Food And Non-Food Classifier.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Food  Food  Food  Food  Food  Food Non-Food Non-Food Non-Food Non-Food Non-Food  Food Non-Food Non-Food  Food Non-Food\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import torch\n",
    "from torch import nn\n",
    "import torchvision\n",
    "from torch.autograd import Variable\n",
    "from torch.utils.data import DataLoader, Sampler\n",
    "from torchvision import datasets\n",
    "from torchvision.transforms import transforms\n",
    "from torch.optim import Adam\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import PIL\n",
    "from pathlib import Path\n",
    "\n",
    "num_classes = 2\n",
    "batch_size = 16\n",
    "num_of_workers = 2\n",
    "\n",
    "DATA_PATH_TRAIN = Path('/home/abhishek/Images/train')\n",
    "DATA_PATH_TEST = Path('/home/abhishek/Images/test')\n",
    "\n",
    "trans = transforms.Compose([\n",
    "    transforms.RandomHorizontalFlip(),\n",
    "    transforms.Resize(32),\n",
    "    transforms.CenterCrop(32),\n",
    "    transforms.ToTensor(),\n",
    "    transforms.Normalize((0.5, 0.5, 0.5),(0.5, 0.5, 0.5))\n",
    "    ])\n",
    "\n",
    "train_dataset = datasets.ImageFolder(root=DATA_PATH_TRAIN, transform=trans)\n",
    "train_loader = DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True, num_workers=num_of_workers)\n",
    "\n",
    "test_dataset = datasets.ImageFolder(root=DATA_PATH_TEST,transform=trans)\n",
    "test_loader = DataLoader(dataset=test_dataset, batch_size=batch_size, shuffle=False, num_workers=num_of_workers)\n",
    "\n",
    "classes=('Food','Non-Food')\n",
    "\n",
    "\n",
    "\n",
    "# functions to show an image\n",
    "def imshow(img):\n",
    "    img = img / 2 + 0.5     # unnormalize\n",
    "    npimg = img.numpy()\n",
    "    plt.imshow(np.transpose(npimg, (1, 2, 0)))\n",
    "\n",
    "# get some random training images\n",
    "dataiter = iter(train_loader)\n",
    "images, labels = dataiter.next()\n",
    "\n",
    "# show images\n",
    "imshow(torchvision.utils.make_grid(images))\n",
    "# print labels\n",
    "print(' '.join('%5s' %   classes[labels[j]] for j in range(16)))\n",
    "   \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000\n"
     ]
    }
   ],
   "source": [
    "print(len(test_dataset))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Net(\n",
      "  (conv1): Conv2d(1, 2, kernel_size=(5, 5), stride=(1, 1))\n",
      "  (pool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  (conv2): Conv2d(6, 16, kernel_size=(5, 5), stride=(1, 1))\n",
      "  (fc1): Linear(in_features=400, out_features=120, bias=True)\n",
      "  (fc2): Linear(in_features=120, out_features=84, bias=True)\n",
      "  (fc3): Linear(in_features=84, out_features=2, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "\n",
    "\n",
    "class Net(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Net, self).__init__()\n",
    "        self.conv1 = nn.Conv2d(1, 2, 5)\n",
    "        self.pool = nn.MaxPool2d(2, 2)\n",
    "        self.conv2 = nn.Conv2d(6, 16, 5)\n",
    "        # an affine operation: y = Wx + b\n",
    "        self.fc1 = nn.Linear(16 * 5 * 5, 120)    # 5*5 from image dimension\n",
    "        self.fc2 = nn.Linear(120, 84)\n",
    "        self.fc3 = nn.Linear(84, 2)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.pool(F.relu(self.conv1(x)))\n",
    "        x = self.pool(F.relu(self.conv2(x)))\n",
    "        x = x.view(-1, 16 * 5 * 5)              #for flattening images\n",
    "        x = F.relu(self.fc1(x))\n",
    "        x = F.relu(self.fc2(x))\n",
    "        x = self.fc3(x)\n",
    "        return x\n",
    "\n",
    "\n",
    "net = Net()\n",
    "print(net)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.optim as optim\n",
    "\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1,    10] loss: 0.003\n",
      "[1,    20] loss: 0.003\n",
      "[1,    30] loss: 0.003\n",
      "[1,    40] loss: 0.003\n",
      "[1,    50] loss: 0.004\n",
      "[1,    60] loss: 0.003\n",
      "[1,    70] loss: 0.003\n",
      "[1,    80] loss: 0.003\n",
      "[1,    90] loss: 0.003\n",
      "[1,   100] loss: 0.003\n",
      "[1,   110] loss: 0.003\n",
      "[1,   120] loss: 0.003\n",
      "[1,   130] loss: 0.003\n",
      "[1,   140] loss: 0.003\n",
      "[1,   150] loss: 0.003\n",
      "[1,   160] loss: 0.003\n",
      "[1,   170] loss: 0.003\n",
      "[1,   180] loss: 0.003\n",
      "[2,    10] loss: 0.003\n",
      "[2,    20] loss: 0.003\n",
      "[2,    30] loss: 0.003\n",
      "[2,    40] loss: 0.003\n",
      "[2,    50] loss: 0.003\n",
      "[2,    60] loss: 0.003\n",
      "[2,    70] loss: 0.003\n",
      "[2,    80] loss: 0.003\n",
      "[2,    90] loss: 0.003\n",
      "[2,   100] loss: 0.003\n",
      "[2,   110] loss: 0.003\n",
      "[2,   120] loss: 0.003\n",
      "[2,   130] loss: 0.003\n",
      "[2,   140] loss: 0.003\n",
      "[2,   150] loss: 0.003\n",
      "[2,   160] loss: 0.003\n",
      "[2,   170] loss: 0.003\n",
      "[2,   180] loss: 0.003\n",
      "[3,    10] loss: 0.003\n",
      "[3,    20] loss: 0.003\n",
      "[3,    30] loss: 0.003\n",
      "[3,    40] loss: 0.003\n",
      "[3,    50] loss: 0.003\n",
      "[3,    60] loss: 0.003\n",
      "[3,    70] loss: 0.003\n",
      "[3,    80] loss: 0.003\n",
      "[3,    90] loss: 0.003\n",
      "[3,   100] loss: 0.003\n",
      "[3,   110] loss: 0.003\n",
      "[3,   120] loss: 0.003\n",
      "[3,   130] loss: 0.003\n",
      "[3,   140] loss: 0.003\n",
      "[3,   150] loss: 0.003\n",
      "[3,   160] loss: 0.003\n",
      "[3,   170] loss: 0.003\n",
      "[3,   180] loss: 0.003\n",
      "[4,    10] loss: 0.003\n",
      "[4,    20] loss: 0.003\n",
      "[4,    30] loss: 0.003\n",
      "[4,    40] loss: 0.002\n",
      "[4,    50] loss: 0.003\n",
      "[4,    60] loss: 0.002\n",
      "[4,    70] loss: 0.003\n",
