3, 4, 6, 4, 3, 4, 0, 4, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 4, 7, 3, 3, 3, 3, 5, 0, 3, 5, 5, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 4, 0, 4, 3, 4, 5, 2, 3, 5, 7, 0, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 5, 0, 7, 2, 0, 3, 3, 0, 3, 6, 6, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 3, 5, 3, 1, 4, 2, 0, 4, 3, 0, 3, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 5, 3, 4, 3, 4, 4, 3, 0, 3, 2, 3, 4, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 4, 3, 3, 5, 3, 3, 3, 7, 2, 5, 5, 3, 5, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 4, 2, 3, 4, 3, 2, 3, 3, 5, 3, 7, 3, 4, 6], device='cuda:0')\n",
      "Epoch 3 (Time: 1.55s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 0, 3, 5, 3, 7, 3, 3, 6, 4, 4, 7, 5, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 6, 6, 5, 3, 3, 2, 0, 4, 4, 3, 3, 3, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 3, 4, 3, 3, 5, 3, 3, 3, 7, 6, 2, 2, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 4, 6, 5, 0, 3, 7, 3, 6, 3, 6, 3, 5, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 2, 2, 7, 5, 5, 7, 3, 7, 3, 0, 3, 0, 6, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 6, 3, 5, 0, 3, 3, 3, 7, 3, 2, 7, 4, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 3, 3, 7, 6, 3, 3, 3, 3, 7, 0, 5, 5, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 6, 7, 3, 3, 6, 2, 3, 6, 7, 2, 0, 5, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 6, 2, 3, 3, 7, 3, 7, 3, 3, 7, 2, 3, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 0, 7, 5, 7, 6, 0, 3, 7, 0, 3, 1, 2, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 7, 1, 0, 6, 2, 5, 0, 7, 3, 2, 5, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 3, 3, 7, 6, 4, 3, 3, 2, 0, 2, 5, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 6, 7, 4, 5, 6, 3, 6, 0, 6, 3, 6, 3, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 5, 2, 6, 6, 2, 6, 2, 4, 1, 5, 4, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 3, 6, 2, 1, 6, 4, 3, 3, 0, 4, 3, 6, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 2, 2, 6, 7, 2, 5, 0, 6, 6, 0, 4, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 5, 7, 6, 3, 5, 3, 6, 7, 5, 3, 3, 6, 3], device='cuda:0')\n",
      "Epoch 4 (Time: 1.57s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 3, 3, 0, 6, 5, 7, 0, 6, 3, 2, 3, 4, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 0, 0, 6, 2, 0, 6, 0, 3, 2, 5, 3, 2, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 1, 6, 6, 0, 0, 0, 5, 3, 5, 5, 5, 6, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 0, 5, 3, 6, 6, 6, 7, 1, 2, 0, 3, 7, 5, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 5, 3, 0, 5, 5, 0, 6, 4, 3, 0, 7, 0, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 4, 0, 5, 2, 0, 6, 0, 3, 4, 3, 3, 3, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 0, 7, 4, 2, 5, 5, 4, 6, 6, 3, 3, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 4, 6, 2, 2, 5, 4, 0, 5, 4, 6, 6, 2, 5, 0, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 4, 7, 4, 3, 6, 3, 4, 3, 3, 0, 0, 4, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 5, 7, 6, 3, 5, 3, 4, 6, 6, 0, 3, 6, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 6, 3, 7, 4, 3, 5, 3, 7, 7, 0, 0, 7, 4, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 6, 6, 6, 3, 5, 7, 3, 3, 3, 5, 4, 2, 4, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 2, 3, 5, 0, 0, 5, 6, 5, 3, 6, 3, 7, 1, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 1, 5, 0, 4, 2, 3, 5, 7, 6, 4, 7, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 3, 3, 3, 4, 3, 2, 3, 7, 0, 7, 0, 5, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 2, 3, 6, 5, 4, 0, 4, 0, 1, 7, 3, 7, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 2, 3, 2, 3, 3, 5, 7, 6, 6, 0, 2, 5, 3], device='cuda:0')\n",
      "Epoch 5 (Time: 1.44s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 4, 4, 3, 3, 2, 4, 3, 3, 3, 7, 0, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 4, 5, 6, 0, 0, 5, 7, 3, 6, 5, 7, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 5, 5, 3, 0, 5, 3, 3, 7, 6, 1, 4, 3, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 5, 4, 3, 7, 3, 0, 3, 4, 7, 7, 6, 4, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 3, 6, 7, 3, 2, 7, 0, 1, 3, 5, 5, 0, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 4, 3, 5, 5, 3, 5, 6, 6, 3, 0, 6, 0, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 2, 7, 7, 5, 4, 0, 3, 2, 7, 7, 5, 5, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 6, 3, 0, 5, 3, 6, 6, 6, 6, 5, 0, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 4, 5, 3, 7, 3, 5, 6, 6, 6, 3, 5, 3, 7, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 3, 0, 6, 5, 3, 0, 6, 7, 6, 6, 1, 6, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 0, 6, 3, 3, 6, 7, 5, 6, 7, 4, 6, 6, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 0, 3, 5, 0, 3, 0, 4, 2, 3, 5, 3, 5, 3, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 6, 7, 3, 3, 6, 6, 6, 3, 3, 3, 6, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 6, 3, 3, 3, 5, 4, 6, 3, 3, 4, 6, 4, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 6, 5, 5, 2, 2, 0, 0, 3, 6, 3, 6, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 7, 3, 4, 3, 3, 6, 6, 3, 1, 5, 0, 0, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 7, 6, 2, 2, 6, 7, 6, 3, 3, 3, 5, 6, 3, 7], device='cuda:0')\n",
      "Epoch 6 (Time: 1.49s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 3, 6, 4, 3, 4, 3, 0, 3, 3, 3, 0, 5, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 4, 7, 6, 3, 3, 5, 7, 3, 3, 0, 3, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 3, 7, 3, 5, 2, 5, 7, 3, 7, 3, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 5, 5, 0, 2, 4, 3, 3, 3, 6, 0, 6, 4, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 0, 3, 5, 5, 6, 6, 7, 0, 7, 0, 3, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 5, 3, 7, 2, 4, 5, 2, 0, 2, 5, 6, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 5, 5, 3, 4, 4, 3, 6, 1, 3, 4, 6, 2, 0, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 7, 0, 2, 7, 7, 3, 4, 3, 3, 2, 4, 1, 0, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 2, 6, 4, 2, 6, 3, 3, 4, 0, 6, 4, 6, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 4, 5, 6, 3, 0, 6, 7, 7, 0, 2, 1, 2, 1, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 1, 2, 2, 7, 3, 3, 6, 0, 2, 7, 6, 5, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 0, 6, 6, 5, 6, 1, 2, 0, 2, 0, 5, 4, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 6, 7, 3, 4, 4, 3, 5, 5, 3, 3, 6, 2, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 1, 6, 2, 6, 0, 3, 2, 7, 5, 2, 0, 0, 4, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 1, 3, 5, 2, 7, 4, 0, 0, 3, 7, 7, 2, 2, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 4, 4, 0, 5, 3, 7, 3, 0, 2, 4, 6, 5, 4, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 3, 5, 3, 7, 0, 3, 6, 6, 4, 6, 2, 5, 6, 1], device='cuda:0')\n",
      "Epoch 7 (Time: 1.61s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 5, 3, 6, 3, 7, 3, 4, 0, 2, 4, 0, 3, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 7, 4, 3, 3, 5, 0, 3, 3, 5, 2, 0, 3, 5, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 5, 5, 5, 0, 3, 6, 6, 3, 3, 6, 7, 4, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 6, 5, 7, 4, 1, 3, 6, 6, 3, 3, 5, 3, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 0, 6, 3, 6, 3, 0, 6, 6, 2, 6, 7, 5, 6, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 0, 3, 5, 6, 3, 6, 3, 6, 3, 0, 0, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 4, 4, 6, 3, 3, 3, 3, 7, 3, 5, 6, 0, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 0, 0, 3, 3, 2, 3, 5, 0, 5, 0, 4, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 6, 4, 3, 1, 7, 1, 3, 4, 4, 3, 0, 1, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 0, 3, 3, 3, 3, 3, 6, 6, 3, 6, 0, 6, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 7, 5, 3, 2, 1, 6, 2, 4, 1, 7, 5, 4, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 0, 5, 0, 5, 4, 5, 3, 7, 2, 7, 3, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 5, 3, 5, 2, 0, 0, 3, 2, 5, 0, 3, 3, 6, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 6, 2, 3, 5, 2, 3, 4, 2, 5, 7, 6, 3, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 3, 3, 0, 3, 0, 6, 3, 4, 6, 3, 3, 6, 5, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 2, 3, 6, 0, 3, 5, 4, 6, 7, 4, 7, 5, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 6, 5, 2, 4, 7, 5, 4, 7, 0, 7, 7, 3, 3, 3], device='cuda:0')\n",
      "Epoch 8 (Time: 1.52s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 4, 3, 3, 2, 6, 4, 3, 6, 5, 6, 6, 3, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 0, 1, 4, 7, 2, 3, 4, 1, 3, 5, 5, 1, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 6, 0, 6, 3, 3, 0, 6, 0, 3, 4, 4, 6, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 1, 3, 4, 5, 5, 2, 3, 3, 4, 7, 3, 2, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 2, 1, 6, 6, 2, 3, 4, 6, 4, 6, 7, 0, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 2, 3, 3, 2, 7, 0, 7, 5, 3, 2, 2, 2, 3, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 4, 7, 5, 3, 5, 4, 6, 2, 6, 3, 3, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 2, 4, 2, 5, 2, 7, 7, 3, 2, 6, 3, 2, 0, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 5, 7, 4, 6, 4, 5, 5, 2, 5, 6, 2, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 6, 3, 2, 0, 5, 6, 0, 2, 2, 7, 4, 7, 2, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 1, 6, 6, 3, 7, 5, 4, 0, 6, 3, 3, 5, 4, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 0, 6, 2, 3, 2, 3, 0, 3, 3, 3, 7, 2, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 6, 6, 3, 6, 5, 1, 4, 3, 4, 7, 6, 3, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 7, 7, 3, 6, 3, 1, 3, 3, 0, 2, 0, 7, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 1, 0, 4, 7, 6, 0, 3, 0, 4, 7, 1, 1, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 2, 7, 5, 3, 4, 7, 6, 3, 3, 0, 3, 6, 2, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 3, 3, 5, 4, 0, 4, 3, 4, 7, 3, 5, 4, 7, 7], device='cuda:0')\n",
      "Epoch 9 (Time: 1.43s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 2, 7, 3, 3, 3, 3, 2, 3, 3, 5, 3, 4, 3, 4, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 2, 7, 4, 7, 3, 7, 3, 4, 0, 3, 3, 2, 0, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 4, 4, 3, 3, 0, 5, 7, 2, 4, 3, 6, 5, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 6, 3, 3, 4, 5, 1, 3, 3, 3, 6, 6, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 3, 1, 6, 6, 5, 6, 3, 4, 3, 3, 3, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 5, 5, 3, 0, 3, 7, 6, 5, 4, 1, 7, 0, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 0, 3, 3, 3, 1, 5, 3, 0, 7, 6, 5, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 5, 6, 4, 0, 6, 0, 3, 6, 0, 3, 4, 6, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 7, 2, 4, 4, 6, 0, 3, 7, 0, 5, 3, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 5, 5, 5, 6, 0, 7, 0, 6, 0, 4, 3, 3, 6, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 2, 7, 6, 7, 0, 3, 3, 4, 7, 3, 6, 3, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 5, 3, 5, 0, 3, 0, 5, 3, 3, 7, 6, 6, 3, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 0, 0, 6, 6, 3, 6, 3, 0, 3, 4, 0, 3, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 3, 5, 3, 3, 3, 5, 5, 6, 6, 4, 5, 0, 3, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 4, 3, 0, 7, 7, 5, 2, 7, 2, 7, 3, 4, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 1, 5, 6, 6, 5, 6, 3, 0, 5, 7, 3, 5, 3, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 5, 3, 3, 5, 5, 6, 5, 6, 0, 5, 0, 0, 3, 4], device='cuda:0')\n",
      "Epoch 10 (Time: 1.49s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 6, 6, 6, 5, 6, 3, 3, 0, 6, 3, 4, 3, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 6, 6, 3, 2, 7, 5, 3, 7, 2, 3, 3, 0, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 6, 7, 6, 6, 7, 7, 3, 6, 3, 6, 7, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 6, 2, 3, 3, 6, 4, 0, 3, 6, 4, 6, 1, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 0, 7, 6, 6, 3, 4, 6, 3, 0, 3, 2, 3, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 6, 5, 3, 7, 5, 4, 5, 2, 4, 3, 5, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 7, 0, 2, 1, 3, 6, 4, 7, 3, 6, 4, 0, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 4, 3, 2, 4, 0, 3, 3, 0, 7, 5, 3, 6, 3, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 0, 3, 7, 7, 4, 3, 4, 7, 6, 3, 5, 0, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 5, 7, 3, 5, 3, 0, 1, 7, 7, 2, 3, 6, 7, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 0, 7, 2, 3, 7, 3, 6, 5, 3, 0, 3, 3, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 5, 5, 6, 7, 4, 3, 7, 2, 4, 3, 3, 4, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 5, 1, 0, 3, 3, 7, 1, 2, 4, 3, 3, 3, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 1, 5, 7, 6, 0, 0, 7, 2, 5, 2, 4, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 7, 5, 6, 4, 0, 7, 6, 4, 0, 2, 4, 5, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 2, 5, 0, 7, 2, 3, 4, 3, 6, 6, 4, 7, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 3, 5, 6, 3, 6, 4, 7, 1, 3, 6, 0, 5, 2, 7], device='cuda:0')\n",