      "[4,    80] loss: 0.003\n",
      "[4,    90] loss: 0.002\n",
      "[4,   100] loss: 0.002\n",
      "[4,   110] loss: 0.003\n",
      "[4,   120] loss: 0.002\n",
      "[4,   130] loss: 0.003\n",
      "[4,   140] loss: 0.002\n",
      "[4,   150] loss: 0.003\n",
      "[4,   160] loss: 0.003\n",
      "[4,   170] loss: 0.002\n",
      "[4,   180] loss: 0.002\n",
      "[5,    10] loss: 0.003\n",
      "[5,    20] loss: 0.003\n",
      "[5,    30] loss: 0.002\n",
      "[5,    40] loss: 0.003\n",
      "[5,    50] loss: 0.002\n",
      "[5,    60] loss: 0.002\n",
      "[5,    70] loss: 0.002\n",
      "[5,    80] loss: 0.002\n",
      "[5,    90] loss: 0.002\n",
      "[5,   100] loss: 0.003\n",
      "[5,   110] loss: 0.003\n",
      "[5,   120] loss: 0.003\n",
      "[5,   130] loss: 0.003\n",
      "[5,   140] loss: 0.002\n",
      "[5,   150] loss: 0.002\n",
      "[5,   160] loss: 0.002\n",
      "[5,   170] loss: 0.003\n",
      "[5,   180] loss: 0.002\n",
      "[6,    10] loss: 0.002\n",
      "[6,    20] loss: 0.002\n",
      "[6,    30] loss: 0.002\n",
      "[6,    40] loss: 0.002\n",
      "[6,    50] loss: 0.002\n",
      "[6,    60] loss: 0.003\n",
      "[6,    70] loss: 0.003\n",
      "[6,    80] loss: 0.002\n",
      "[6,    90] loss: 0.002\n",
      "[6,   100] loss: 0.002\n",
      "[6,   110] loss: 0.002\n",
      "[6,   120] loss: 0.003\n",
      "[6,   130] loss: 0.003\n",
      "[6,   140] loss: 0.002\n",
      "[6,   150] loss: 0.002\n",
      "[6,   160] loss: 0.003\n",
      "[6,   170] loss: 0.003\n",
      "[6,   180] loss: 0.002\n",
      "[7,    10] loss: 0.002\n",
      "[7,    20] loss: 0.002\n",
      "[7,    30] loss: 0.002\n",
      "[7,    40] loss: 0.002\n",
      "[7,    50] loss: 0.002\n",
      "[7,    60] loss: 0.003\n",
      "[7,    70] loss: 0.002\n",
      "[7,    80] loss: 0.002\n",
      "[7,    90] loss: 0.002\n",
      "[7,   100] loss: 0.002\n",
      "[7,   110] loss: 0.002\n",
      "[7,   120] loss: 0.002\n",
      "[7,   130] loss: 0.002\n",
      "[7,   140] loss: 0.003\n",
      "[7,   150] loss: 0.003\n",
      "[7,   160] loss: 0.003\n",
      "[7,   170] loss: 0.002\n",
      "[7,   180] loss: 0.002\n",
      "[8,    10] loss: 0.002\n",
      "[8,    20] loss: 0.002\n",
      "[8,    30] loss: 0.003\n",
      "[8,    40] loss: 0.002\n",
      "[8,    50] loss: 0.002\n",
      "[8,    60] loss: 0.002\n",
      "[8,    70] loss: 0.002\n",
      "[8,    80] loss: 0.002\n",
      "[8,    90] loss: 0.003\n",
      "[8,   100] loss: 0.003\n",
      "[8,   110] loss: 0.003\n",
      "[8,   120] loss: 0.002\n",
      "[8,   130] loss: 0.002\n",
      "[8,   140] loss: 0.002\n",
      "[8,   150] loss: 0.002\n",
      "[8,   160] loss: 0.002\n",
      "[8,   170] loss: 0.002\n",
      "[8,   180] loss: 0.002\n",
      "[9,    10] loss: 0.002\n",
      "[9,    20] loss: 0.002\n",
      "[9,    30] loss: 0.002\n",
      "[9,    40] loss: 0.002\n",
      "[9,    50] loss: 0.002\n",
      "[9,    60] loss: 0.002\n",
      "[9,    70] loss: 0.002\n",
      "[9,    80] loss: 0.002\n",
      "[9,    90] loss: 0.002\n",
      "[9,   100] loss: 0.002\n",
      "[9,   110] loss: 0.003\n",
      "[9,   120] loss: 0.002\n",
      "[9,   130] loss: 0.002\n",
      "[9,   140] loss: 0.002\n",
      "[9,   150] loss: 0.002\n",
      "[9,   160] loss: 0.003\n",
      "[9,   170] loss: 0.002\n",
      "[9,   180] loss: 0.002\n",
      "[10,    10] loss: 0.002\n",
      "[10,    20] loss: 0.002\n",
      "[10,    30] loss: 0.002\n",
      "[10,    40] loss: 0.002\n",
      "[10,    50] loss: 0.002\n",
      "[10,    60] loss: 0.002\n",
      "[10,    70] loss: 0.002\n",
      "[10,    80] loss: 0.002\n",
      "[10,    90] loss: 0.002\n",
      "[10,   100] loss: 0.002\n",
      "[10,   110] loss: 0.002\n",
      "[10,   120] loss: 0.002\n",
      "[10,   130] loss: 0.002\n",
      "[10,   140] loss: 0.002\n",
      "[10,   150] loss: 0.002\n",
      "[10,   160] loss: 0.002\n",
      "[10,   170] loss: 0.002\n",
      "[10,   180] loss: 0.002\n",
      "[11,    10] loss: 0.002\n",
      "[11,    20] loss: 0.002\n",
      "[11,    30] loss: 0.002\n",
      "[11,    40] loss: 0.002\n",
      "[11,    50] loss: 0.002\n",
      "[11,    60] loss: 0.002\n",
      "[11,    70] loss: 0.002\n",
      "[11,    80] loss: 0.002\n",
      "[11,    90] loss: 0.002\n",
      "[11,   100] loss: 0.002\n",
      "[11,   110] loss: 0.002\n",
      "[11,   120] loss: 0.002\n",
      "[11,   130] loss: 0.003\n",
      "[11,   140] loss: 0.002\n",
      "[11,   150] loss: 0.002\n",
      "[11,   160] loss: 0.002\n",
      "[11,   170] loss: 0.002\n",
      "[11,   180] loss: 0.002\n",
      "[12,    10] loss: 0.002\n",
      "[12,    20] loss: 0.002\n",
      "[12,    30] loss: 0.002\n",
      "[12,    40] loss: 0.002\n",
      "[12,    50] loss: 0.002\n",
      "[12,    60] loss: 0.002\n",
      "[12,    70] loss: 0.002\n",
      "[12,    80] loss: 0.002\n",
      "[12,    90] loss: 0.002\n",
      "[12,   100] loss: 0.002\n",
      "[12,   110] loss: 0.001\n",
      "[12,   120] loss: 0.002\n",
      "[12,   130] loss: 0.002\n",
      "[12,   140] loss: 0.002\n",
      "[12,   150] loss: 0.002\n",
      "[12,   160] loss: 0.002\n",
      "[12,   170] loss: 0.002\n",
      "[12,   180] loss: 0.002\n",
      "[13,    10] loss: 0.002\n",
      "[13,    20] loss: 0.002\n",
      "[13,    30] loss: 0.002\n",
      "[13,    40] loss: 0.002\n",
      "[13,    50] loss: 0.002\n",
      "[13,    60] loss: 0.002\n",
      "[13,    70] loss: 0.002\n",
      "[13,    80] loss: 0.002\n",
      "[13,    90] loss: 0.002\n",
      "[13,   100] loss: 0.002\n",
      "[13,   110] loss: 0.002\n",
      "[13,   120] loss: 0.002\n",
      "[13,   130] loss: 0.002\n",
      "[13,   140] loss: 0.002\n",
      "[13,   150] loss: 0.002\n",
      "[13,   160] loss: 0.002\n",
      "[13,   170] loss: 0.002\n",
      "[13,   180] loss: 0.002\n",
      "[14,    10] loss: 0.002\n",
      "[14,    20] loss: 0.002\n",
      "[14,    30] loss: 0.002\n",
      "[14,    40] loss: 0.002\n",
      "[14,    50] loss: 0.002\n",
      "[14,    60] loss: 0.002\n",
      "[14,    70] loss: 0.002\n",
      "[14,    80] loss: 0.002\n",
      "[14,    90] loss: 0.002\n",
      "[14,   100] loss: 0.002\n",
      "[14,   110] loss: 0.002\n",
      "[14,   120] loss: 0.002\n",
      "[14,   130] loss: 0.002\n",
      "[14,   140] loss: 0.002\n",
      "[14,   150] loss: 0.002\n",
      "[14,   160] loss: 0.002\n",
      "[14,   170] loss: 0.002\n",
      "[14,   180] loss: 0.002\n",
      "[15,    10] loss: 0.002\n",
      "[15,    20] loss: 0.002\n",
      "[15,    30] loss: 0.002\n",
      "[15,    40] loss: 0.002\n",
      "[15,    50] loss: 0.002\n",
      "[15,    60] loss: 0.002\n",
      "[15,    70] loss: 0.001\n",
      "[15,    80] loss: 0.002\n",
      "[15,    90] loss: 0.002\n",
      "[15,   100] loss: 0.002\n",
      "[15,   110] loss: 0.002\n",
      "[15,   120] loss: 0.002\n",
      "[15,   130] loss: 0.002\n",
      "[15,   140] loss: 0.002\n",
      "[15,   150] loss: 0.002\n",
      "[15,   160] loss: 0.002\n",
      "[15,   170] loss: 0.002\n",
      "[15,   180] loss: 0.002\n",
      "[16,    10] loss: 0.002\n",
      "[16,    20] loss: 0.002\n",
      "[16,    30] loss: 0.002\n",
      "[16,    40] loss: 0.002\n",
      "[16,    50] loss: 0.002\n",
      "[16,    60] loss: 0.002\n",
      "[16,    70] loss: 0.002\n",
      "[16,    80] loss: 0.002\n",
      "[16,    90] loss: 0.002\n",
      "[16,   100] loss: 0.002\n",
      "[16,   110] loss: 0.002\n",
      "[16,   120] loss: 0.002\n",
      "[16,   130] loss: 0.002\n",
      "[16,   140] loss: 0.002\n",
      "[16,   150] loss: 0.002\n",
      "[16,   160] loss: 0.002\n",
      "[16,   170] loss: 0.002\n",
      "[16,   180] loss: 0.002\n",
      "[17,    10] loss: 0.002\n",
      "[17,    20] loss: 0.002\n",
      "[17,    30] loss: 0.002\n",
      "[17,    40] loss: 0.002\n",
      "[17,    50] loss: 0.002\n",
      "[17,    60] loss: 0.002\n",
      "[17,    70] loss: 0.001\n",
      "[17,    80] loss: 0.002\n",
      "[17,    90] loss: 0.002\n",
      "[17,   100] loss: 0.002\n",
      "[17,   110] loss: 0.002\n",
      "[17,   120] loss: 0.002\n",
      "[17,   130] loss: 0.002\n",
      "[17,   140] loss: 0.002\n",
      "[17,   150] loss: 0.002\n",
      "[17,   160] loss: 0.002\n",
      "[17,   170] loss: 0.002\n",
      "[17,   180] loss: 0.002\n",
      "[18,    10] loss: 0.002\n",
      "[18,    20] loss: 0.001\n",
      "[18,    30] loss: 0.002\n",
      "[18,    40] loss: 0.002\n",
      "[18,    50] loss: 0.002\n",
      "[18,    60] loss: 0.002\n",
      "[18,    70] loss: 0.002\n",
      "[18,    80] loss: 0.002\n",
      "[18,    90] loss: 0.001\n",
      "[18,   100] loss: 0.001\n",
      "[18,   110] loss: 0.002\n",
      "[18,   120] loss: 0.002\n",
      "[18,   130] loss: 0.002\n",
      "[18,   140] loss: 0.002\n",
      "[18,   150] loss: 0.002\n",
      "[18,   160] loss: 0.002\n",
      "[18,   170] loss: 0.002\n",
      "[18,   180] loss: 0.002\n",
      "[19,    10] loss: 0.002\n",
      "[19,    20] loss: 0.002\n",
      "[19,    30] loss: 0.001\n",
      "[19,    40] loss: 0.002\n",
      "[19,    50] loss: 0.002\n",
      "[19,    60] loss: 0.002\n",
      "[19,    70] loss: 0.002\n",
      "[19,    80] loss: 0.002\n",
      "[19,    90] loss: 0.002\n",
      "[19,   100] loss: 0.002\n",
      "[19,   110] loss: 0.002\n",
      "[19,   120] loss: 0.001\n",
      "[19,   130] loss: 0.002\n",
      "[19,   140] loss: 0.002\n",
      "[19,   150] loss: 0.001\n",
      "[19,   160] loss: 0.002\n",
      "[19,   170] loss: 0.002\n",
      "[19,   180] loss: 0.002\n",
      "[20,    10] loss: 0.002\n",
      "[20,    20] loss: 0.001\n",
      "[20,    30] loss: 0.002\n",
      "[20,    40] loss: 0.002\n",
      "[20,    50] loss: 0.002\n",
      "[20,    60] loss: 0.002\n",
      "[20,    70] loss: 0.002\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[20,    80] loss: 0.002\n",
      "[20,    90] loss: 0.002\n",
      "[20,   100] loss: 0.002\n",
      "[20,   110] loss: 0.002\n",
      "[20,   120] loss: 0.002\n",
      "[20,   130] loss: 0.001\n",
      "[20,   140] loss: 0.002\n",
      "[20,   150] loss: 0.001\n",
      "[20,   160] loss: 0.002\n",
      "[20,   170] loss: 0.002\n",
      "[20,   180] loss: 0.002\n",
      "[21,    10] loss: 0.002\n",
      "[21,    20] loss: 0.002\n",
      "[21,    30] loss: 0.001\n",
      "[21,    40] loss: 0.002\n",
      "[21,    50] loss: 0.001\n",
      "[21,    60] loss: 0.002\n",
      "[21,    70] loss: 0.001\n",
      "[21,    80] loss: 0.001\n",
      "[21,    90] loss: 0.002\n",
      "[21,   100] loss: 0.002\n",
      "[21,   110] loss: 0.001\n",
      "[21,   120] loss: 0.002\n",
      "[21,   130] loss: 0.002\n",
      "[21,   140] loss: 0.002\n",
      "[21,   150] loss: 0.002\n",
      "[21,   160] loss: 0.002\n",
      "[21,   170] loss: 0.002\n",
      "[21,   180] loss: 0.001\n",
      "[22,    10] loss: 0.001\n",
      "[22,    20] loss: 0.001\n",
      "[22,    30] loss: 0.002\n",
      "[22,    40] loss: 0.002\n",
      "[22,    50] loss: 0.002\n",
      "[22,    60] loss: 0.002\n",
      "[22,    70] loss: 0.001\n",
      "[22,    80] loss: 0.002\n",
      "[22,    90] loss: 0.001\n",
      "[22,   100] loss: 0.002\n",
      "[22,   110] loss: 0.002\n",
      "[22,   120] loss: 0.002\n",
      "[22,   130] loss: 0.002\n",
      "[22,   140] loss: 0.001\n",
      "[22,   150] loss: 0.002\n",
      "[22,   160] loss: 0.002\n",
      "[22,   170] loss: 0.002\n",
      "[22,   180] loss: 0.002\n",
      "[23,    10] loss: 0.001\n",
      "[23,    20] loss: 0.002\n",
      "[23,    30] loss: 0.001\n",
      "[23,    40] loss: 0.002\n",
      "[23,    50] loss: 0.002\n",
      "[23,    60] loss: 0.002\n",