      "Epoch 11 (Time: 1.54s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 7, 0, 3, 6, 5, 3, 3, 5, 0, 3, 3, 2, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 7, 0, 0, 5, 6, 2, 2, 4, 4, 5, 7, 3, 0, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 1, 7, 5, 6, 3, 3, 6, 1, 2, 3, 2, 4, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 4, 2, 4, 3, 4, 4, 3, 6, 7, 3, 6, 1, 2, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 7, 0, 5, 0, 3, 3, 6, 3, 3, 7, 6, 4, 4, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 2, 2, 2, 3, 2, 5, 7, 0, 2, 7, 3, 1, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 7, 4, 6, 3, 5, 1, 4, 2, 3, 0, 4, 3, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 5, 5, 7, 6, 3, 2, 6, 2, 7, 0, 6, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 7, 6, 3, 5, 5, 3, 5, 3, 4, 7, 6, 5, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 4, 3, 7, 0, 6, 0, 3, 0, 2, 6, 5, 2, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 2, 3, 0, 3, 4, 6, 7, 0, 6, 4, 6, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 3, 3, 0, 5, 4, 3, 7, 3, 3, 4, 5, 0, 5, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 6, 4, 2, 0, 0, 5, 3, 7, 4, 5, 4, 3, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 6, 6, 5, 6, 5, 3, 2, 4, 3, 0, 1, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 6, 4, 7, 7, 3, 0, 5, 3, 6, 2, 4, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 5, 0, 0, 3, 3, 7, 5, 3, 3, 4, 7, 1, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 3, 5, 0, 4, 6, 6, 7, 5, 6, 3, 2, 6, 2, 6], device='cuda:0')\n",
      "Epoch 12 (Time: 1.53s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 4, 0, 3, 3, 7, 6, 6, 0, 0, 6, 6, 3, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 4, 3, 7, 7, 3, 6, 3, 6, 3, 2, 3, 3, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 0, 0, 5, 0, 6, 4, 4, 4, 1, 6, 7, 2, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 0, 6, 7, 0, 7, 3, 6, 6, 0, 7, 3, 6, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 3, 4, 3, 2, 6, 7, 4, 0, 2, 7, 7, 3, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 3, 6, 3, 5, 3, 3, 1, 6, 6, 3, 4, 3, 2, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 4, 6, 3, 3, 6, 5, 3, 0, 3, 1, 6, 0, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 3, 0, 1, 3, 4, 5, 4, 5, 3, 3, 7, 6, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 5, 3, 2, 3, 3, 0, 4, 7, 3, 3, 2, 4, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 2, 1, 3, 5, 6, 0, 3, 6, 4, 0, 3, 7, 5, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 4, 6, 3, 6, 7, 3, 7, 4, 5, 5, 6, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 6, 3, 3, 7, 4, 5, 0, 3, 3, 5, 3, 3, 1, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 5, 4, 3, 0, 0, 6, 3, 3, 3, 2, 6, 7, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 1, 6, 5, 6, 6, 0, 2, 3, 3, 2, 6, 0, 0, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 4, 7, 3, 3, 4, 6, 7, 5, 7, 0, 3, 7, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 5, 5, 7, 0, 7, 0, 6, 6, 3, 4, 4, 2, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 3, 3, 7, 3, 5, 3, 6, 7, 7, 7, 3, 2, 2, 6], device='cuda:0')\n",
      "Epoch 13 (Time: 1.43s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 4, 2, 2, 6, 6, 6, 3, 7, 3, 4, 3, 0, 6, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 3, 3, 6, 3, 7, 1, 3, 0, 2, 6, 3, 4, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 3, 0, 1, 2, 6, 7, 3, 7, 4, 0, 3, 3, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 1, 5, 2, 0, 3, 7, 3, 2, 4, 4, 3, 5, 3, 2, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 3, 1, 6, 6, 0, 5, 6, 3, 0, 4, 4, 3, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 3, 2, 4, 6, 1, 6, 2, 6, 0, 3, 2, 2, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 6, 7, 3, 5, 3, 3, 3, 3, 2, 7, 7, 4, 3, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 5, 6, 7, 2, 6, 4, 2, 5, 0, 3, 4, 4, 2, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 4, 1, 4, 3, 5, 7, 3, 7, 3, 5, 3, 5, 0, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 5, 3, 6, 5, 3, 3, 5, 3, 6, 5, 2, 6, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 7, 3, 5, 6, 4, 5, 2, 4, 2, 5, 0, 7, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 3, 0, 6, 3, 2, 2, 7, 1, 4, 7, 5, 3, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 3, 6, 3, 5, 2, 6, 6, 6, 0, 0, 2, 0, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 2, 6, 6, 5, 4, 3, 5, 4, 7, 3, 0, 7, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 7, 2, 0, 3, 7, 4, 5, 3, 4, 5, 3, 7, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 1, 5, 0, 5, 6, 7, 2, 7, 3, 4, 0, 3, 6, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 2, 1, 4, 3, 4, 7, 3, 0, 5, 7, 5, 3, 6, 6, 3], device='cuda:0')\n",
      "Epoch 14 (Time: 1.56s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 3, 2, 5, 3, 6, 3, 0, 6, 1, 3, 2, 3, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 1, 2, 4, 7, 0, 2, 3, 3, 5, 6, 0, 6, 3, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 1, 3, 4, 6, 0, 2, 3, 5, 3, 0, 3, 0, 3, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 5, 3, 3, 5, 7, 3, 3, 3, 6, 7, 3, 3, 1, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 2, 0, 4, 3, 7, 5, 3, 3, 7, 4, 3, 5, 4, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 7, 0, 6, 4, 4, 3, 1, 2, 7, 2, 0, 3, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 0, 3, 5, 6, 7, 0, 0, 7, 0, 3, 1, 0, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 5, 7, 7, 5, 5, 3, 7, 4, 2, 7, 4, 1, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 0, 2, 3, 3, 4, 3, 7, 3, 4, 4, 0, 2, 4, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 6, 3, 4, 3, 6, 3, 5, 3, 3, 5, 4, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 2, 6, 5, 3, 5, 2, 3, 6, 7, 2, 3, 3, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 0, 3, 3, 3, 6, 0, 2, 7, 6, 4, 7, 4, 6, 1, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 0, 5, 3, 5, 6, 4, 3, 3, 3, 6, 6, 5, 5, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 3, 0, 4, 6, 0, 0, 2, 4, 3, 6, 6, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 3, 0, 0, 7, 7, 3, 0, 5, 3, 7, 3, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 5, 3, 3, 4, 7, 0, 6, 6, 3, 5, 6, 0, 4, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 4, 7, 4, 5, 3, 7, 7, 6, 3, 6, 2, 6, 3, 5], device='cuda:0')\n",
      "Epoch 15 (Time: 1.45s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 0, 2, 6, 2, 6, 6, 1, 0, 5, 3, 0, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 3, 5, 0, 7, 4, 6, 0, 5, 3, 7, 2, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 6, 5, 4, 7, 3, 7, 3, 3, 1, 0, 7, 2, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 3, 6, 3, 3, 1, 3, 0, 0, 3, 3, 6, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 0, 3, 2, 7, 2, 3, 6, 3, 6, 3, 1, 5, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 3, 7, 0, 0, 2, 4, 7, 0, 7, 3, 7, 5, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 3, 3, 2, 3, 4, 3, 3, 2, 3, 7, 3, 3, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 5, 7, 5, 4, 0, 3, 3, 4, 3, 6, 7, 7, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 3, 2, 6, 4, 6, 0, 3, 4, 6, 1, 6, 7, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 2, 4, 5, 2, 3, 0, 5, 3, 3, 3, 6, 3, 4, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 6, 6, 3, 5, 7, 2, 1, 5, 3, 5, 3, 4, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 1, 2, 7, 0, 4, 4, 3, 6, 0, 5, 2, 7, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 4, 2, 6, 7, 4, 2, 6, 2, 1, 1, 4, 2, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 4, 6, 3, 4, 3, 3, 5, 0, 6, 4, 3, 5, 4, 2, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 5, 5, 6, 6, 3, 0, 6, 2, 0, 4, 3, 6, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 4, 6, 5, 6, 6, 5, 3, 3, 0, 0, 0, 5, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 2, 3, 0, 6, 7, 2, 0, 5, 5, 3, 0, 7, 3, 3], device='cuda:0')\n",
      "Epoch 16 (Time: 1.48s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 5, 3, 7, 0, 5, 3, 7, 6, 5, 6, 0, 0, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 4, 7, 6, 1, 0, 5, 6, 7, 3, 7, 6, 3, 0, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 7, 2, 5, 5, 2, 5, 4, 0, 3, 3, 4, 4, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 0, 0, 6, 5, 3, 3, 7, 5, 4, 5, 3, 0, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 1, 2, 4, 6, 0, 2, 3, 3, 3, 0, 2, 6, 4, 4, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 3, 5, 6, 7, 7, 6, 7, 5, 1, 2, 3, 2, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 6, 6, 2, 0, 7, 6, 3, 5, 4, 3, 7, 6, 7, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 5, 4, 4, 5, 1, 4, 2, 2, 0, 6, 3, 4, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 2, 5, 6, 7, 6, 7, 4, 5, 6, 3, 5, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 6, 6, 7, 5, 4, 5, 6, 3, 6, 2, 4, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 6, 5, 6, 3, 3, 5, 5, 5, 7, 6, 3, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 7, 6, 3, 5, 2, 5, 7, 3, 3, 2, 2, 3, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 3, 3, 7, 5, 5, 4, 5, 6, 7, 4, 4, 5, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 0, 6, 3, 6, 3, 3, 3, 3, 5, 6, 2, 0, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 1, 2, 3, 0, 3, 3, 1, 0, 5, 1, 5, 2, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 4, 3, 7, 6, 4, 7, 6, 3, 5, 1, 3, 3, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 0, 0, 2, 3, 7, 3, 2, 3, 2, 5, 4, 1, 0, 6], device='cuda:0')\n",
      "Epoch 17 (Time: 1.52s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 7, 7, 5, 3, 6, 6, 0, 1, 3, 5, 2, 1, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 2, 4, 3, 3, 4, 6, 3, 5, 5, 5, 1, 7, 0, 0, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 4, 2, 5, 2, 0, 6, 3, 2, 4, 3, 0, 4, 7, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 2, 3, 4, 6, 6, 0, 3, 3, 3, 2, 3, 5, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 2, 4, 4, 5, 4, 4, 3, 7, 3, 2, 3, 6, 5, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 5, 5, 3, 3, 3, 4, 3, 6, 3, 3, 0, 3, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 7, 6, 3, 5, 0, 3, 4, 0, 7, 7, 1, 2, 2, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 0, 5, 3, 0, 0, 3, 7, 3, 6, 7, 3, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 3, 7, 3, 5, 0, 3, 7, 7, 0, 0, 6, 7, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 3, 3, 3, 3, 5, 7, 4, 3, 6, 6, 0, 2, 3, 4, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 5, 6, 3, 3, 0, 4, 5, 7, 3, 7, 4, 3, 2, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 5, 4, 2, 3, 5, 3, 4, 0, 2, 7, 0, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 5, 2, 2, 3, 3, 3, 2, 6, 7, 7, 3, 4, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 7, 3, 7, 5, 2, 0, 4, 5, 7, 7, 7, 2, 0, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 3, 5, 1, 0, 4, 3, 0, 4, 7, 3, 5, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 3, 3, 6, 6, 3, 5, 6, 2, 0, 2, 7, 7, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 4, 4, 7, 6, 3, 2, 6, 2, 0, 6, 0, 5, 6, 3, 7], device='cuda:0')\n",
      "Epoch 18 (Time: 1.43s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 5, 7, 5, 6, 5, 2, 3, 2, 5, 2, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 7, 2, 3, 1, 6, 2, 0, 3, 6, 0, 3, 5, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 7, 6, 3, 2, 4, 3, 5, 7, 3, 3, 3, 4, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 0, 2, 2, 3, 3, 5, 7, 5, 3, 5, 3, 1, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 1, 7, 6, 5, 6, 6, 6, 6, 7, 5, 1, 7, 6, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 3, 7, 6, 3, 3, 5, 7, 3, 5, 6, 1, 4, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 6, 0, 7, 0, 4, 0, 3, 7, 6, 5, 5, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 7, 3, 2, 1, 6, 0, 7, 7, 7, 3, 3, 3, 4, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 0, 3, 3, 5, 6, 5, 6, 3, 5, 3, 7, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 4, 7, 5, 6, 7, 0, 6, 3, 3, 7, 3, 2, 7, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 5, 3, 2, 5, 3, 4, 4, 0, 7, 0, 3, 6, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 1, 3, 0, 3, 4, 3, 7, 4, 3, 5, 6, 7, 2, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 5, 7, 6, 4, 3, 0, 6, 3, 5, 3, 4, 3, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 2, 3, 7, 6, 6, 3, 3, 3, 4, 3, 5, 3, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 3, 4, 0, 2, 4, 3, 1, 6, 2, 5, 6, 3, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 3, 0, 0, 6, 2, 3, 1, 6, 4, 6, 0, 6, 5, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 6, 3, 7, 5, 2, 3, 2, 3, 0, 4, 5, 5, 0], device='cuda:0')\n",
      "Epoch 19 (Time: 1.44s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 7, 5, 3, 4, 3, 1, 2, 2, 3, 2, 5, 4, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 6, 3, 6, 5, 7, 3, 4, 0, 6, 5, 7, 0, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 4, 4, 5, 4, 6, 4, 6, 6, 2, 4, 6, 1, 5, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 6, 2, 3, 3, 3, 2, 5, 0, 4, 1, 7, 3, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 5, 6, 3, 3, 3, 3, 3, 3, 0, 7, 5, 4, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 3, 4, 2, 6, 4, 6, 7, 5, 6, 7, 3, 6, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 2, 0, 5, 2, 3, 0, 2, 4, 3, 5, 6, 4, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 4, 3, 2, 2, 0, 2, 7, 7, 3, 0, 2, 6, 3, 2, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 7, 6, 5, 5, 3, 2, 1, 3, 0, 0, 0, 4, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 7, 0, 3, 6, 3, 1, 2, 1, 4, 6, 2, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 2, 6, 6, 2, 4, 6, 6, 6, 0, 2, 5, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 2, 4, 3, 3, 6, 0, 0, 6, 5, 6, 2, 4, 1, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 2, 3, 0, 3, 5, 6, 4, 3, 3, 5, 6, 5, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 4, 1, 7, 3, 4, 7, 3, 4, 2, 0, 4, 0, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 7, 4, 3, 3, 3, 7, 7, 5, 0, 4, 2, 5, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 5, 3, 5, 4, 3, 7, 7, 3, 6, 0, 6, 0, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 5, 7, 0, 7, 6, 3, 7, 2, 5, 3, 4, 3, 3], device='cuda:0')\n",