      "[23,    70] loss: 0.002\n",
      "[23,    80] loss: 0.001\n",
      "[23,    90] loss: 0.001\n",
      "[23,   100] loss: 0.002\n",
      "[23,   110] loss: 0.002\n",
      "[23,   120] loss: 0.002\n",
      "[23,   130] loss: 0.002\n",
      "[23,   140] loss: 0.002\n",
      "[23,   150] loss: 0.002\n",
      "[23,   160] loss: 0.001\n",
      "[23,   170] loss: 0.002\n",
      "[23,   180] loss: 0.002\n",
      "[24,    10] loss: 0.002\n",
      "[24,    20] loss: 0.001\n",
      "[24,    30] loss: 0.002\n",
      "[24,    40] loss: 0.002\n",
      "[24,    50] loss: 0.002\n",
      "[24,    60] loss: 0.002\n",
      "[24,    70] loss: 0.001\n",
      "[24,    80] loss: 0.001\n",
      "[24,    90] loss: 0.002\n",
      "[24,   100] loss: 0.001\n",
      "[24,   110] loss: 0.001\n",
      "[24,   120] loss: 0.002\n",
      "[24,   130] loss: 0.002\n",
      "[24,   140] loss: 0.001\n",
      "[24,   150] loss: 0.002\n",
      "[24,   160] loss: 0.001\n",
      "[24,   170] loss: 0.002\n",
      "[24,   180] loss: 0.002\n",
      "[25,    10] loss: 0.002\n",
      "[25,    20] loss: 0.001\n",
      "[25,    30] loss: 0.002\n",
      "[25,    40] loss: 0.002\n",
      "[25,    50] loss: 0.002\n",
      "[25,    60] loss: 0.002\n",
      "[25,    70] loss: 0.002\n",
      "[25,    80] loss: 0.001\n",
      "[25,    90] loss: 0.002\n",
      "[25,   100] loss: 0.002\n",
      "[25,   110] loss: 0.001\n",
      "[25,   120] loss: 0.002\n",
      "[25,   130] loss: 0.001\n",
      "[25,   140] loss: 0.002\n",
      "[25,   150] loss: 0.001\n",
      "[25,   160] loss: 0.001\n",
      "[25,   170] loss: 0.001\n",
      "[25,   180] loss: 0.002\n",
      "[26,    10] loss: 0.001\n",
      "[26,    20] loss: 0.002\n",
      "[26,    30] loss: 0.002\n",
      "[26,    40] loss: 0.002\n",
      "[26,    50] loss: 0.002\n",
      "[26,    60] loss: 0.001\n",
      "[26,    70] loss: 0.002\n",
      "[26,    80] loss: 0.002\n",
      "[26,    90] loss: 0.002\n",
      "[26,   100] loss: 0.001\n",
      "[26,   110] loss: 0.001\n",
      "[26,   120] loss: 0.002\n",
      "[26,   130] loss: 0.001\n",
      "[26,   140] loss: 0.002\n",
      "[26,   150] loss: 0.002\n",
      "[26,   160] loss: 0.001\n",
      "[26,   170] loss: 0.001\n",
      "[26,   180] loss: 0.001\n",
      "[27,    10] loss: 0.002\n",
      "[27,    20] loss: 0.002\n",
      "[27,    30] loss: 0.002\n",
      "[27,    40] loss: 0.001\n",
      "[27,    50] loss: 0.002\n",
      "[27,    60] loss: 0.001\n",
      "[27,    70] loss: 0.001\n",
      "[27,    80] loss: 0.001\n",
      "[27,    90] loss: 0.001\n",
      "[27,   100] loss: 0.001\n",
      "[27,   110] loss: 0.002\n",
      "[27,   120] loss: 0.002\n",
      "[27,   130] loss: 0.002\n",
      "[27,   140] loss: 0.002\n",
      "[27,   150] loss: 0.002\n",
      "[27,   160] loss: 0.001\n",
      "[27,   170] loss: 0.002\n",
      "[27,   180] loss: 0.002\n",
      "[28,    10] loss: 0.002\n",
      "[28,    20] loss: 0.002\n",
      "[28,    30] loss: 0.001\n",
      "[28,    40] loss: 0.002\n",
      "[28,    50] loss: 0.002\n",
      "[28,    60] loss: 0.002\n",
      "[28,    70] loss: 0.001\n",
      "[28,    80] loss: 0.002\n",
      "[28,    90] loss: 0.002\n",
      "[28,   100] loss: 0.001\n",
      "[28,   110] loss: 0.001\n",
      "[28,   120] loss: 0.001\n",
      "[28,   130] loss: 0.001\n",
      "[28,   140] loss: 0.001\n",
      "[28,   150] loss: 0.002\n",
      "[28,   160] loss: 0.001\n",
      "[28,   170] loss: 0.001\n",
      "[28,   180] loss: 0.002\n",
      "[29,    10] loss: 0.001\n",
      "[29,    20] loss: 0.002\n",
      "[29,    30] loss: 0.002\n",
      "[29,    40] loss: 0.002\n",
      "[29,    50] loss: 0.001\n",
      "[29,    60] loss: 0.001\n",
      "[29,    70] loss: 0.001\n",
      "[29,    80] loss: 0.002\n",
      "[29,    90] loss: 0.001\n",
      "[29,   100] loss: 0.001\n",
      "[29,   110] loss: 0.001\n",
      "[29,   120] loss: 0.002\n",
      "[29,   130] loss: 0.002\n",
      "[29,   140] loss: 0.001\n",
      "[29,   150] loss: 0.002\n",
      "[29,   160] loss: 0.001\n",
      "[29,   170] loss: 0.002\n",
      "[29,   180] loss: 0.002\n",
      "[30,    10] loss: 0.002\n",
      "[30,    20] loss: 0.001\n",
      "[30,    30] loss: 0.002\n",
      "[30,    40] loss: 0.001\n",
      "[30,    50] loss: 0.001\n",
      "[30,    60] loss: 0.002\n",
      "[30,    70] loss: 0.002\n",
      "[30,    80] loss: 0.001\n",
      "[30,    90] loss: 0.002\n",
      "[30,   100] loss: 0.001\n",
      "[30,   110] loss: 0.001\n",
      "[30,   120] loss: 0.001\n",
      "[30,   130] loss: 0.001\n",
      "[30,   140] loss: 0.002\n",
      "[30,   150] loss: 0.002\n",
      "[30,   160] loss: 0.001\n",
      "[30,   170] loss: 0.001\n",
      "[30,   180] loss: 0.002\n",
      "[31,    10] loss: 0.001\n",
      "[31,    20] loss: 0.002\n",
      "[31,    30] loss: 0.001\n",
      "[31,    40] loss: 0.002\n",
      "[31,    50] loss: 0.001\n",
      "[31,    60] loss: 0.002\n",
      "[31,    70] loss: 0.002\n",
      "[31,    80] loss: 0.001\n",
      "[31,    90] loss: 0.002\n",
      "[31,   100] loss: 0.002\n",
      "[31,   110] loss: 0.001\n",
      "[31,   120] loss: 0.001\n",
      "[31,   130] loss: 0.001\n",
      "[31,   140] loss: 0.002\n",
      "[31,   150] loss: 0.001\n",
      "[31,   160] loss: 0.002\n",
      "[31,   170] loss: 0.002\n",
      "[31,   180] loss: 0.002\n",
      "[32,    10] loss: 0.001\n",
      "[32,    20] loss: 0.001\n",
      "[32,    30] loss: 0.002\n",
      "[32,    40] loss: 0.001\n",
      "[32,    50] loss: 0.001\n",
      "[32,    60] loss: 0.001\n",
      "[32,    70] loss: 0.002\n",
      "[32,    80] loss: 0.001\n",
      "[32,    90] loss: 0.002\n",
      "[32,   100] loss: 0.001\n",
      "[32,   110] loss: 0.001\n",
      "[32,   120] loss: 0.001\n",
      "[32,   130] loss: 0.001\n",
      "[32,   140] loss: 0.002\n",
      "[32,   150] loss: 0.001\n",
      "[32,   160] loss: 0.001\n",
      "[32,   170] loss: 0.002\n",
      "[32,   180] loss: 0.001\n",
      "[33,    10] loss: 0.001\n",
      "[33,    20] loss: 0.002\n",
      "[33,    30] loss: 0.001\n",
      "[33,    40] loss: 0.002\n",
      "[33,    50] loss: 0.002\n",
      "[33,    60] loss: 0.002\n",
      "[33,    70] loss: 0.001\n",
      "[33,    80] loss: 0.001\n",
      "[33,    90] loss: 0.001\n",
      "[33,   100] loss: 0.001\n",
      "[33,   110] loss: 0.002\n",
      "[33,   120] loss: 0.001\n",
      "[33,   130] loss: 0.001\n",
      "[33,   140] loss: 0.001\n",
      "[33,   150] loss: 0.001\n",
      "[33,   160] loss: 0.001\n",
      "[33,   170] loss: 0.001\n",
      "[33,   180] loss: 0.001\n",
      "[34,    10] loss: 0.001\n",
      "[34,    20] loss: 0.001\n",
      "[34,    30] loss: 0.001\n",
      "[34,    40] loss: 0.001\n",
      "[34,    50] loss: 0.001\n",
      "[34,    60] loss: 0.001\n",
      "[34,    70] loss: 0.001\n",
      "[34,    80] loss: 0.002\n",
      "[34,    90] loss: 0.001\n",
      "[34,   100] loss: 0.002\n",
      "[34,   110] loss: 0.002\n",
      "[34,   120] loss: 0.002\n",
      "[34,   130] loss: 0.001\n",
      "[34,   140] loss: 0.001\n",
      "[34,   150] loss: 0.001\n",
      "[34,   160] loss: 0.002\n",
      "[34,   170] loss: 0.001\n",
      "[34,   180] loss: 0.002\n",
      "[35,    10] loss: 0.001\n",
      "[35,    20] loss: 0.001\n",
      "[35,    30] loss: 0.002\n",
      "[35,    40] loss: 0.001\n",
      "[35,    50] loss: 0.001\n",
      "[35,    60] loss: 0.002\n",
      "[35,    70] loss: 0.002\n",
      "[35,    80] loss: 0.001\n",
      "[35,    90] loss: 0.001\n",
      "[35,   100] loss: 0.001\n",
      "[35,   110] loss: 0.001\n",
      "[35,   120] loss: 0.001\n",
      "[35,   130] loss: 0.002\n",
      "[35,   140] loss: 0.001\n",
      "[35,   150] loss: 0.001\n",
      "[35,   160] loss: 0.002\n",
      "[35,   170] loss: 0.002\n",
      "[35,   180] loss: 0.001\n",
      "[36,    10] loss: 0.001\n",
      "[36,    20] loss: 0.002\n",
      "[36,    30] loss: 0.001\n",
      "[36,    40] loss: 0.001\n",
      "[36,    50] loss: 0.001\n",
      "[36,    60] loss: 0.002\n",
      "[36,    70] loss: 0.001\n",
      "[36,    80] loss: 0.001\n",
      "[36,    90] loss: 0.001\n",
      "[36,   100] loss: 0.001\n",
      "[36,   110] loss: 0.002\n",
      "[36,   120] loss: 0.001\n",
      "[36,   130] loss: 0.002\n",
      "[36,   140] loss: 0.001\n",
      "[36,   150] loss: 0.001\n",
      "[36,   160] loss: 0.001\n",
      "[36,   170] loss: 0.001\n",
      "[36,   180] loss: 0.001\n",
      "[37,    10] loss: 0.001\n",
      "[37,    20] loss: 0.001\n",
      "[37,    30] loss: 0.001\n",
      "[37,    40] loss: 0.001\n",
      "[37,    50] loss: 0.001\n",
      "[37,    60] loss: 0.001\n",
      "[37,    70] loss: 0.001\n",
      "[37,    80] loss: 0.002\n",
      "[37,    90] loss: 0.001\n",
      "[37,   100] loss: 0.002\n",
      "[37,   110] loss: 0.001\n",
      "[37,   120] loss: 0.002\n",
      "[37,   130] loss: 0.001\n",
      "[37,   140] loss: 0.002\n",
      "[37,   150] loss: 0.001\n",
      "[37,   160] loss: 0.001\n",
      "[37,   170] loss: 0.001\n",
      "[37,   180] loss: 0.001\n",
      "[38,    10] loss: 0.002\n",
      "[38,    20] loss: 0.002\n",
      "[38,    30] loss: 0.002\n",
      "[38,    40] loss: 0.001\n",
      "[38,    50] loss: 0.001\n",
      "[38,    60] loss: 0.002\n",
      "[38,    70] loss: 0.001\n",
      "[38,    80] loss: 0.001\n",
      "[38,    90] loss: 0.001\n",
      "[38,   100] loss: 0.001\n",
      "[38,   110] loss: 0.001\n",
      "[38,   120] loss: 0.001\n",
      "[38,   130] loss: 0.001\n",
      "[38,   140] loss: 0.001\n",
      "[38,   150] loss: 0.001\n",
      "[38,   160] loss: 0.001\n",
      "[38,   170] loss: 0.001\n",
      "[38,   180] loss: 0.001\n",
      "[39,    10] loss: 0.002\n",
      "[39,    20] loss: 0.002\n",
      "[39,    30] loss: 0.001\n",
      "[39,    40] loss: 0.001\n",
      "[39,    50] loss: 0.001\n",
      "[39,    60] loss: 0.001\n",
      "[39,    70] loss: 0.001\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[39,    80] loss: 0.001\n",
      "[39,    90] loss: 0.001\n",
      "[39,   100] loss: 0.001\n",
      "[39,   110] loss: 0.001\n",
      "[39,   120] loss: 0.002\n",
      "[39,   130] loss: 0.001\n",
      "[39,   140] loss: 0.001\n",
      "[39,   150] loss: 0.001\n",
      "[39,   160] loss: 0.001\n",
      "[39,   170] loss: 0.002\n",
      "[39,   180] loss: 0.001\n",
      "[40,    10] loss: 0.002\n",
      "[40,    20] loss: 0.001\n",
      "[40,    30] loss: 0.001\n",
      "[40,    40] loss: 0.001\n",
      "[40,    50] loss: 0.001\n",
      "[40,    60] loss: 0.001\n",
      "[40,    70] loss: 0.001\n",
      "[40,    80] loss: 0.001\n",
      "[40,    90] loss: 0.001\n",
      "[40,   100] loss: 0.001\n",
      "[40,   110] loss: 0.001\n",
      "[40,   120] loss: 0.002\n",
      "[40,   130] loss: 0.001\n",
      "[40,   140] loss: 0.002\n",
      "[40,   150] loss: 0.001\n",
      "[40,   160] loss: 0.001\n",
      "[40,   170] loss: 0.002\n",
      "[40,   180] loss: 0.001\n",
      "[41,    10] loss: 0.002\n",
      "[41,    20] loss: 0.001\n",
      "[41,    30] loss: 0.001\n",
      "[41,    40] loss: 0.001\n",
      "[41,    50] loss: 0.001\n",
      "[41,    60] loss: 0.001\n",
      "[41,    70] loss: 0.001\n",
      "[41,    80] loss: 0.001\n",
      "[41,    90] loss: 0.001\n",
      "[41,   100] loss: 0.001\n",
      "[41,   110] loss: 0.002\n",
      "[41,   120] loss: 0.001\n",
      "[41,   130] loss: 0.001\n",
      "[41,   140] loss: 0.002\n",
      "[41,   150] loss: 0.001\n",
      "[41,   160] loss: 0.001\n",
      "[41,   170] loss: 0.001\n",
      "[41,   180] loss: 0.001\n",
      "[42,    10] loss: 0.001\n",
      "[42,    20] loss: 0.001\n",
      "[42,    30] loss: 0.001\n",