      "Epoch 20 (Time: 1.43s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 4, 0, 2, 3, 5, 3, 4, 6, 7, 4, 2, 3, 2, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 5, 7, 3, 6, 2, 2, 4, 6, 7, 2, 6, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 5, 7, 3, 6, 4, 7, 3, 2, 6, 4, 4, 6, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 0, 3, 7, 6, 3, 5, 7, 5, 5, 4, 3, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 7, 0, 5, 3, 7, 4, 4, 7, 3, 3, 3, 5, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 3, 2, 3, 6, 6, 3, 2, 6, 6, 2, 0, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 3, 7, 1, 3, 5, 6, 5, 7, 5, 3, 1, 3, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 0, 1, 3, 3, 0, 4, 3, 5, 3, 5, 2, 2, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 7, 7, 6, 6, 1, 3, 4, 5, 0, 7, 3, 4, 5, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 0, 0, 6, 3, 3, 4, 3, 6, 3, 6, 7, 3, 3, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 1, 5, 4, 6, 3, 3, 3, 3, 6, 3, 0, 7, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 7, 3, 5, 3, 3, 6, 6, 3, 3, 5, 3, 0, 2, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 7, 3, 3, 3, 3, 3, 4, 2, 4, 3, 1, 5, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 7, 5, 6, 6, 6, 6, 7, 1, 5, 3, 3, 6, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 3, 7, 2, 3, 1, 4, 0, 3, 0, 0, 0, 2, 0, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 5, 4, 7, 2, 1, 5, 7, 4, 5, 3, 4, 5, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 6, 5, 0, 7, 3, 4, 0, 0, 6, 2, 5, 3, 6, 0], device='cuda:0')\n",
      "Epoch 21 (Time: 1.40s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 3, 7, 3, 3, 1, 4, 1, 6, 5, 2, 2, 5, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 2, 0, 3, 5, 0, 6, 6, 3, 6, 6, 5, 5, 3, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 2, 7, 5, 6, 0, 6, 7, 0, 5, 4, 7, 0, 5, 1, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 3, 5, 4, 3, 0, 6, 4, 0, 3, 6, 3, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 3, 5, 4, 0, 5, 2, 5, 3, 2, 1, 6, 0, 2, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 3, 5, 0, 0, 2, 4, 6, 3, 2, 0, 6, 2, 0, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 4, 4, 5, 0, 2, 3, 3, 3, 4, 2, 6, 3, 0, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 2, 5, 5, 7, 5, 3, 4, 6, 3, 4, 0, 3, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 0, 3, 4, 2, 5, 6, 6, 6, 2, 4, 2, 6, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 5, 4, 4, 3, 3, 3, 6, 4, 4, 6, 3, 7, 1, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 7, 4, 0, 2, 3, 1, 0, 4, 3, 3, 3, 7, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 5, 3, 6, 7, 5, 7, 5, 3, 0, 5, 4, 6, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 6, 2, 6, 7, 3, 3, 5, 0, 3, 4, 4, 0, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 3, 7, 6, 7, 6, 6, 0, 5, 4, 0, 5, 6, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 6, 3, 5, 5, 5, 3, 6, 7, 4, 0, 3, 3, 7, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 6, 4, 1, 3, 5, 5, 0, 2, 6, 3, 2, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 1, 2, 2, 6, 4, 4, 4, 5, 3, 3, 4, 3, 3], device='cuda:0')\n",
      "Epoch 22 (Time: 1.51s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 2, 3, 3, 3, 3, 3, 0, 3, 0, 3, 7, 3, 6, 0, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 3, 6, 6, 2, 2, 0, 0, 5, 3, 5, 6, 7, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 2, 6, 3, 0, 7, 5, 4, 2, 0, 7, 3, 2, 2, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 7, 5, 3, 3, 2, 3, 0, 2, 6, 0, 4, 1, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 5, 3, 6, 3, 1, 7, 4, 1, 3, 7, 3, 5, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 6, 2, 3, 6, 2, 3, 6, 7, 5, 6, 3, 6, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 6, 3, 3, 7, 3, 4, 7, 6, 0, 0, 3, 6, 2, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 0, 3, 4, 3, 5, 4, 2, 2, 4, 3, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 1, 5, 7, 5, 7, 0, 3, 0, 3, 3, 7, 6, 3, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 4, 5, 7, 3, 5, 6, 3, 2, 5, 6, 3, 5, 3, 2, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 3, 3, 3, 6, 2, 3, 3, 3, 3, 3, 0, 7, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 2, 5, 5, 3, 5, 4, 0, 6, 7, 4, 0, 2, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 6, 7, 3, 0, 5, 6, 7, 6, 7, 6, 3, 4, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 1, 6, 0, 3, 3, 7, 3, 5, 2, 3, 3, 3, 1, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 4, 3, 4, 0, 3, 0, 7, 6, 4, 5, 5, 5, 5, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 7, 3, 4, 3, 2, 3, 3, 5, 3, 5, 0, 4, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 5, 7, 2, 1, 5, 6, 5, 3, 5, 6, 4, 3, 2, 4], device='cuda:0')\n",
      "Epoch 23 (Time: 1.61s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 6, 4, 6, 7, 5, 4, 3, 7, 5, 7, 2, 0, 6, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 2, 3, 7, 3, 4, 5, 3, 3, 7, 0, 2, 0, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 3, 5, 7, 7, 0, 0, 3, 0, 6, 0, 4, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 3, 4, 7, 1, 6, 3, 3, 3, 0, 3, 7, 3, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 5, 5, 6, 7, 7, 2, 6, 0, 3, 6, 5, 5, 0, 5, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 3, 4, 5, 0, 4, 0, 6, 3, 3, 3, 7, 1, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 0, 4, 0, 3, 5, 3, 4, 7, 3, 6, 5, 2, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 2, 7, 2, 5, 7, 5, 5, 3, 4, 6, 0, 3, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 4, 5, 7, 2, 3, 7, 7, 6, 6, 2, 0, 4, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 5, 3, 6, 3, 4, 0, 3, 7, 5, 4, 4, 4, 4, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 0, 0, 5, 3, 4, 7, 0, 6, 3, 4, 0, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 6, 1, 4, 2, 3, 4, 4, 3, 3, 2, 0, 3, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 0, 2, 4, 7, 4, 3, 6, 3, 5, 2, 6, 5, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 2, 7, 5, 6, 6, 5, 0, 3, 7, 5, 1, 3, 3, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 2, 1, 2, 4, 5, 2, 3, 4, 3, 0, 7, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 3, 3, 5, 0, 7, 3, 6, 2, 2, 5, 3, 2, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 5, 3, 3, 7, 1, 6, 3, 3, 5, 6, 3, 7, 3, 3], device='cuda:0')\n",
      "Epoch 24 (Time: 1.53s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 2, 7, 0, 3, 3, 7, 0, 7, 6, 2, 7, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 3, 7, 3, 3, 5, 0, 5, 2, 6, 7, 4, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 1, 4, 0, 3, 5, 5, 3, 3, 3, 7, 5, 7, 3, 0, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 4, 3, 7, 0, 1, 7, 5, 0, 4, 7, 5, 7, 7, 2, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 7, 2, 5, 5, 3, 2, 4, 0, 0, 3, 0, 2, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 0, 4, 5, 2, 6, 2, 7, 4, 3, 7, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 1, 7, 6, 6, 4, 6, 6, 1, 4, 3, 2, 3, 7, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 4, 6, 5, 6, 4, 7, 0, 3, 0, 0, 3, 3, 4, 2, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 5, 6, 1, 7, 0, 5, 7, 4, 2, 2, 5, 3, 7, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 2, 3, 6, 1, 3, 5, 6, 3, 4, 0, 3, 2, 3, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 7, 3, 3, 0, 3, 6, 3, 6, 4, 3, 5, 3, 6, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 6, 3, 5, 0, 3, 2, 2, 6, 2, 7, 0, 7, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 1, 3, 0, 5, 3, 7, 5, 0, 4, 3, 5, 7, 6, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 1, 7, 3, 3, 7, 3, 3, 4, 5, 3, 4, 6, 7, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 3, 2, 6, 2, 6, 4, 4, 5, 4, 3, 7, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 2, 3, 6, 4, 3, 7, 3, 0, 3, 6, 5, 3, 3, 0, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 5, 3, 3, 2, 6, 0, 5, 3, 3, 0, 5, 6, 6, 4], device='cuda:0')\n",
      "Epoch 25 (Time: 1.51s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 2, 6, 6, 0, 3, 3, 3, 3, 0, 6, 5, 3, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 7, 3, 4, 6, 7, 5, 5, 3, 0, 3, 3, 0, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 0, 5, 2, 2, 1, 3, 2, 6, 1, 3, 2, 0, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 6, 3, 3, 5, 6, 3, 4, 3, 6, 3, 3, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 7, 7, 6, 6, 2, 5, 5, 4, 4, 3, 7, 0, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 6, 4, 0, 7, 1, 3, 4, 3, 3, 2, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 3, 6, 3, 3, 2, 6, 0, 4, 6, 4, 5, 4, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 0, 6, 4, 3, 5, 5, 5, 3, 3, 6, 7, 7, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 4, 6, 2, 5, 7, 7, 6, 3, 4, 3, 3, 6, 5, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 6, 0, 0, 5, 7, 4, 4, 4, 4, 3, 5, 7, 1, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 0, 2, 6, 6, 3, 3, 3, 4, 3, 2, 5, 4, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 2, 1, 5, 4, 4, 7, 3, 2, 6, 5, 0, 0, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 6, 0, 3, 4, 3, 0, 7, 3, 5, 3, 7, 6, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 6, 6, 6, 3, 3, 3, 5, 0, 3, 5, 5, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 1, 3, 7, 5, 3, 5, 6, 3, 7, 4, 5, 2, 1, 0, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 0, 3, 5, 7, 7, 3, 6, 0, 4, 3, 6, 4, 2, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 5, 1, 7, 0, 5, 5, 0, 3, 3, 0, 3, 4, 6, 0], device='cuda:0')\n",
      "Epoch 26 (Time: 1.50s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 6, 7, 7, 6, 6, 6, 3, 6, 3, 4, 0, 2, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 6, 0, 0, 5, 3, 5, 3, 3, 2, 1, 6, 0, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 2, 0, 0, 3, 3, 7, 5, 6, 5, 7, 2, 7, 2, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 3, 4, 3, 3, 7, 7, 2, 2, 4, 3, 6, 5, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 7, 2, 6, 3, 0, 2, 2, 7, 3, 5, 3, 3, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 6, 3, 5, 6, 3, 6, 5, 2, 3, 3, 4, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 1, 0, 4, 7, 3, 3, 2, 0, 4, 0, 5, 5, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 0, 2, 3, 3, 0, 2, 5, 3, 7, 2, 3, 6, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 1, 0, 1, 2, 7, 5, 6, 4, 3, 7, 3, 2, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 1, 3, 5, 3, 5, 4, 3, 7, 4, 2, 5, 5, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 4, 7, 3, 4, 0, 3, 0, 4, 7, 1, 5, 2, 3, 7, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 4, 4, 3, 2, 5, 3, 3, 7, 4, 3, 3, 6, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 7, 6, 5, 0, 7, 3, 5, 5, 3, 7, 6, 6, 4, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 0, 6, 5, 6, 2, 7, 4, 5, 1, 0, 3, 4, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 4, 3, 4, 0, 4, 6, 2, 3, 6, 3, 3, 6, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 2, 3, 0, 4, 2, 6, 4, 3, 0, 6, 1, 4, 0, 0, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 7, 2, 2, 0, 7, 3, 3, 3, 2, 3, 4, 5, 5, 3], device='cuda:0')\n",
      "Epoch 27 (Time: 1.41s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 2, 3, 3, 3, 0, 1, 0, 4, 3, 3, 5, 3, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 6, 3, 1, 0, 7, 0, 1, 2, 3, 6, 3, 3, 5, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 5, 2, 6, 7, 7, 0, 6, 3, 4, 0, 3, 2, 3, 0, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 7, 5, 7, 7, 3, 0, 4, 3, 6, 3, 2, 3, 5, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 2, 4, 6, 6, 4, 2, 3, 5, 6, 3, 3, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 5, 7, 7, 5, 3, 6, 3, 6, 3, 3, 0, 5, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 4, 6, 3, 3, 5, 3, 2, 7, 2, 7, 5, 3, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 3, 1, 4, 5, 6, 0, 0, 0, 3, 6, 2, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 3, 5, 7, 3, 2, 7, 6, 3, 6, 3, 5, 3, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 2, 0, 3, 3, 4, 0, 0, 5, 0, 0, 4, 0, 6, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 5, 4, 0, 7, 6, 6, 2, 3, 6, 6, 3, 2, 3, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 5, 3, 5, 7, 7, 3, 4, 3, 2, 6, 5, 5, 6, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 7, 0, 4, 4, 3, 6, 3, 3, 7, 2, 3, 7, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 7, 6, 3, 0, 5, 7, 5, 0, 1, 3, 4, 6, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 2, 7, 6, 3, 0, 2, 2, 3, 1, 5, 3, 4, 0, 0, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 0, 0, 3, 3, 0, 3, 2, 4, 5, 7, 4, 7, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 7, 2, 3, 4, 5, 4, 0, 4, 6, 4, 3, 0, 6], device='cuda:0')\n",
      "Epoch 28 (Time: 1.51s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 3, 7, 5, 3, 3, 6, 0, 0, 3, 2, 2, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 0, 4, 5, 5, 3, 2, 3, 1, 7, 6, 6, 6, 5, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 7, 4, 6, 3, 4, 1, 4, 7, 3, 4, 3, 6, 1, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 4, 3, 3, 6, 7, 6, 4, 1, 2, 2, 3, 4, 7, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 6, 3, 5, 2, 3, 3, 6, 0, 0, 5, 7, 0, 0, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 0, 5, 4, 2, 3, 3, 6, 6, 3, 5, 3, 4, 4, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 4, 6, 5, 6, 6, 6, 3, 0, 1, 2, 2, 3, 0, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 0, 3, 7, 3, 3, 6, 2, 6, 2, 3, 4, 3, 4, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 4, 1, 3, 0, 3, 6, 3, 3, 5, 1, 4, 7, 0, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 7, 7, 3, 5, 4, 7, 3, 6, 3, 3, 0, 2, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 0, 7, 3, 5, 2, 0, 3, 5, 5, 5, 4, 7, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 5, 7, 3, 6, 3, 7, 4, 6, 6, 6, 2, 6, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 6, 3, 3, 0, 3, 2, 5, 4, 3, 3, 3, 5, 3, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 2, 3, 6, 4, 7, 3, 3, 3, 0, 3, 5, 3, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 0, 0, 5, 6, 5, 2, 6, 7, 6, 5, 6, 0, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 5, 5, 6, 0, 2, 3, 0, 3, 6, 7, 3, 1, 0, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 3, 6, 6, 4, 1, 3, 3, 3, 0, 4, 3, 3, 3, 6], device='cuda:0')\n",