      "[42,    40] loss: 0.001\n",
      "[42,    50] loss: 0.001\n",
      "[42,    60] loss: 0.001\n",
      "[42,    70] loss: 0.001\n",
      "[42,    80] loss: 0.001\n",
      "[42,    90] loss: 0.001\n",
      "[42,   100] loss: 0.001\n",
      "[42,   110] loss: 0.002\n",
      "[42,   120] loss: 0.001\n",
      "[42,   130] loss: 0.001\n",
      "[42,   140] loss: 0.001\n",
      "[42,   150] loss: 0.002\n",
      "[42,   160] loss: 0.002\n",
      "[42,   170] loss: 0.001\n",
      "[42,   180] loss: 0.001\n",
      "[43,    10] loss: 0.001\n",
      "[43,    20] loss: 0.001\n",
      "[43,    30] loss: 0.001\n",
      "[43,    40] loss: 0.001\n",
      "[43,    50] loss: 0.002\n",
      "[43,    60] loss: 0.001\n",
      "[43,    70] loss: 0.001\n",
      "[43,    80] loss: 0.001\n",
      "[43,    90] loss: 0.001\n",
      "[43,   100] loss: 0.001\n",
      "[43,   110] loss: 0.001\n",
      "[43,   120] loss: 0.001\n",
      "[43,   130] loss: 0.002\n",
      "[43,   140] loss: 0.001\n",
      "[43,   150] loss: 0.001\n",
      "[43,   160] loss: 0.001\n",
      "[43,   170] loss: 0.001\n",
      "[43,   180] loss: 0.001\n",
      "[44,    10] loss: 0.002\n",
      "[44,    20] loss: 0.001\n",
      "[44,    30] loss: 0.001\n",
      "[44,    40] loss: 0.001\n",
      "[44,    50] loss: 0.001\n",
      "[44,    60] loss: 0.001\n",
      "[44,    70] loss: 0.001\n",
      "[44,    80] loss: 0.001\n",
      "[44,    90] loss: 0.001\n",
      "[44,   100] loss: 0.001\n",
      "[44,   110] loss: 0.002\n",
      "[44,   120] loss: 0.001\n",
      "[44,   130] loss: 0.001\n",
      "[44,   140] loss: 0.001\n",
      "[44,   150] loss: 0.001\n",
      "[44,   160] loss: 0.001\n",
      "[44,   170] loss: 0.001\n",
      "[44,   180] loss: 0.001\n",
      "[45,    10] loss: 0.001\n",
      "[45,    20] loss: 0.002\n",
      "[45,    30] loss: 0.002\n",
      "[45,    40] loss: 0.002\n",
      "[45,    50] loss: 0.001\n",
      "[45,    60] loss: 0.001\n",
      "[45,    70] loss: 0.002\n",
      "[45,    80] loss: 0.001\n",
      "[45,    90] loss: 0.001\n",
      "[45,   100] loss: 0.001\n",
      "[45,   110] loss: 0.001\n",
      "[45,   120] loss: 0.001\n",
      "[45,   130] loss: 0.001\n",
      "[45,   140] loss: 0.001\n",
      "[45,   150] loss: 0.001\n",
      "[45,   160] loss: 0.002\n",
      "[45,   170] loss: 0.002\n",
      "[45,   180] loss: 0.001\n",
      "[46,    10] loss: 0.001\n",
      "[46,    20] loss: 0.001\n",
      "[46,    30] loss: 0.001\n",
      "[46,    40] loss: 0.001\n",
      "[46,    50] loss: 0.001\n",
      "[46,    60] loss: 0.001\n",
      "[46,    70] loss: 0.001\n",
      "[46,    80] loss: 0.001\n",
      "[46,    90] loss: 0.001\n",
      "[46,   100] loss: 0.001\n",
      "[46,   110] loss: 0.001\n",
      "[46,   120] loss: 0.001\n",
      "[46,   130] loss: 0.001\n",
      "[46,   140] loss: 0.001\n",
      "[46,   150] loss: 0.001\n",
      "[46,   160] loss: 0.001\n",
      "[46,   170] loss: 0.002\n",
      "[46,   180] loss: 0.001\n",
      "[47,    10] loss: 0.001\n",
      "[47,    20] loss: 0.001\n",
      "[47,    30] loss: 0.001\n",
      "[47,    40] loss: 0.001\n",
      "[47,    50] loss: 0.002\n",
      "[47,    60] loss: 0.001\n",
      "[47,    70] loss: 0.001\n",
      "[47,    80] loss: 0.001\n",
      "[47,    90] loss: 0.001\n",
      "[47,   100] loss: 0.001\n",
      "[47,   110] loss: 0.001\n",
      "[47,   120] loss: 0.001\n",
      "[47,   130] loss: 0.001\n",
      "[47,   140] loss: 0.001\n",
      "[47,   150] loss: 0.001\n",
      "[47,   160] loss: 0.001\n",
      "[47,   170] loss: 0.002\n",
      "[47,   180] loss: 0.001\n",
      "[48,    10] loss: 0.001\n",
      "[48,    20] loss: 0.001\n",
      "[48,    30] loss: 0.001\n",
      "[48,    40] loss: 0.001\n",
      "[48,    50] loss: 0.001\n",
      "[48,    60] loss: 0.001\n",
      "[48,    70] loss: 0.001\n",
      "[48,    80] loss: 0.002\n",
      "[48,    90] loss: 0.001\n",
      "[48,   100] loss: 0.001\n",
      "[48,   110] loss: 0.001\n",
      "[48,   120] loss: 0.001\n",
      "[48,   130] loss: 0.001\n",
      "[48,   140] loss: 0.001\n",
      "[48,   150] loss: 0.001\n",
      "[48,   160] loss: 0.002\n",
      "[48,   170] loss: 0.001\n",
      "[48,   180] loss: 0.001\n",
      "[49,    10] loss: 0.001\n",
      "[49,    20] loss: 0.001\n",
      "[49,    30] loss: 0.001\n",
      "[49,    40] loss: 0.001\n",
      "[49,    50] loss: 0.001\n",
      "[49,    60] loss: 0.001\n",
      "[49,    70] loss: 0.001\n",
      "[49,    80] loss: 0.002\n",
      "[49,    90] loss: 0.001\n",
      "[49,   100] loss: 0.001\n",
      "[49,   110] loss: 0.001\n",
      "[49,   120] loss: 0.001\n",
      "[49,   130] loss: 0.002\n",
      "[49,   140] loss: 0.001\n",
      "[49,   150] loss: 0.001\n",
      "[49,   160] loss: 0.001\n",
      "[49,   170] loss: 0.001\n",
      "[49,   180] loss: 0.001\n",
      "[50,    10] loss: 0.002\n",
      "[50,    20] loss: 0.001\n",
      "[50,    30] loss: 0.002\n",
      "[50,    40] loss: 0.001\n",
      "[50,    50] loss: 0.002\n",
      "[50,    60] loss: 0.001\n",
      "[50,    70] loss: 0.001\n",
      "[50,    80] loss: 0.001\n",
      "[50,    90] loss: 0.001\n",
      "[50,   100] loss: 0.001\n",
      "[50,   110] loss: 0.001\n",
      "[50,   120] loss: 0.001\n",
      "[50,   130] loss: 0.001\n",
      "[50,   140] loss: 0.001\n",
      "[50,   150] loss: 0.001\n",
      "[50,   160] loss: 0.001\n",
      "[50,   170] loss: 0.001\n",
      "[50,   180] loss: 0.001\n",
      "[51,    10] loss: 0.001\n",
      "[51,    20] loss: 0.002\n",
      "[51,    30] loss: 0.001\n",
      "[51,    40] loss: 0.001\n",
      "[51,    50] loss: 0.001\n",
      "[51,    60] loss: 0.001\n",
      "[51,    70] loss: 0.001\n",
      "[51,    80] loss: 0.001\n",
      "[51,    90] loss: 0.001\n",
      "[51,   100] loss: 0.001\n",