      "Epoch 29 (Time: 1.48s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 3, 2, 3, 6, 3, 0, 5, 6, 6, 1, 0, 0, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 7, 3, 7, 5, 2, 5, 6, 2, 3, 5, 1, 3, 3, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 3, 6, 1, 3, 7, 0, 5, 6, 3, 3, 5, 3, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 1, 2, 0, 5, 4, 7, 0, 3, 7, 4, 5, 4, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 4, 1, 3, 0, 0, 3, 0, 4, 2, 3, 6, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 5, 4, 3, 3, 7, 3, 0, 3, 6, 7, 6, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 0, 3, 0, 3, 3, 2, 7, 2, 6, 6, 7, 2, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 0, 6, 3, 7, 3, 6, 4, 2, 3, 4, 5, 6, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 3, 3, 3, 3, 6, 0, 2, 6, 2, 4, 3, 3, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 3, 4, 0, 3, 3, 7, 3, 6, 3, 5, 5, 4, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 4, 4, 0, 6, 2, 1, 2, 5, 7, 7, 4, 0, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 0, 3, 3, 4, 0, 6, 0, 5, 5, 7, 7, 2, 5, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 0, 3, 3, 0, 2, 3, 7, 0, 3, 5, 3, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 4, 5, 6, 0, 7, 3, 6, 4, 3, 0, 7, 3, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 1, 3, 4, 3, 3, 3, 0, 0, 5, 2, 5, 7, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 7, 4, 3, 7, 0, 6, 3, 2, 5, 5, 4, 6, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 0, 6, 6, 7, 5, 7, 5, 6, 5, 2, 4, 4, 5], device='cuda:0')\n",
      "Epoch 30 (Time: 1.56s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 5, 5, 6, 4, 0, 7, 2, 0, 3, 4, 7, 6, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 7, 3, 3, 5, 6, 3, 3, 6, 2, 4, 2, 7, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 0, 2, 7, 4, 3, 7, 0, 0, 6, 2, 5, 7, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 1, 3, 3, 2, 3, 6, 7, 3, 4, 6, 3, 3, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 2, 5, 1, 0, 3, 3, 4, 4, 4, 0, 7, 3, 1, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 2, 1, 2, 6, 4, 6, 3, 5, 0, 0, 5, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 3, 5, 6, 0, 4, 4, 3, 6, 3, 7, 7, 4, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 6, 3, 2, 0, 7, 6, 4, 6, 0, 2, 1, 2, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 3, 0, 5, 0, 7, 3, 0, 6, 6, 5, 5, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 4, 0, 0, 3, 4, 0, 6, 2, 5, 0, 5, 3, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 5, 7, 2, 6, 6, 3, 0, 4, 3, 7, 3, 5, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 3, 4, 3, 7, 3, 4, 6, 5, 2, 5, 5, 3, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 0, 5, 6, 4, 6, 0, 3, 3, 4, 1, 1, 5, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 6, 0, 2, 6, 5, 0, 4, 2, 3, 5, 3, 4, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 3, 7, 0, 3, 7, 0, 7, 3, 5, 3, 7, 7, 6, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 6, 4, 5, 3, 5, 3, 3, 4, 6, 0, 0, 2, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 2, 3, 6, 5, 3, 3, 2, 5, 0, 4, 5, 2, 3, 6], device='cuda:0')\n",
      "Epoch 31 (Time: 1.40s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 2, 4, 6, 6, 6, 5, 3, 0, 2, 4, 3, 2, 4, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 1, 0, 0, 2, 1, 3, 2, 5, 4, 3, 2, 6, 7, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 7, 3, 7, 7, 5, 6, 2, 3, 6, 5, 1, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 7, 6, 6, 4, 3, 3, 6, 3, 7, 3, 0, 0, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 3, 0, 0, 5, 5, 3, 2, 4, 3, 2, 1, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 4, 2, 4, 0, 5, 6, 4, 6, 0, 6, 2, 6, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 5, 5, 3, 3, 4, 3, 0, 3, 7, 6, 3, 5, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 5, 4, 0, 5, 3, 7, 7, 0, 6, 2, 0, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 5, 3, 5, 4, 4, 3, 3, 7, 3, 4, 3, 0, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 6, 4, 6, 3, 5, 3, 3, 7, 4, 2, 3, 7, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 7, 3, 5, 0, 7, 7, 2, 0, 3, 4, 3, 6, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 7, 0, 7, 3, 3, 5, 3, 6, 3, 7, 3, 7, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 5, 3, 3, 0, 4, 3, 6, 0, 5, 5, 6, 6, 6, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 7, 0, 3, 1, 7, 3, 4, 6, 3, 4, 0, 2, 0, 1, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 4, 5, 4, 3, 4, 5, 3, 5, 3, 2, 3, 3, 3, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 1, 6, 0, 4, 3, 6, 2, 6, 3, 7, 3, 3, 5, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 6, 3, 6, 3, 7, 3, 4, 3, 0, 0, 0, 0, 5, 7], device='cuda:0')\n",
      "Epoch 32 (Time: 1.45s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 5, 6, 3, 2, 3, 1, 3, 6, 0, 7, 0, 7, 7, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 1, 2, 5, 5, 3, 4, 4, 3, 4, 5, 3, 0, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 1, 2, 3, 5, 6, 4, 1, 5, 3, 2, 0, 1, 5, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 5, 3, 4, 2, 2, 6, 1, 4, 3, 5, 3, 6, 3, 2, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 4, 6, 6, 2, 0, 6, 6, 4, 0, 6, 7, 5, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 4, 0, 4, 6, 0, 7, 0, 3, 0, 5, 0, 0, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 5, 3, 6, 4, 3, 4, 2, 7, 5, 3, 2, 2, 7, 4, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 6, 7, 0, 6, 5, 2, 7, 2, 7, 6, 4, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 3, 5, 0, 6, 5, 7, 0, 7, 4, 3, 3, 5, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 4, 0, 6, 4, 5, 2, 3, 4, 2, 2, 3, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 7, 0, 7, 7, 3, 6, 2, 3, 3, 3, 3, 5, 2, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 2, 0, 2, 2, 6, 1, 6, 3, 6, 5, 3, 3, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 6, 4, 6, 7, 3, 3, 2, 1, 7, 3, 6, 0, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 4, 4, 5, 7, 6, 5, 4, 3, 0, 4, 0, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 6, 2, 6, 3, 0, 7, 3, 5, 6, 3, 1, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 5, 3, 3, 4, 6, 5, 3, 5, 5, 5, 7, 6, 3, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 5, 3, 3, 6, 0, 5, 1, 0, 7, 2, 3, 7, 3], device='cuda:0')\n",
      "Epoch 33 (Time: 1.32s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 3, 6, 3, 3, 7, 0, 7, 4, 3, 3, 5, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 6, 6, 5, 4, 1, 4, 2, 3, 3, 6, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 0, 5, 0, 6, 6, 2, 2, 3, 0, 5, 7, 4, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 6, 3, 5, 4, 2, 2, 4, 6, 3, 0, 4, 3, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 6, 4, 4, 0, 1, 3, 5, 3, 4, 3, 7, 7, 2, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 5, 4, 3, 6, 7, 4, 3, 3, 7, 6, 3, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 4, 3, 0, 6, 0, 5, 1, 7, 6, 3, 7, 4, 3, 0, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 6, 4, 7, 3, 6, 6, 3, 3, 6, 3, 5, 3, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 2, 6, 1, 3, 1, 3, 3, 3, 5, 0, 3, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 0, 2, 5, 4, 3, 0, 6, 7, 2, 3, 5, 0, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 5, 5, 3, 3, 7, 6, 0, 3, 5, 2, 3, 4, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 2, 7, 3, 0, 3, 0, 6, 7, 6, 3, 3, 0, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 3, 3, 3, 5, 0, 7, 6, 3, 6, 7, 4, 2, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 5, 2, 4, 2, 4, 3, 3, 4, 3, 0, 3, 0, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 1, 6, 7, 3, 3, 3, 5, 3, 7, 6, 5, 5, 0, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 6, 5, 3, 4, 6, 3, 7, 3, 3, 4, 7, 6, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 4, 0, 3, 6, 6, 4, 5, 2, 3, 4, 3, 3, 3, 5, 3], device='cuda:0')\n",
      "Epoch 34 (Time: 1.40s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 5, 2, 3, 3, 6, 5, 3, 5, 7, 2, 5, 0, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 6, 5, 4, 3, 4, 7, 0, 2, 6, 3, 0, 3, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 3, 6, 0, 2, 3, 0, 3, 6, 3, 4, 3, 6, 3, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 1, 7, 5, 3, 1, 0, 2, 0, 3, 6, 3, 5, 2, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 1, 4, 3, 3, 4, 0, 2, 4, 0, 3, 2, 3, 5, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 2, 1, 5, 4, 7, 7, 3, 2, 6, 6, 0, 7, 3, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 0, 3, 3, 3, 5, 5, 6, 0, 5, 4, 5, 5, 6, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 3, 3, 6, 3, 5, 5, 7, 0, 3, 4, 3, 0, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 5, 0, 7, 3, 7, 2, 3, 4, 6, 2, 4, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 3, 6, 5, 5, 3, 0, 3, 6, 6, 1, 3, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 4, 2, 1, 5, 3, 3, 4, 6, 3, 3, 3, 7, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 6, 2, 3, 7, 2, 5, 3, 2, 0, 5, 3, 7, 4, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 7, 2, 0, 6, 0, 3, 7, 7, 2, 5, 2, 3, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 3, 2, 5, 1, 5, 5, 5, 6, 4, 6, 6, 7, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 1, 2, 3, 7, 5, 3, 0, 2, 5, 3, 4, 2, 5, 5, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 7, 3, 3, 6, 0, 3, 0, 4, 3, 6, 3, 3, 5, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 6, 2, 4, 7, 0, 3, 7, 3, 7, 3, 5, 6, 3, 3, 0], device='cuda:0')\n",
      "Epoch 35 (Time: 1.50s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 1, 2, 7, 5, 2, 3, 6, 3, 0, 5, 3, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 7, 6, 1, 0, 6, 3, 3, 6, 2, 0, 3, 6, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 2, 6, 3, 4, 4, 3, 2, 3, 0, 7, 4, 2, 2, 0, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 7, 2, 0, 4, 5, 0, 7, 0, 6, 5, 3, 3, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 6, 6, 5, 4, 7, 2, 7, 4, 2, 4, 3, 3, 4, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 5, 6, 3, 4, 7, 5, 6, 0, 6, 3, 0, 3, 6, 0, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 3, 6, 6, 4, 4, 7, 7, 7, 3, 7, 3, 4, 3, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 6, 0, 7, 6, 7, 2, 7, 4, 5, 7, 0, 6, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 5, 6, 4, 4, 4, 1, 3, 6, 6, 6, 7, 3, 7, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 1, 7, 2, 7, 6, 4, 2, 4, 3, 5, 0, 3, 2, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 7, 3, 0, 1, 7, 3, 0, 0, 3, 6, 4, 6, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 3, 3, 5, 1, 3, 2, 5, 7, 3, 7, 0, 6, 5, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 3, 2, 6, 2, 0, 7, 4, 1, 3, 5, 2, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 3, 1, 0, 6, 3, 5, 3, 7, 0, 0, 6, 7, 2, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 4, 3, 4, 0, 6, 7, 7, 4, 0, 5, 3, 3, 6, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 5, 2, 3, 5, 6, 4, 7, 3, 5, 6, 3, 0, 3, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 6, 5, 3, 2, 3, 5, 3, 7, 3, 7, 0, 0, 2, 6], device='cuda:0')\n",
      "Epoch 36 (Time: 1.44s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 2, 7, 3, 4, 3, 4, 3, 2, 3, 3, 0, 5, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 6, 3, 3, 3, 3, 7, 2, 0, 6, 3, 3, 3, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 2, 7, 4, 4, 4, 5, 7, 7, 3, 2, 6, 0, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 4, 0, 7, 4, 3, 2, 6, 7, 0, 3, 3, 3, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 7, 2, 1, 3, 3, 3, 4, 5, 0, 5, 6, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 4, 4, 6, 3, 3, 5, 3, 6, 2, 7, 3, 7, 6, 0, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 1, 6, 5, 3, 2, 6, 7, 3, 3, 0, 3, 0, 6, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 4, 3, 3, 7, 4, 3, 7, 2, 3, 6, 6, 1, 0, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 7, 3, 1, 5, 2, 3, 3, 7, 4, 3, 5, 7, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 3, 2, 4, 3, 6, 6, 5, 5, 6, 6, 0, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 0, 6, 6, 6, 0, 5, 7, 5, 0, 6, 3, 5, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 7, 2, 3, 6, 7, 3, 3, 2, 2, 4, 1, 7, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 5, 4, 3, 5, 3, 1, 3, 0, 7, 3, 6, 3, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 5, 7, 4, 2, 3, 3, 6, 6, 6, 5, 3, 3, 5, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 0, 1, 1, 3, 7, 6, 3, 6, 6, 7, 3, 4, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 5, 7, 6, 6, 7, 6, 3, 5, 3, 7, 4, 2, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 0, 4, 4, 3, 5, 5, 2, 3, 2, 0, 5, 4, 2, 3], device='cuda:0')\n",
      "Epoch 37 (Time: 1.44s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 3, 1, 0, 6, 7, 5, 2, 3, 4, 0, 3, 4, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 7, 5, 0, 3, 3, 6, 3, 3, 5, 6, 3, 3, 0, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 5, 3, 4, 4, 7, 6, 6, 0, 3, 3, 3, 7, 7, 5, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 0, 5, 0, 3, 5, 2, 4, 4, 4, 0, 5, 0, 1, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 5, 1, 0, 2, 7, 4, 6, 3, 5, 6, 0, 6, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 6, 3, 3, 6, 5, 3, 3, 6, 3, 1, 4, 0, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 5, 6, 3, 4, 0, 3, 5, 3, 6, 6, 3, 5, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 7, 7, 2, 4, 6, 4, 0, 3, 3, 3, 5, 0, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 5, 5, 0, 4, 2, 7, 5, 5, 5, 5, 0, 4, 3, 1, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 0, 3, 3, 5, 4, 7, 2, 7, 2, 2, 7, 0, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 6, 1, 5, 6, 5, 2, 6, 5, 3, 0, 3, 3, 2, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 0, 4, 3, 3, 2, 7, 2, 6, 4, 2, 4, 3, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 3, 7, 7, 5, 3, 0, 4, 2, 0, 1, 7, 2, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 3, 0, 6, 3, 5, 2, 3, 5, 3, 1, 4, 5, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 2, 7, 0, 2, 3, 5, 3, 3, 4, 2, 6, 6, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 6, 3, 0, 6, 7, 5, 3, 7, 5, 2, 0, 5, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 3, 6, 6, 3, 5, 3, 5, 7, 3, 7, 5, 3, 6], device='cuda:0')\n",
      "Epoch 38 (Time: 1.54s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 4, 3, 3, 7, 4, 3, 0, 5, 3, 0, 4, 4, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 6, 3, 3, 1, 7, 2, 6, 6, 5, 5, 0, 3, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 4, 2, 0, 6, 3, 6, 4, 3, 3, 6, 5, 7, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 1, 0, 7, 6, 3, 1, 2, 6, 4, 5, 6, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 6, 3, 4, 4, 4, 4, 5, 6, 2, 7, 3, 3, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 6, 0, 0, 5, 1, 0, 2, 3, 5, 3, 3, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 0, 3, 4, 5, 3, 0, 5, 2, 3, 3, 6, 0, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 0, 6, 5, 1, 7, 3, 3, 4, 2, 4, 7, 4, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 7, 3, 7, 2, 6, 3, 4, 6, 5, 2, 4, 5, 3, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 2, 3, 3, 5, 6, 4, 0, 6, 2, 2, 1, 4, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 2, 6, 7, 3, 2, 5, 3, 3, 7, 3, 7, 6, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 3, 7, 0, 6, 7, 3, 2, 4, 7, 3, 6, 6, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 1, 6, 6, 5, 5, 4, 3, 0, 0, 2, 7, 2, 0, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 6, 7, 6, 2, 3, 3, 0, 3, 4, 7, 2, 5, 3, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 7, 0, 3, 6, 4, 3, 0, 5, 6, 5, 3, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 4, 0, 7, 6, 7, 3, 0, 7, 3, 3, 0, 4, 2, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 3, 0, 0, 3, 6, 6, 3, 3, 3, 5, 3, 5, 2, 3, 3], device='cuda:0')\n",
      "Epoch 39 (Time: 1.54s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 7, 6, 6, 6, 6, 0, 0, 1, 4, 6, 0, 6, 2, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 6, 3, 5, 2, 2, 0, 6, 5, 6, 5, 4, 3, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 4, 3, 0, 4, 7, 7, 2, 2, 4, 5, 3, 4, 2, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 5, 5, 0, 7, 7, 3, 3, 6, 3, 4, 4, 3, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 6, 3, 6, 0, 3, 3, 2, 3, 3, 4, 3, 2, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 6, 0, 0, 2, 6, 3, 2, 5, 0, 0, 6, 3, 3, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 3, 7, 6, 4, 7, 5, 0, 6, 4, 3, 6, 5, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 3, 4, 4, 6, 1, 7, 2, 7, 3, 2, 7, 0, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 3, 6, 2, 5, 7, 7, 0, 0, 6, 5, 6, 3, 0, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 7, 6, 1, 5, 3, 6, 3, 0, 5, 3, 3, 2, 0, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 7, 4, 6, 3, 5, 1, 6, 0, 5, 6, 6, 0, 7, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 6, 0, 6, 0, 7, 3, 3, 7, 4, 0, 5, 1, 5, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 7, 6, 0, 3, 1, 6, 7, 6, 4, 5, 3, 3, 0, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 5, 3, 3, 4, 4, 4, 5, 3, 0, 3, 3, 3, 7, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 3, 3, 4, 4, 5, 5, 3, 7, 7, 3, 2, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 5, 5, 2, 6, 6, 5, 7, 3, 3, 3, 3, 5, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 3, 7, 4, 3, 4, 1, 7, 5, 6, 3, 3, 5, 0, 7], device='cuda:0')\n",
      "Epoch 40 (Time: 1.47s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 7, 4, 0, 6, 7, 2, 6, 5, 3, 5, 7, 4, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 2, 2, 3, 3, 3, 3, 1, 7, 3, 3, 3, 7, 7, 2, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 4, 0, 6, 5, 3, 2, 7, 4, 7, 3, 7, 5, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 4, 4, 2, 2, 0, 6, 5, 5, 3, 5, 3, 5, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 6, 2, 2, 6, 6, 3, 2, 3, 7, 6, 0, 5, 3, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 2, 3, 7, 3, 1, 7, 3, 6, 0, 2, 4, 5, 4, 0, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 3, 2, 3, 7, 0, 3, 5, 5, 7, 6, 4, 6, 0, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 5, 7, 3, 0, 1, 5, 3, 1, 3, 3, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 2, 5, 3, 7, 4, 1, 5, 7, 5, 0, 0, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 6, 3, 3, 5, 3, 6, 3, 3, 7, 3, 6, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 5, 6, 2, 6, 6, 0, 3, 5, 5, 3, 3, 3, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 3, 2, 2, 3, 4, 3, 1, 7, 5, 5, 3, 2, 2, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 4, 4, 5, 2, 5, 4, 0, 6, 3, 0, 4, 1, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 3, 0, 3, 5, 6, 0, 3, 4, 3, 6, 0, 6, 6, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 6, 0, 3, 7, 3, 1, 4, 6, 2, 0, 7, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 6, 2, 7, 4, 3, 3, 0, 3, 6, 3, 4, 2, 0, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 7, 0, 6, 2, 7, 5, 0, 7, 3, 2, 6, 7, 1, 5], device='cuda:0')\n",
      "Epoch 41 (Time: 1.43s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 3, 5, 1, 7, 5, 7, 0, 6, 3, 4, 4, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 4, 2, 3, 0, 2, 0, 3, 7, 4, 6, 5, 5, 3, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 7, 5, 6, 7, 2, 3, 0, 6, 5, 3, 6, 2, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 3, 0, 0, 5, 5, 4, 6, 6, 3, 5, 0, 0, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 3, 2, 2, 6, 5, 3, 3, 5, 3, 1, 4, 4, 7, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 7, 3, 4, 0, 6, 3, 3, 0, 6, 0, 5, 7, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 1, 6, 0, 2, 3, 4, 6, 6, 4, 3, 2, 0, 5, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 4, 6, 4, 6, 3, 3, 3, 5, 4, 3, 1, 5, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 3, 3, 2, 7, 7, 5, 7, 6, 5, 5, 5, 2, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 7, 7, 6, 1, 3, 3, 3, 5, 0, 3, 7, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 0, 3, 2, 7, 6, 4, 4, 5, 5, 5, 6, 3, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 2, 0, 3, 0, 2, 6, 5, 2, 3, 3, 0, 2, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 3, 5, 3, 7, 3, 0, 0, 3, 1, 4, 7, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 7, 1, 6, 3, 0, 6, 4, 4, 3, 4, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 0, 5, 4, 6, 3, 5, 4, 4, 3, 2, 5, 4, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 6, 4, 3, 7, 0, 2, 7, 3, 7, 2, 2, 0, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 6, 7, 3, 6, 1, 3, 3, 2, 3, 3, 0, 4, 6, 6], device='cuda:0')\n",
      "Epoch 42 (Time: 1.57s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 7, 5, 3, 1, 2, 5, 6, 3, 5, 2, 7, 3, 0, 0, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 6, 3, 3, 5, 3, 0, 0, 4, 3, 3, 6, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 7, 3, 0, 6, 2, 6, 4, 3, 2, 2, 2, 7, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 7, 7, 2, 3, 3, 3, 5, 7, 7, 7, 3, 7, 5, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 3, 3, 3, 6, 4, 4, 3, 5, 6, 3, 0, 2, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 2, 3, 5, 3, 0, 3, 4, 4, 1, 6, 4, 6, 5, 1, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 3, 6, 6, 6, 6, 3, 3, 7, 2, 3, 0, 0, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 7, 2, 5, 2, 6, 7, 3, 4, 5, 6, 0, 3, 6, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 6, 5, 5, 6, 4, 1, 2, 0, 0, 5, 3, 2, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 0, 3, 5, 0, 6, 4, 6, 0, 2, 4, 3, 7, 3, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 4, 3, 5, 3, 3, 2, 6, 5, 3, 5, 3, 7, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 0, 3, 5, 0, 5, 6, 4, 6, 3, 4, 3, 4, 3, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 6, 4, 6, 6, 3, 2, 4, 7, 0, 3, 6, 5, 1, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 3, 3, 6, 3, 3, 3, 3, 0, 0, 6, 2, 3, 0, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 0, 5, 0, 3, 3, 3, 7, 4, 2, 6, 4, 0, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 7, 6, 5, 4, 7, 2, 7, 2, 7, 5, 6, 3, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 5, 7, 5, 6, 3, 3, 5, 4, 3, 3, 2, 4, 7, 0], device='cuda:0')\n",
      "Epoch 43 (Time: 1.54s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 0, 3, 3, 4, 5, 6, 2, 5, 4, 7, 0, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 2, 3, 0, 3, 7, 1, 3, 1, 4, 0, 1, 7, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 0, 3, 0, 5, 2, 1, 6, 3, 5, 0, 4, 3, 5, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 5, 6, 4, 5, 4, 6, 6, 6, 0, 3, 0, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 4, 5, 2, 5, 3, 3, 0, 5, 0, 4, 2, 5, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 4, 2, 5, 7, 0, 5, 6, 3, 2, 3, 4, 3, 1, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 7, 4, 6, 3, 5, 3, 5, 5, 0, 3, 7, 4, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 2, 3, 3, 0, 3, 5, 0, 3, 7, 0, 7, 0, 3, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 4, 0, 0, 3, 3, 2, 7, 1, 3, 2, 3, 2, 2, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 3, 0, 5, 2, 0, 6, 4, 2, 3, 5, 6, 2, 4, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 7, 3, 4, 5, 5, 2, 5, 4, 2, 6, 4, 3, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 7, 0, 3, 5, 0, 7, 3, 5, 2, 0, 0, 1, 2, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 3, 5, 6, 3, 1, 2, 4, 3, 6, 6, 7, 7, 6, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 7, 4, 3, 3, 6, 0, 6, 6, 2, 6, 0, 6, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 4, 4, 3, 3, 7, 5, 6, 2, 4, 6, 2, 3, 7, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 7, 3, 0, 3, 5, 2, 3, 3, 3, 5, 7, 4, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 3, 3, 0, 5, 5, 6, 6, 3, 2, 4, 2, 1, 3, 7], device='cuda:0')\n",
      "Epoch 44 (Time: 1.47s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 2, 4, 0, 4, 3, 0, 0, 3, 6, 4, 2, 4, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 3, 7, 0, 6, 0, 3, 7, 7, 0, 1, 6, 1, 2, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 3, 0, 6, 7, 4, 6, 5, 0, 5, 3, 6, 6, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 1, 1, 2, 4, 3, 3, 4, 3, 6, 0, 7, 3, 1, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 2, 3, 4, 7, 2, 5, 3, 5, 4, 5, 3, 5, 7, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 0, 5, 5, 3, 2, 7, 7, 4, 0, 6, 6, 3, 6, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 3, 3, 4, 0, 0, 4, 7, 2, 0, 3, 3, 3, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 0, 5, 4, 3, 7, 5, 3, 3, 4, 3, 3, 5, 7, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 2, 2, 4, 0, 3, 5, 7, 6, 6, 3, 6, 5, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 3, 3, 3, 3, 0, 4, 2, 5, 7, 7, 3, 7, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 3, 7, 5, 5, 5, 2, 3, 0, 0, 5, 6, 2, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 0, 3, 3, 6, 3, 3, 0, 7, 5, 6, 2, 5, 2, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 0, 5, 2, 3, 3, 2, 3, 3, 6, 7, 4, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 7, 1, 3, 3, 5, 4, 5, 5, 3, 1, 4, 7, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 5, 7, 3, 7, 6, 3, 3, 3, 6, 0, 0, 6, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 3, 7, 0, 3, 6, 3, 1, 2, 6, 7, 6, 5, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 3, 6, 3, 5, 3, 3, 3, 0, 0, 6, 3, 4, 4, 5], device='cuda:0')\n",
      "Epoch 45 (Time: 1.49s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 7, 6, 4, 7, 7, 2, 5, 3, 6, 4, 5, 7, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 1, 3, 0, 6, 5, 7, 3, 5, 6, 6, 6, 2, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 3, 2, 5, 4, 7, 7, 5, 0, 3, 7, 3, 5, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 4, 3, 3, 3, 3, 0, 3, 4, 2, 7, 3, 7, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 6, 3, 5, 6, 1, 5, 6, 5, 6, 6, 5, 2, 4, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 0, 3, 4, 4, 6, 2, 3, 2, 7, 5, 0, 3, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 0, 1, 6, 3, 3, 6, 3, 5, 4, 2, 7, 1, 1, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 3, 6, 3, 6, 4, 5, 4, 5, 5, 3, 3, 7, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 3, 6, 6, 5, 7, 0, 3, 3, 4, 3, 6, 7, 0, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 6, 7, 3, 3, 3, 0, 2, 4, 3, 4, 3, 5, 5, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 0, 4, 3, 6, 3, 5, 5, 3, 7, 7, 0, 6, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 2, 6, 2, 2, 4, 0, 1, 3, 6, 3, 5, 1, 0, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 7, 5, 3, 6, 0, 6, 4, 0, 6, 2, 5, 0, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 5, 2, 3, 3, 0, 6, 1, 3, 6, 3, 3, 7, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 7, 0, 6, 0, 2, 2, 0, 6, 2, 3, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 5, 0, 4, 6, 3, 0, 3, 7, 7, 4, 3, 3, 0, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 6, 3, 2, 0, 3, 3, 6, 4, 5, 2, 0, 7, 7, 5, 6], device='cuda:0')\n",