      "[51,   110] loss: 0.001\n",
      "[51,   120] loss: 0.001\n",
      "[51,   130] loss: 0.001\n",
      "[51,   140] loss: 0.001\n",
      "[51,   150] loss: 0.001\n",
      "[51,   160] loss: 0.001\n",
      "[51,   170] loss: 0.001\n",
      "[51,   180] loss: 0.001\n",
      "[52,    10] loss: 0.001\n",
      "[52,    20] loss: 0.001\n",
      "[52,    30] loss: 0.002\n",
      "[52,    40] loss: 0.001\n",
      "[52,    50] loss: 0.001\n",
      "[52,    60] loss: 0.001\n",
      "[52,    70] loss: 0.001\n",
      "[52,    80] loss: 0.001\n",
      "[52,    90] loss: 0.001\n",
      "[52,   100] loss: 0.001\n",
      "[52,   110] loss: 0.001\n",
      "[52,   120] loss: 0.001\n",
      "[52,   130] loss: 0.001\n",
      "[52,   140] loss: 0.001\n",
      "[52,   150] loss: 0.001\n",
      "[52,   160] loss: 0.001\n",
      "[52,   170] loss: 0.001\n",
      "[52,   180] loss: 0.001\n",
      "[53,    10] loss: 0.001\n",
      "[53,    20] loss: 0.001\n",
      "[53,    30] loss: 0.002\n",
      "[53,    40] loss: 0.001\n",
      "[53,    50] loss: 0.001\n",
      "[53,    60] loss: 0.001\n",
      "[53,    70] loss: 0.001\n",
      "[53,    80] loss: 0.001\n",
      "[53,    90] loss: 0.001\n",
      "[53,   100] loss: 0.001\n",
      "[53,   110] loss: 0.001\n",
      "[53,   120] loss: 0.001\n",
      "[53,   130] loss: 0.001\n",
      "[53,   140] loss: 0.001\n",
      "[53,   150] loss: 0.001\n",
      "[53,   160] loss: 0.001\n",
      "[53,   170] loss: 0.002\n",
      "[53,   180] loss: 0.001\n",
      "[54,    10] loss: 0.001\n",
      "[54,    20] loss: 0.001\n",
      "[54,    30] loss: 0.001\n",
      "[54,    40] loss: 0.001\n",
      "[54,    50] loss: 0.001\n",
      "[54,    60] loss: 0.002\n",
      "[54,    70] loss: 0.001\n",
      "[54,    80] loss: 0.001\n",
      "[54,    90] loss: 0.001\n",
      "[54,   100] loss: 0.001\n",
      "[54,   110] loss: 0.001\n",
      "[54,   120] loss: 0.001\n",
      "[54,   130] loss: 0.001\n",
      "[54,   140] loss: 0.001\n",
      "[54,   150] loss: 0.001\n",
      "[54,   160] loss: 0.001\n",
      "[54,   170] loss: 0.001\n",
      "[54,   180] loss: 0.001\n",
      "[55,    10] loss: 0.001\n",
      "[55,    20] loss: 0.001\n",
      "[55,    30] loss: 0.001\n",
      "[55,    40] loss: 0.001\n",
      "[55,    50] loss: 0.001\n",
      "[55,    60] loss: 0.001\n",
      "[55,    70] loss: 0.001\n",
      "[55,    80] loss: 0.001\n",
      "[55,    90] loss: 0.001\n",
      "[55,   100] loss: 0.001\n",
      "[55,   110] loss: 0.001\n",
      "[55,   120] loss: 0.001\n",
      "[55,   130] loss: 0.001\n",
      "[55,   140] loss: 0.001\n",
      "[55,   150] loss: 0.001\n",
      "[55,   160] loss: 0.001\n",
      "[55,   170] loss: 0.001\n",
      "[55,   180] loss: 0.002\n",
      "[56,    10] loss: 0.001\n",
      "[56,    20] loss: 0.001\n",
      "[56,    30] loss: 0.001\n",
      "[56,    40] loss: 0.001\n",
      "[56,    50] loss: 0.001\n",
      "[56,    60] loss: 0.001\n",
      "[56,    70] loss: 0.001\n",
      "[56,    80] loss: 0.001\n",
      "[56,    90] loss: 0.001\n",
      "[56,   100] loss: 0.001\n",
      "[56,   110] loss: 0.001\n",
      "[56,   120] loss: 0.001\n",
      "[56,   130] loss: 0.001\n",
      "[56,   140] loss: 0.001\n",
      "[56,   150] loss: 0.001\n",
      "[56,   160] loss: 0.001\n",
      "[56,   170] loss: 0.001\n",
      "[56,   180] loss: 0.001\n",
      "[57,    10] loss: 0.001\n",
      "[57,    20] loss: 0.001\n",
      "[57,    30] loss: 0.001\n",
      "[57,    40] loss: 0.001\n",
      "[57,    50] loss: 0.001\n",
      "[57,    60] loss: 0.001\n",
      "[57,    70] loss: 0.001\n",
      "[57,    80] loss: 0.001\n",
      "[57,    90] loss: 0.001\n",
      "[57,   100] loss: 0.002\n",
      "[57,   110] loss: 0.001\n",
      "[57,   120] loss: 0.001\n",
      "[57,   130] loss: 0.001\n",
      "[57,   140] loss: 0.001\n",
      "[57,   150] loss: 0.001\n",
      "[57,   160] loss: 0.001\n",
      "[57,   170] loss: 0.001\n",
      "[57,   180] loss: 0.001\n",
      "[58,    10] loss: 0.001\n",
      "[58,    20] loss: 0.001\n",
      "[58,    30] loss: 0.001\n",
      "[58,    40] loss: 0.001\n",
      "[58,    50] loss: 0.001\n",
      "[58,    60] loss: 0.001\n",
      "[58,    70] loss: 0.001\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[58,    80] loss: 0.001\n",
      "[58,    90] loss: 0.001\n",
      "[58,   100] loss: 0.001\n",
      "[58,   110] loss: 0.001\n",
      "[58,   120] loss: 0.001\n",
      "[58,   130] loss: 0.001\n",
      "[58,   140] loss: 0.001\n",
      "[58,   150] loss: 0.001\n",
      "[58,   160] loss: 0.001\n",
      "[58,   170] loss: 0.001\n",
      "[58,   180] loss: 0.001\n",
      "[59,    10] loss: 0.001\n",
      "[59,    20] loss: 0.001\n",
      "[59,    30] loss: 0.001\n",
      "[59,    40] loss: 0.001\n",
      "[59,    50] loss: 0.001\n",
      "[59,    60] loss: 0.001\n",
      "[59,    70] loss: 0.001\n",
      "[59,    80] loss: 0.001\n",
      "[59,    90] loss: 0.002\n",
      "[59,   100] loss: 0.002\n",
      "[59,   110] loss: 0.001\n",
      "[59,   120] loss: 0.001\n",
      "[59,   130] loss: 0.001\n",
      "[59,   140] loss: 0.001\n",
      "[59,   150] loss: 0.001\n",
      "[59,   160] loss: 0.001\n",
      "[59,   170] loss: 0.001\n",
      "[59,   180] loss: 0.001\n",
      "[60,    10] loss: 0.001\n",
      "[60,    20] loss: 0.001\n",
      "[60,    30] loss: 0.001\n",
      "[60,    40] loss: 0.001\n",
      "[60,    50] loss: 0.001\n",
      "[60,    60] loss: 0.001\n",
      "[60,    70] loss: 0.001\n",
      "[60,    80] loss: 0.001\n",
      "[60,    90] loss: 0.001\n",
      "[60,   100] loss: 0.001\n",
      "[60,   110] loss: 0.001\n",
      "[60,   120] loss: 0.001\n",
      "[60,   130] loss: 0.001\n",
      "[60,   140] loss: 0.001\n",
      "[60,   150] loss: 0.001\n",
      "[60,   160] loss: 0.001\n",
      "[60,   170] loss: 0.001\n",