      "Epoch 46 (Time: 1.47s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 5, 6, 5, 2, 3, 2, 7, 6, 0, 5, 7, 0, 7, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 5, 6, 6, 3, 3, 7, 3, 0, 0, 4, 6, 6, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 5, 3, 6, 3, 3, 1, 4, 5, 3, 7, 3, 4, 6, 6, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 7, 2, 2, 4, 0, 7, 4, 3, 3, 6, 2, 0, 0, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 7, 7, 4, 3, 5, 1, 3, 2, 5, 1, 0, 3, 0, 6, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 3, 7, 3, 3, 5, 3, 0, 0, 3, 7, 5, 7, 6, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 4, 7, 5, 0, 3, 1, 0, 3, 6, 7, 7, 4, 0, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 4, 4, 7, 6, 6, 3, 2, 6, 7, 0, 5, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 4, 0, 3, 7, 0, 6, 6, 3, 2, 7, 7, 6, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 4, 6, 6, 2, 3, 3, 0, 3, 3, 7, 6, 6, 4, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 4, 4, 4, 5, 3, 0, 3, 4, 6, 4, 4, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 6, 2, 3, 7, 2, 5, 3, 3, 3, 6, 0, 6, 3, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 3, 1, 0, 2, 3, 1, 5, 0, 3, 2, 4, 7, 4, 5, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 4, 7, 3, 3, 6, 0, 6, 0, 2, 5, 3, 5, 6, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 4, 0, 5, 3, 6, 5, 3, 2, 2, 3, 6, 4, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 3, 6, 5, 0, 5, 2, 6, 5, 6, 2, 3, 5, 4, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 6, 5, 3, 0, 2, 3, 3, 3, 3, 3, 2, 2, 0, 3, 6], device='cuda:0')\n",
      "Epoch 47 (Time: 1.56s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 4, 3, 0, 5, 3, 6, 5, 3, 4, 6, 3, 7, 5, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 4, 3, 6, 6, 6, 4, 0, 3, 2, 5, 7, 2, 0, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 5, 6, 0, 5, 0, 5, 0, 5, 6, 7, 4, 4, 7, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 7, 2, 3, 3, 5, 6, 3, 3, 5, 3, 7, 3, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 5, 3, 2, 4, 0, 7, 7, 2, 0, 6, 4, 0, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 1, 4, 2, 3, 3, 6, 6, 6, 1, 2, 7, 2, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 6, 3, 3, 6, 2, 7, 3, 2, 4, 5, 4, 6, 5, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 6, 7, 5, 7, 2, 2, 2, 3, 3, 3, 1, 7, 6, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 6, 3, 3, 1, 6, 2, 5, 7, 0, 0, 0, 5, 3, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 2, 0, 0, 0, 2, 3, 3, 3, 3, 7, 0, 6, 0, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 1, 3, 5, 7, 7, 2, 4, 3, 5, 7, 4, 5, 4, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 1, 0, 5, 0, 3, 5, 4, 2, 6, 5, 6, 3, 4, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 6, 2, 2, 4, 6, 3, 3, 3, 0, 5, 3, 5, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([1, 3, 4, 3, 6, 4, 6, 3, 7, 6, 2, 0, 7, 0, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 7, 6, 6, 4, 1, 6, 3, 6, 0, 0, 7, 2, 2, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 5, 7, 3, 2, 6, 5, 2, 3, 3, 6, 0, 5, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 6, 6, 4, 7, 6, 3, 3, 0, 4, 2, 2, 0, 0, 5], device='cuda:0')\n",
      "Epoch 48 (Time: 1.51s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 3, 4, 1, 3, 4, 5, 5, 0, 5, 7, 4, 6, 3, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 5, 6, 0, 7, 2, 0, 2, 6, 5, 2, 1, 5, 6, 4, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 3, 5, 3, 0, 0, 3, 2, 3, 3, 0, 3, 6, 3, 7, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 7, 0, 5, 0, 4, 6, 4, 5, 4, 3, 2, 3, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 3, 5, 0, 2, 3, 0, 3, 6, 4, 6, 6, 5, 0, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 6, 3, 6, 0, 4, 6, 0, 0, 3, 4, 6, 2, 2, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 4, 5, 7, 6, 6, 6, 6, 2, 3, 3, 5, 4, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 4, 3, 3, 2, 0, 0, 5, 1, 5, 7, 4, 3, 5, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 0, 3, 6, 4, 7, 6, 1, 6, 0, 3, 3, 5, 3, 7, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 7, 3, 7, 4, 3, 3, 3, 5, 6, 6, 5, 7, 2, 5, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 3, 0, 2, 4, 5, 2, 7, 0, 7, 0, 4, 3, 3, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 3, 6, 7, 3, 2, 4, 3, 4, 3, 3, 6, 2, 3, 3, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 5, 4, 7, 5, 2, 6, 3, 0, 3, 0, 6, 3, 1, 6, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([4, 3, 3, 4, 7, 4, 7, 6, 3, 7, 3, 6, 4, 5, 1, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 7, 2, 6, 3, 3, 5, 7, 4, 5, 1, 7, 3, 5, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 4, 3, 7, 2, 5, 6, 0, 0, 3, 6, 3, 2, 3, 0, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([5, 1, 3, 0, 3, 3, 0, 3, 2, 5, 3, 6, 7, 5, 7, 4], device='cuda:0')\n",
      "Epoch 49 (Time: 1.50s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 4, 3, 3, 2, 2, 3, 7, 3, 4, 0, 0, 0, 3, 3, 1], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 3, 5, 0, 3, 3, 7, 6, 5, 5, 0, 7, 4, 3, 0, 5], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 6, 5, 7, 3, 0, 3, 3, 5, 7, 6, 5, 3, 3, 2, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([0, 4, 3, 5, 6, 3, 6, 7, 3, 0, 5, 6, 3, 3, 4, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 7, 3, 4, 3, 2, 0, 5, 3, 6, 3, 3, 7, 2, 6, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 1, 1, 3, 3, 4, 4, 6, 1, 7, 6, 0, 0, 3, 7, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 3, 4, 0, 6, 4, 6, 6, 7, 0, 3, 3, 7, 3, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 4, 5, 5, 4, 0, 2, 7, 5, 5, 0, 7, 3, 3, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 4, 3, 6, 2, 2, 3, 5, 6, 3, 7, 5, 7, 3, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 2, 7, 3, 5, 7, 4, 6, 2, 0, 3, 5, 1, 6, 4, 2], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 6, 3, 6, 3, 7, 0, 3, 7, 0, 4, 6, 0, 3, 6, 0], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 5, 4, 6, 6, 4, 7, 7, 5, 5, 3, 7, 3, 7, 2, 3], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([7, 2, 5, 1, 3, 5, 3, 1, 3, 5, 4, 6, 5, 3, 3, 7], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 0, 7, 4, 0, 5, 7, 6, 6, 5, 3, 4, 3, 2, 2, 4], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([3, 2, 2, 3, 5, 7, 5, 0, 5, 6, 3, 3, 6, 3, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([6, 0, 7, 7, 0, 3, 3, 4, 1, 4, 4, 5, 7, 6, 6, 6], device='cuda:0')\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "<class 'PIL.Image.Image'>\n",
      "tensor([2, 7, 4, 3, 3, 0, 3, 7, 1, 0, 3, 4, 2, 4, 0, 5], device='cuda:0')\n",
      "Epoch 50 (Time: 1.49s)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finish training for 50 epochs\n"
     ]
    }
   ],
   "source": [
    "fig, axes = plt.subplots(ncols=2, figsize=(7, 3))\n",
    "\n",
    "exp1.run(num_epochs=50, plot=lambda exp: plot(exp, fig=fig, axes=axes))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "/* Put everything inside the global mpl namespace */\n",
       "window.mpl = {};\n",
       "\n",
       "\n",
       "mpl.get_websocket_type = function() {\n",
       "    if (typeof(WebSocket) !== 'undefined') {\n",
       "        return WebSocket;\n",
       "    } else if (typeof(MozWebSocket) !== 'undefined') {\n",
       "        return MozWebSocket;\n",
       "    } else {\n",
       "        alert('Your browser does not have WebSocket support. ' +\n",
       "              'Please try Chrome, Safari or Firefox ≥ 6. ' +\n",
       "              'Firefox 4 and 5 are also supported but you ' +\n",
       "              'have to enable WebSockets in about:config.');\n",
       "    };\n",
       "}\n",
       "\n",
       "mpl.figure = function(figure_id, websocket, ondownload, parent_element) {\n",
       "    this.id = figure_id;\n",
       "\n",
       "    this.ws = websocket;\n",
       "\n",
       "    this.supports_binary = (this.ws.binaryType != undefined);\n",
       "\n",
       "    if (!this.supports_binary) {\n",
       "        var warnings = document.getElementById(\"mpl-warnings\");\n",
       "        if (warnings) {\n",
       "            warnings.style.display = 'block';\n",
       "            warnings.textContent = (\n",
       "                \"This browser does not support binary websocket messages. \" +\n",
       "                    \"Performance may be slow.\");\n",
       "        }\n",
       "    }\n",
       "\n",
       "    this.imageObj = new Image();\n",
       "\n",
       "    this.context = undefined;\n",
       "    this.message = undefined;\n",
       "    this.canvas = undefined;\n",
       "    this.rubberband_canvas = undefined;\n",
       "    this.rubberband_context = undefined;\n",
       "    this.format_dropdown = undefined;\n",
       "\n",
       "    this.image_mode = 'full';\n",
       "\n",
       "    this.root = $('<div/>');\n",
       "    this._root_extra_style(this.root)\n",
       "    this.root.attr('style', 'display: inline-block');\n",
       "\n",
       "    $(parent_element).append(this.root);\n",
       "\n",
       "    this._init_header(this);\n",
       "    this._init_canvas(this);\n",
       "    this._init_toolbar(this);\n",
       "\n",
       "    var fig = this;\n",
       "\n",
       "    this.waiting = false;\n",
       "\n",
       "    this.ws.onopen =  function () {\n",
       "            fig.send_message(\"supports_binary\", {value: fig.supports_binary});\n",
       "            fig.send_message(\"send_image_mode\", {});\n",
       "            if (mpl.ratio != 1) {\n",
       "                fig.send_message(\"set_dpi_ratio\", {'dpi_ratio': mpl.ratio});\n",
       "            }\n",
       "            fig.send_message(\"refresh\", {});\n",
       "        }\n",
       "\n",
       "    this.imageObj.onload = function() {\n",
       "            if (fig.image_mode == 'full') {\n",
       "                // Full images could contain transparency (where diff images\n",
       "                // almost always do), so we need to clear the canvas so that\n",
       "                // there is no ghosting.\n",
       "                fig.context.clearRect(0, 0, fig.canvas.width, fig.canvas.height);\n",
       "            }\n",
       "            fig.context.drawImage(fig.imageObj, 0, 0);\n",
       "        };\n",
       "\n",
       "    this.imageObj.onunload = function() {\n",
       "        fig.ws.close();\n",
       "    }\n",
       "\n",
       "    this.ws.onmessage = this._make_on_message_function(this);\n",
       "\n",
       "    this.ondownload = ondownload;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_header = function() {\n",
       "    var titlebar = $(\n",
       "        '<div class=\"ui-dialog-titlebar ui-widget-header ui-corner-all ' +\n",
       "        'ui-helper-clearfix\"/>');\n",
       "    var titletext = $(\n",
       "        '<div class=\"ui-dialog-title\" style=\"width: 100%; ' +\n",
       "        'text-align: center; padding: 3px;\"/>');\n",
       "    titlebar.append(titletext)\n",
       "    this.root.append(titlebar);\n",
       "    this.header = titletext[0];\n",
       "}\n",
       "\n",
       "\n",
       "\n",
       "mpl.figure.prototype._canvas_extra_style = function(canvas_div) {\n",
       "\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype._root_extra_style = function(canvas_div) {\n",
       "\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_canvas = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var canvas_div = $('<div/>');\n",
       "\n",
       "    canvas_div.attr('style', 'position: relative; clear: both; outline: 0');\n",
       "\n",
       "    function canvas_keyboard_event(event) {\n",
       "        return fig.key_event(event, event['data']);\n",
       "    }\n",
       "\n",
       "    canvas_div.keydown('key_press', canvas_keyboard_event);\n",
       "    canvas_div.keyup('key_release', canvas_keyboard_event);\n",
       "    this.canvas_div = canvas_div\n",
       "    this._canvas_extra_style(canvas_div)\n",
       "    this.root.append(canvas_div);\n",
       "\n",
       "    var canvas = $('<canvas/>');\n",
       "    canvas.addClass('mpl-canvas');\n",
       "    canvas.attr('style', \"left: 0; top: 0; z-index: 0; outline: 0\")\n",
       "\n",
       "    this.canvas = canvas[0];\n",
       "    this.context = canvas[0].getContext(\"2d\");\n",
       "\n",
       "    var backingStore = this.context.backingStorePixelRatio ||\n",
       "\tthis.context.webkitBackingStorePixelRatio ||\n",
       "\tthis.context.mozBackingStorePixelRatio ||\n",
       "\tthis.context.msBackingStorePixelRatio ||\n",
       "\tthis.context.oBackingStorePixelRatio ||\n",
       "\tthis.context.backingStorePixelRatio || 1;\n",
       "\n",
       "    mpl.ratio = (window.devicePixelRatio || 1) / backingStore;\n",
       "\n",
       "    var rubberband = $('<canvas/>');\n",
       "    rubberband.attr('style', \"position: absolute; left: 0; top: 0; z-index: 1;\")\n",
       "\n",
       "    var pass_mouse_events = true;\n",
       "\n",
       "    canvas_div.resizable({\n",
       "        start: function(event, ui) {\n",
       "            pass_mouse_events = false;\n",
       "        },\n",
       "        resize: function(event, ui) {\n",
       "            fig.request_resize(ui.size.width, ui.size.height);\n",
       "        },\n",
       "        stop: function(event, ui) {\n",
       "            pass_mouse_events = true;\n",
       "            fig.request_resize(ui.size.width, ui.size.height);\n",
       "        },\n",
       "    });\n",
       "\n",
       "    function mouse_event_fn(event) {\n",
       "        if (pass_mouse_events)\n",
       "            return fig.mouse_event(event, event['data']);\n",
       "    }\n",
       "\n",
       "    rubberband.mousedown('button_press', mouse_event_fn);\n",
       "    rubberband.mouseup('button_release', mouse_event_fn);\n",
       "    // Throttle sequential mouse events to 1 every 20ms.\n",
       "    rubberband.mousemove('motion_notify', mouse_event_fn);\n",
       "\n",
       "    rubberband.mouseenter('figure_enter', mouse_event_fn);\n",
       "    rubberband.mouseleave('figure_leave', mouse_event_fn);\n",
       "\n",
       "    canvas_div.on(\"wheel\", function (event) {\n",
       "        event = event.originalEvent;\n",
       "        event['data'] = 'scroll'\n",
       "        if (event.deltaY < 0) {\n",
       "            event.step = 1;\n",
       "        } else {\n",
       "            event.step = -1;\n",
       "        }\n",
       "        mouse_event_fn(event);\n",
       "    });\n",
       "\n",
       "    canvas_div.append(canvas);\n",
       "    canvas_div.append(rubberband);\n",
       "\n",
       "    this.rubberband = rubberband;\n",
       "    this.rubberband_canvas = rubberband[0];\n",
       "    this.rubberband_context = rubberband[0].getContext(\"2d\");\n",
       "    this.rubberband_context.strokeStyle = \"#000000\";\n",
       "\n",
       "    this._resize_canvas = function(width, height) {\n",
       "        // Keep the size of the canvas, canvas container, and rubber band\n",
       "        // canvas in synch.\n",
       "        canvas_div.css('width', width)\n",
       "        canvas_div.css('height', height)\n",
       "\n",
       "        canvas.attr('width', width * mpl.ratio);\n",
       "        canvas.attr('height', height * mpl.ratio);\n",
       "        canvas.attr('style', 'width: ' + width + 'px; height: ' + height + 'px;');\n",
       "\n",
       "        rubberband.attr('width', width);\n",
       "        rubberband.attr('height', height);\n",
       "    }\n",
       "\n",
       "    // Set the figure to an initial 600x600px, this will subsequently be updated\n",
       "    // upon first draw.\n",
       "    this._resize_canvas(600, 600);\n",
       "\n",
       "    // Disable right mouse context menu.\n",
       "    $(this.rubberband_canvas).bind(\"contextmenu\",function(e){\n",
       "        return false;\n",
       "    });\n",
       "\n",
       "    function set_focus () {\n",
       "        canvas.focus();\n",
       "        canvas_div.focus();\n",
       "    }\n",
       "\n",
       "    window.setTimeout(set_focus, 100);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_toolbar = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var nav_element = $('<div/>');\n",
       "    nav_element.attr('style', 'width: 100%');\n",
       "    this.root.append(nav_element);\n",
       "\n",
       "    // Define a callback function for later on.\n",
       "    function toolbar_event(event) {\n",
       "        return fig.toolbar_button_onclick(event['data']);\n",
       "    }\n",
       "    function toolbar_mouse_event(event) {\n",
       "        return fig.toolbar_button_onmouseover(event['data']);\n",
       "    }\n",
       "\n",
       "    for(var toolbar_ind in mpl.toolbar_items) {\n",
       "        var name = mpl.toolbar_items[toolbar_ind][0];\n",
       "        var tooltip = mpl.toolbar_items[toolbar_ind][1];\n",
       "        var image = mpl.toolbar_items[toolbar_ind][2];\n",
       "        var method_name = mpl.toolbar_items[toolbar_ind][3];\n",
       "\n",
       "        if (!name) {\n",
       "            // put a spacer in here.\n",
       "            continue;\n",
       "        }\n",
       "        var button = $('<button/>');\n",
       "        button.addClass('ui-button ui-widget ui-state-default ui-corner-all ' +\n",
       "                        'ui-button-icon-only');\n",
       "        button.attr('role', 'button');\n",
       "        button.attr('aria-disabled', 'false');\n",
       "        button.click(method_name, toolbar_event);\n",
       "        button.mouseover(tooltip, toolbar_mouse_event);\n",
       "\n",
       "        var icon_img = $('<span/>');\n",
       "        icon_img.addClass('ui-button-icon-primary ui-icon');\n",
       "        icon_img.addClass(image);\n",
       "        icon_img.addClass('ui-corner-all');\n",
       "\n",
       "        var tooltip_span = $('<span/>');\n",
       "        tooltip_span.addClass('ui-button-text');\n",
       "        tooltip_span.html(tooltip);\n",
       "\n",
       "        button.append(icon_img);\n",
       "        button.append(tooltip_span);\n",
       "\n",
       "        nav_element.append(button);\n",
       "    }\n",
       "\n",
       "    var fmt_picker_span = $('<span/>');\n",
       "\n",
       "    var fmt_picker = $('<select/>');\n",
       "    fmt_picker.addClass('mpl-toolbar-option ui-widget ui-widget-content');\n",
       "    fmt_picker_span.append(fmt_picker);\n",
       "    nav_element.append(fmt_picker_span);\n",
       "    this.format_dropdown = fmt_picker[0];\n",
       "\n",
       "    for (var ind in mpl.extensions) {\n",
       "        var fmt = mpl.extensions[ind];\n",
       "        var option = $(\n",
       "            '<option/>', {selected: fmt === mpl.default_extension}).html(fmt);\n",
       "        fmt_picker.append(option);\n",
       "    }\n",
       "\n",
       "    // Add hover states to the ui-buttons\n",
       "    $( \".ui-button\" ).hover(\n",
       "        function() { $(this).addClass(\"ui-state-hover\");},\n",
       "        function() { $(this).removeClass(\"ui-state-hover\");}\n",
       "    );\n",
       "\n",
       "    var status_bar = $('<span class=\"mpl-message\"/>');\n",
       "    nav_element.append(status_bar);\n",
       "    this.message = status_bar[0];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.request_resize = function(x_pixels, y_pixels) {\n",
       "    // Request matplotlib to resize the figure. Matplotlib will then trigger a resize in the client,\n",
       "    // which will in turn request a refresh of the image.\n",
       "    this.send_message('resize', {'width': x_pixels, 'height': y_pixels});\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.send_message = function(type, properties) {\n",
       "    properties['type'] = type;\n",
       "    properties['figure_id'] = this.id;\n",
       "    this.ws.send(JSON.stringify(properties));\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.send_draw_message = function() {\n",
       "    if (!this.waiting) {\n",
       "        this.waiting = true;\n",
       "        this.ws.send(JSON.stringify({type: \"draw\", figure_id: this.id}));\n",
       "    }\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype.handle_save = function(fig, msg) {\n",
       "    var format_dropdown = fig.format_dropdown;\n",
       "    var format = format_dropdown.options[format_dropdown.selectedIndex].value;\n",
       "    fig.ondownload(fig, format);\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype.handle_resize = function(fig, msg) {\n",
       "    var size = msg['size'];\n",
       "    if (size[0] != fig.canvas.width || size[1] != fig.canvas.height) {\n",
       "        fig._resize_canvas(size[0], size[1]);\n",
       "        fig.send_message(\"refresh\", {});\n",
       "    };\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_rubberband = function(fig, msg) {\n",
       "    var x0 = msg['x0'] / mpl.ratio;\n",
       "    var y0 = (fig.canvas.height - msg['y0']) / mpl.ratio;\n",
       "    var x1 = msg['x1'] / mpl.ratio;\n",
       "    var y1 = (fig.canvas.height - msg['y1']) / mpl.ratio;\n",
       "    x0 = Math.floor(x0) + 0.5;\n",
       "    y0 = Math.floor(y0) + 0.5;\n",
       "    x1 = Math.floor(x1) + 0.5;\n",
       "    y1 = Math.floor(y1) + 0.5;\n",
       "    var min_x = Math.min(x0, x1);\n",
       "    var min_y = Math.min(y0, y1);\n",
       "    var width = Math.abs(x1 - x0);\n",
       "    var height = Math.abs(y1 - y0);\n",
       "\n",
       "    fig.rubberband_context.clearRect(\n",
       "        0, 0, fig.canvas.width / mpl.ratio, fig.canvas.height / mpl.ratio);\n",
       "\n",
       "    fig.rubberband_context.strokeRect(min_x, min_y, width, height);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_figure_label = function(fig, msg) {\n",
       "    // Updates the figure title.\n",
       "    fig.header.textContent = msg['label'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_cursor = function(fig, msg) {\n",
       "    var cursor = msg['cursor'];\n",
       "    switch(cursor)\n",
       "    {\n",
       "    case 0:\n",
       "        cursor = 'pointer';\n",
       "        break;\n",
       "    case 1:\n",
       "        cursor = 'default';\n",
       "        break;\n",
       "    case 2:\n",
       "        cursor = 'crosshair';\n",
       "        break;\n",
       "    case 3:\n",
       "        cursor = 'move';\n",
       "        break;\n",
       "    }\n",
       "    fig.rubberband_canvas.style.cursor = cursor;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_message = function(fig, msg) {\n",
       "    fig.message.textContent = msg['message'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_draw = function(fig, msg) {\n",
       "    // Request the server to send over a new figure.\n",
       "    fig.send_draw_message();\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_image_mode = function(fig, msg) {\n",
       "    fig.image_mode = msg['mode'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.updated_canvas_event = function() {\n",
       "    // Called whenever the canvas gets updated.\n",
       "    this.send_message(\"ack\", {});\n",
       "}\n",
       "\n",
       "// A function to construct a web socket function for onmessage handling.\n",
       "// Called in the figure constructor.\n",
       "mpl.figure.prototype._make_on_message_function = function(fig) {\n",
       "    return function socket_on_message(evt) {\n",
       "        if (evt.data instanceof Blob) {\n",
       "            /* FIXME: We get \"Resource interpreted as Image but\n",
       "             * transferred with MIME type text/plain:\" errors on\n",
       "             * Chrome.  But how to set the MIME type?  It doesn't seem\n",
       "             * to be part of the websocket stream */\n",
       "            evt.data.type = \"image/png\";\n",
       "\n",
       "            /* Free the memory for the previous frames */\n",
       "            if (fig.imageObj.src) {\n",
       "                (window.URL || window.webkitURL).revokeObjectURL(\n",
       "                    fig.imageObj.src);\n",
       "            }\n",
       "\n",
       "            fig.imageObj.src = (window.URL || window.webkitURL).createObjectURL(\n",
       "                evt.data);\n",
       "            fig.updated_canvas_event();\n",
       "            fig.waiting = false;\n",
       "            return;\n",
       "        }\n",
       "        else if (typeof evt.data === 'string' && evt.data.slice(0, 21) == \"data:image/png;base64\") {\n",
       "            fig.imageObj.src = evt.data;\n",
       "            fig.updated_canvas_event();\n",
       "            fig.waiting = false;\n",
       "            return;\n",
       "        }\n",
       "\n",
       "        var msg = JSON.parse(evt.data);\n",
       "        var msg_type = msg['type'];\n",
       "\n",
       "        // Call the  \"handle_{type}\" callback, which takes\n",
       "        // the figure and JSON message as its only arguments.\n",
       "        try {\n",
       "            var callback = fig[\"handle_\" + msg_type];\n",
       "        } catch (e) {\n",
       "            console.log(\"No handler for the '\" + msg_type + \"' message type: \", msg);\n",
       "            return;\n",
       "        }\n",
       "\n",
       "        if (callback) {\n",
       "            try {\n",
       "                // console.log(\"Handling '\" + msg_type + \"' message: \", msg);\n",
       "                callback(fig, msg);\n",
       "            } catch (e) {\n",
       "                console.log(\"Exception inside the 'handler_\" + msg_type + \"' callback:\", e, e.stack, msg);\n",
       "            }\n",
       "        }\n",
       "    };\n",
       "}\n",
       "\n",
       "// from http://stackoverflow.com/questions/1114465/getting-mouse-location-in-canvas\n",
       "mpl.findpos = function(e) {\n",
       "    //this section is from http://www.quirksmode.org/js/events_properties.html\n",
       "    var targ;\n",
       "    if (!e)\n",
       "        e = window.event;\n",
       "    if (e.target)\n",
       "        targ = e.target;\n",
       "    else if (e.srcElement)\n",
       "        targ = e.srcElement;\n",
       "    if (targ.nodeType == 3) // defeat Safari bug\n",
       "        targ = targ.parentNode;\n",
       "\n",
       "    // jQuery normalizes the pageX and pageY\n",
       "    // pageX,Y are the mouse positions relative to the document\n",
       "    // offset() returns the position of the element relative to the document\n",
       "    var x = e.pageX - $(targ).offset().left;\n",
       "    var y = e.pageY - $(targ).offset().top;\n",
       "\n",
       "    return {\"x\": x, \"y\": y};\n",
       "};\n",
       "\n",
       "/*\n",
       " * return a copy of an object with only non-object keys\n",
       " * we need this to avoid circular references\n",
       " * http://stackoverflow.com/a/24161582/3208463\n",
       " */\n",
       "function simpleKeys (original) {\n",
       "  return Object.keys(original).reduce(function (obj, key) {\n",
       "    if (typeof original[key] !== 'object')\n",
       "        obj[key] = original[key]\n",
       "    return obj;\n",
       "  }, {});\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.mouse_event = function(event, name) {\n",
       "    var canvas_pos = mpl.findpos(event)\n",
       "\n",
       "    if (name === 'button_press')\n",
       "    {\n",
       "        this.canvas.focus();\n",
       "        this.canvas_div.focus();\n",
       "    }\n",
       "\n",
       "    var x = canvas_pos.x * mpl.ratio;\n",
       "    var y = canvas_pos.y * mpl.ratio;\n",
       "\n",
       "    this.send_message(name, {x: x, y: y, button: event.button,\n",
       "                             step: event.step,\n",
       "                             guiEvent: simpleKeys(event)});\n",
       "\n",
       "    /* This prevents the web browser from automatically changing to\n",
       "     * the text insertion cursor when the button is pressed.  We want\n",
       "     * to control all of the cursor setting manually through the\n",
       "     * 'cursor' event from matplotlib */\n",
       "    event.preventDefault();\n",
       "    return false;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._key_event_extra = function(event, name) {\n",
       "    // Handle any extra behaviour associated with a key event\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.key_event = function(event, name) {\n",
       "\n",
       "    // Prevent repeat events\n",
       "    if (name == 'key_press')\n",
       "    {\n",
       "        if (event.which === this._key)\n",
       "            return;\n",
       "        else\n",
       "            this._key = event.which;\n",
       "    }\n",
       "    if (name == 'key_release')\n",
       "        this._key = null;\n",
       "\n",
       "    var value = '';\n",
       "    if (event.ctrlKey && event.which != 17)\n",
       "        value += \"ctrl+\";\n",
       "    if (event.altKey && event.which != 18)\n",
       "        value += \"alt+\";\n",
       "    if (event.shiftKey && event.which != 16)\n",
       "        value += \"shift+\";\n",
       "\n",
       "    value += 'k';\n",
       "    value += event.which.toString();\n",
       "\n",
       "    this._key_event_extra(event, name);\n",
       "\n",
       "    this.send_message(name, {key: value,\n",
       "                             guiEvent: simpleKeys(event)});\n",
       "    return false;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.toolbar_button_onclick = function(name) {\n",
       "    if (name == 'download') {\n",
       "        this.handle_save(this, null);\n",
       "    } else {\n",
       "        this.send_message(\"toolbar_button\", {name: name});\n",
       "    }\n",
       "};\n",
       "\n",
       "mpl.figure.prototype.toolbar_button_onmouseover = function(tooltip) {\n",
       "    this.message.textContent = tooltip;\n",
       "};\n",
       "mpl.toolbar_items = [[\"Home\", \"Reset original view\", \"fa fa-home icon-home\", \"home\"], [\"Back\", \"Back to previous view\", \"fa fa-arrow-left icon-arrow-left\", \"back\"], [\"Forward\", \"Forward to next view\", \"fa fa-arrow-right icon-arrow-right\", \"forward\"], [\"\", \"\", \"\", \"\"], [\"Pan\", \"Pan axes with left mouse, zoom with right\", \"fa fa-arrows icon-move\", \"pan\"], [\"Zoom\", \"Zoom to rectangle\", \"fa fa-square-o icon-check-empty\", \"zoom\"], [\"\", \"\", \"\", \"\"], [\"Download\", \"Download plot\", \"fa fa-floppy-o icon-save\", \"download\"]];\n",
       "\n",
       "mpl.extensions = [\"eps\", \"jpeg\", \"pdf\", \"png\", \"ps\", \"raw\", \"svg\", \"tif\"];\n",
       "\n",
       "mpl.default_extension = \"png\";var comm_websocket_adapter = function(comm) {\n",
       "    // Create a \"websocket\"-like object which calls the given IPython comm\n",
       "    // object with the appropriate methods. Currently this is a non binary\n",
       "    // socket, so there is still some room for performance tuning.\n",
       "    var ws = {};\n",
       "\n",
       "    ws.close = function() {\n",
       "        comm.close()\n",
       "    };\n",
       "    ws.send = function(m) {\n",
       "        //console.log('sending', m);\n",
       "        comm.send(m);\n",
       "    };\n",
       "    // Register the callback with on_msg.\n",
       "    comm.on_msg(function(msg) {\n",
       "        //console.log('receiving', msg['content']['data'], msg);\n",
       "        // Pass the mpl event to the overridden (by mpl) onmessage function.\n",
       "        ws.onmessage(msg['content']['data'])\n",
       "    });\n",
       "    return ws;\n",
       "}\n",
       "\n",
       "mpl.mpl_figure_comm = function(comm, msg) {\n",
       "    // This is the function which gets called when the mpl process\n",
       "    // starts-up an IPython Comm through the \"matplotlib\" channel.\n",
       "\n",
       "    var id = msg.content.data.id;\n",
       "    // Get hold of the div created by the display call when the Comm\n",
       "    // socket was opened in Python.\n",
       "    var element = $(\"#\" + id);\n",
       "    var ws_proxy = comm_websocket_adapter(comm)\n",
       "\n",
       "    function ondownload(figure, format) {\n",
       "        window.open(figure.imageObj.src);\n",
       "    }\n",
       "\n",
       "    var fig = new mpl.figure(id, ws_proxy,\n",
       "                           ondownload,\n",
       "                           element.get(0));\n",
       "\n",
       "    // Call onopen now - mpl needs it, as it is assuming we've passed it a real\n",
       "    // web socket which is closed, not our websocket->open comm proxy.\n",
       "    ws_proxy.onopen();\n",
       "\n",
       "    fig.parent_element = element.get(0);\n",
       "    fig.cell_info = mpl.find_output_cell(\"<div id='\" + id + \"'></div>\");\n",
       "    if (!fig.cell_info) {\n",
       "        console.error(\"Failed to find cell for figure\", id, fig);\n",
       "        return;\n",
       "    }\n",
       "\n",
       "    var output_index = fig.cell_info[2]\n",
       "    var cell = fig.cell_info[0];\n",
       "\n",
       "};\n",
       "\n",
       "mpl.figure.prototype.handle_close = function(fig, msg) {\n",
       "    var width = fig.canvas.width/mpl.ratio\n",
       "    fig.root.unbind('remove')\n",
       "\n",
       "    // Update the output cell to use the data from the current canvas.\n",
       "    fig.push_to_output();\n",
       "    var dataURL = fig.canvas.toDataURL();\n",
       "    // Re-enable the keyboard manager in IPython - without this line, in FF,\n",
       "    // the notebook keyboard shortcuts fail.\n",
       "    IPython.keyboard_manager.enable()\n",
       "    $(fig.parent_element).html('<img src=\"' + dataURL + '\" width=\"' + width + '\">');\n",
       "    fig.close_ws(fig, msg);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.close_ws = function(fig, msg){\n",
       "    fig.send_message('closing', msg);\n",
       "    // fig.ws.close()\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.push_to_output = function(remove_interactive) {\n",
       "    // Turn the data on the canvas into data in the output cell.\n",
       "    var width = this.canvas.width/mpl.ratio\n",
       "    var dataURL = this.canvas.toDataURL();\n",
       "    this.cell_info[1]['text/html'] = '<img src=\"' + dataURL + '\" width=\"' + width + '\">';\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.updated_canvas_event = function() {\n",
       "    // Tell IPython that the notebook contents must change.\n",
       "    IPython.notebook.set_dirty(true);\n",
       "    this.send_message(\"ack\", {});\n",
       "    var fig = this;\n",
       "    // Wait a second, then push the new image to the DOM so\n",
       "    // that it is saved nicely (might be nice to debounce this).\n",
       "    setTimeout(function () { fig.push_to_output() }, 1000);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_toolbar = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var nav_element = $('<div/>');\n",
       "    nav_element.attr('style', 'width: 100%');\n",
       "    this.root.append(nav_element);\n",
       "\n",
       "    // Define a callback function for later on.\n",
       "    function toolbar_event(event) {\n",
       "        return fig.toolbar_button_onclick(event['data']);\n",
       "    }\n",
       "    function toolbar_mouse_event(event) {\n",
       "        return fig.toolbar_button_onmouseover(event['data']);\n",
       "    }\n",
       "\n",
       "    for(var toolbar_ind in mpl.toolbar_items){\n",
       "        var name = mpl.toolbar_items[toolbar_ind][0];\n",
       "        var tooltip = mpl.toolbar_items[toolbar_ind][1];\n",
       "        var image = mpl.toolbar_items[toolbar_ind][2];\n",
       "        var method_name = mpl.toolbar_items[toolbar_ind][3];\n",
       "\n",
       "        if (!name) { continue; };\n",
       "\n",
       "        var button = $('<button class=\"btn btn-default\" href=\"#\" title=\"' + name + '\"><i class=\"fa ' + image + ' fa-lg\"></i></button>');\n",
       "        button.click(method_name, toolbar_event);\n",
       "        button.mouseover(tooltip, toolbar_mouse_event);\n",
       "        nav_element.append(button);\n",
       "    }\n",
       "\n",
       "    // Add the status bar.\n",
       "    var status_bar = $('<span class=\"mpl-message\" style=\"text-align:right; float: right;\"/>');\n",
       "    nav_element.append(status_bar);\n",
       "    this.message = status_bar[0];\n",
       "\n",
       "    // Add the close button to the window.\n",
       "    var buttongrp = $('<div class=\"btn-group inline pull-right\"></div>');\n",
       "    var button = $('<button class=\"btn btn-mini btn-primary\" href=\"#\" title=\"Stop Interaction\"><i class=\"fa fa-power-off icon-remove icon-large\"></i></button>');\n",
       "    button.click(function (evt) { fig.handle_close(fig, {}); } );\n",
       "    button.mouseover('Stop Interaction', toolbar_mouse_event);\n",
       "    buttongrp.append(button);\n",
       "    var titlebar = this.root.find($('.ui-dialog-titlebar'));\n",
       "    titlebar.prepend(buttongrp);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._root_extra_style = function(el){\n",
       "    var fig = this\n",
       "    el.on(\"remove\", function(){\n",
       "\tfig.close_ws(fig, {});\n",
       "    });\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._canvas_extra_style = function(el){\n",
       "    // this is important to make the div 'focusable\n",
       "    el.attr('tabindex', 0)\n",
       "    // reach out to IPython and tell the keyboard manager to turn it's self\n",
       "    // off when our div gets focus\n",
       "\n",
       "    // location in version 3\n",
       "    if (IPython.notebook.keyboard_manager) {\n",
       "        IPython.notebook.keyboard_manager.register_events(el);\n",
       "    }\n",
       "    else {\n",
       "        // location in version 2\n",
       "        IPython.keyboard_manager.register_events(el);\n",
       "    }\n",
       "\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._key_event_extra = function(event, name) {\n",
       "    var manager = IPython.notebook.keyboard_manager;\n",
       "    if (!manager)\n",
       "        manager = IPython.keyboard_manager;\n",
       "\n",
       "    // Check for shift+enter\n",
       "    if (event.shiftKey && event.which == 13) {\n",
       "        this.canvas_div.blur();\n",
       "        event.shiftKey = false;\n",
       "        // Send a \"J\" for go to next cell\n",
       "        event.which = 74;\n",
       "        event.keyCode = 74;\n",
       "        manager.command_mode();\n",
       "        manager.handle_keydown(event);\n",
       "    }\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_save = function(fig, msg) {\n",
       "    fig.ondownload(fig, null);\n",
       "}\n",
       "\n",
       "\n",
       "mpl.find_output_cell = function(html_output) {\n",
       "    // Return the cell and output element which can be found *uniquely* in the notebook.\n",
       "    // Note - this is a bit hacky, but it is done because the \"notebook_saving.Notebook\"\n",
       "    // IPython event is triggered only after the cells have been serialised, which for\n",
       "    // our purposes (turning an active figure into a static one), is too late.\n",
       "    var cells = IPython.notebook.get_cells();\n",
       "    var ncells = cells.length;\n",
       "    for (var i=0; i<ncells; i++) {\n",
       "        var cell = cells[i];\n",
       "        if (cell.cell_type === 'code'){\n",
       "            for (var j=0; j<cell.output_area.outputs.length; j++) {\n",
       "                var data = cell.output_area.outputs[j];\n",
       "                if (data.data) {\n",
       "                    // IPython >= 3 moved mimebundle to data attribute of output\n",
       "                    data = data.data;\n",
       "                }\n",
       "                if (data['text/html'] == html_output) {\n",
       "                    return [cell, data, j];\n",
       "                }\n",
       "            }\n",
       "        }\n",
       "    }\n",
       "}\n",
       "\n",
       "// Register the function which deals with the matplotlib target/channel.\n",
       "// The kernel may be null if the page has been refreshed.\n",
       "if (IPython.notebook.kernel != null) {\n",
       "    IPython.notebook.kernel.comm_manager.register_target('matplotlib', mpl.mpl_figure_comm);\n",
       "}\n"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" width=\"639.4444613839378\">"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig1, axes1 = plt.subplots(2, 2, sharex='all', sharey='all')\n",
    "\n",
    "map_dict = {'Amusement': 0,'Anger': 1,'Awe': 2,'Content': 3,'Disgust': 4,'Excitement': 5,'Fear': 6,'Sad': 7}\n",
    "rev = dict((v,k) for k,v in map_dict.items())\n",
    "\n",
    "img_list = []\n",
    "label_list = []\n",
    "\n",
    "for i in range(len(pred_set)):\n",
    "    with torch.no_grad():\n",
    "        output = exp1.net(pred_set[i][np.newaxis].to(exp1.net.device))[0]\n",
    "    values, indices = torch.max(output, 0)\n",
    "    label = rev[int(indices)]\n",
    "    img_list.append(pred_set[i])\n",
    "    label_list.append(label)\n",
    "    if i == 0 or i == 1:\n",
    "        axes1[0][i].set_title(label)\n",
    "        myimshow(pred_set[i], ax=axes1[0][i])\n",
    "    binary = [int(x) for x in list('{0:0b}'.format(i))]\n",
    "    if i > 1:\n",
    "        axes1[binary[0]][binary[1]].set_title(label)\n",
    "        myimshow(pred_set[i], ax=axes1[binary[0]][binary[1]])\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}