      "[60,   180] loss: 0.001\n",
      "[61,    10] loss: 0.001\n",
      "[61,    20] loss: 0.001\n",
      "[61,    30] loss: 0.001\n",
      "[61,    40] loss: 0.001\n",
      "[61,    50] loss: 0.001\n",
      "[61,    60] loss: 0.001\n",
      "[61,    70] loss: 0.001\n",
      "[61,    80] loss: 0.001\n",
      "[61,    90] loss: 0.001\n",
      "[61,   100] loss: 0.001\n",
      "[61,   110] loss: 0.001\n",
      "[61,   120] loss: 0.001\n",
      "[61,   130] loss: 0.001\n",
      "[61,   140] loss: 0.001\n",
      "[61,   150] loss: 0.001\n",
      "[61,   160] loss: 0.001\n",
      "[61,   170] loss: 0.001\n",
      "[61,   180] loss: 0.001\n",
      "[62,    10] loss: 0.001\n",
      "[62,    20] loss: 0.001\n",
      "[62,    30] loss: 0.001\n",
      "[62,    40] loss: 0.001\n",
      "[62,    50] loss: 0.001\n",
      "[62,    60] loss: 0.001\n",
      "[62,    70] loss: 0.001\n",
      "[62,    80] loss: 0.001\n",
      "[62,    90] loss: 0.001\n",
      "[62,   100] loss: 0.001\n",
      "[62,   110] loss: 0.001\n",
      "[62,   120] loss: 0.001\n",
      "[62,   130] loss: 0.001\n",
      "[62,   140] loss: 0.001\n",
      "[62,   150] loss: 0.001\n",
      "[62,   160] loss: 0.001\n",
      "[62,   170] loss: 0.001\n",
      "[62,   180] loss: 0.001\n",
      "[63,    10] loss: 0.001\n",
      "[63,    20] loss: 0.001\n",
      "[63,    30] loss: 0.001\n",
      "[63,    40] loss: 0.001\n",
      "[63,    50] loss: 0.001\n",
      "[63,    60] loss: 0.001\n",
      "[63,    70] loss: 0.001\n",
      "[63,    80] loss: 0.001\n",
      "[63,    90] loss: 0.001\n",
      "[63,   100] loss: 0.001\n",
      "[63,   110] loss: 0.001\n",
      "[63,   120] loss: 0.001\n",
      "[63,   130] loss: 0.001\n",
      "[63,   140] loss: 0.001\n",
      "[63,   150] loss: 0.002\n",
      "[63,   160] loss: 0.001\n",
      "[63,   170] loss: 0.001\n",
      "[63,   180] loss: 0.001\n",
      "[64,    10] loss: 0.001\n",
      "[64,    20] loss: 0.001\n",
      "[64,    30] loss: 0.001\n",
      "[64,    40] loss: 0.001\n",
      "[64,    50] loss: 0.001\n",
      "[64,    60] loss: 0.001\n",
      "[64,    70] loss: 0.001\n",
      "[64,    80] loss: 0.001\n",
      "[64,    90] loss: 0.001\n",
      "[64,   100] loss: 0.001\n",
      "[64,   110] loss: 0.001\n",
      "[64,   120] loss: 0.001\n",
      "[64,   130] loss: 0.001\n",
      "[64,   140] loss: 0.001\n",
      "[64,   150] loss: 0.001\n",
      "[64,   160] loss: 0.001\n",
      "[64,   170] loss: 0.001\n",
      "[64,   180] loss: 0.001\n",
      "[65,    10] loss: 0.001\n",
      "[65,    20] loss: 0.001\n",
      "[65,    30] loss: 0.001\n",
      "[65,    40] loss: 0.001\n",
      "[65,    50] loss: 0.001\n",
      "[65,    60] loss: 0.001\n",
      "[65,    70] loss: 0.001\n",
      "[65,    80] loss: 0.001\n",
      "[65,    90] loss: 0.001\n",
      "[65,   100] loss: 0.001\n",
      "[65,   110] loss: 0.001\n",
      "[65,   120] loss: 0.001\n",
      "[65,   130] loss: 0.001\n",
      "[65,   140] loss: 0.001\n",
      "[65,   150] loss: 0.001\n",
      "[65,   160] loss: 0.001\n",
      "[65,   170] loss: 0.001\n",
      "[65,   180] loss: 0.001\n",
      "[66,    10] loss: 0.001\n",
      "[66,    20] loss: 0.001\n",
      "[66,    30] loss: 0.001\n",
      "[66,    40] loss: 0.001\n",
      "[66,    50] loss: 0.001\n",
      "[66,    60] loss: 0.001\n",
      "[66,    70] loss: 0.001\n",
      "[66,    80] loss: 0.001\n",
      "[66,    90] loss: 0.001\n",
      "[66,   100] loss: 0.001\n",
      "[66,   110] loss: 0.001\n",
      "[66,   120] loss: 0.001\n",
      "[66,   130] loss: 0.001\n",
      "[66,   140] loss: 0.001\n",
      "[66,   150] loss: 0.001\n",
      "[66,   160] loss: 0.001\n",
      "[66,   170] loss: 0.001\n",
      "[66,   180] loss: 0.001\n",
      "[67,    10] loss: 0.001\n",
      "[67,    20] loss: 0.001\n",
      "[67,    30] loss: 0.001\n",
      "[67,    40] loss: 0.001\n",
      "[67,    50] loss: 0.001\n",
      "[67,    60] loss: 0.001\n",
      "[67,    70] loss: 0.001\n",
      "[67,    80] loss: 0.001\n",
      "[67,    90] loss: 0.001\n",
      "[67,   100] loss: 0.001\n",
      "[67,   110] loss: 0.001\n",
      "[67,   120] loss: 0.001\n",
      "[67,   130] loss: 0.001\n",
      "[67,   140] loss: 0.001\n",
      "[67,   150] loss: 0.001\n",
      "[67,   160] loss: 0.001\n",
      "[67,   170] loss: 0.001\n",
      "[67,   180] loss: 0.001\n",
      "[68,    10] loss: 0.001\n",
      "[68,    20] loss: 0.001\n",
      "[68,    30] loss: 0.001\n",
      "[68,    40] loss: 0.001\n",
      "[68,    50] loss: 0.001\n",
      "[68,    60] loss: 0.001\n",
      "[68,    70] loss: 0.001\n",
      "[68,    80] loss: 0.001\n",
      "[68,    90] loss: 0.001\n",
      "[68,   100] loss: 0.000\n",
      "[68,   110] loss: 0.001\n",
      "[68,   120] loss: 0.001\n",
      "[68,   130] loss: 0.001\n",
      "[68,   140] loss: 0.001\n",
      "[68,   150] loss: 0.001\n",
      "[68,   160] loss: 0.001\n",
      "[68,   170] loss: 0.001\n",
      "[68,   180] loss: 0.001\n",
      "[69,    10] loss: 0.001\n",
      "[69,    20] loss: 0.001\n",
      "[69,    30] loss: 0.001\n",
      "[69,    40] loss: 0.001\n",
      "[69,    50] loss: 0.001\n",
      "[69,    60] loss: 0.001\n",
      "[69,    70] loss: 0.000\n",
      "[69,    80] loss: 0.001\n",
      "[69,    90] loss: 0.001\n",
      "[69,   100] loss: 0.001\n",
      "[69,   110] loss: 0.001\n",
      "[69,   120] loss: 0.001\n",
      "[69,   130] loss: 0.001\n",
      "[69,   140] loss: 0.001\n",
      "[69,   150] loss: 0.001\n",
      "[69,   160] loss: 0.001\n",
      "[69,   170] loss: 0.001\n",
      "[69,   180] loss: 0.001\n",
      "[70,    10] loss: 0.001\n",
      "[70,    20] loss: 0.001\n",
      "[70,    30] loss: 0.001\n",
      "[70,    40] loss: 0.001\n",
      "[70,    50] loss: 0.001\n",
      "[70,    60] loss: 0.001\n",
      "[70,    70] loss: 0.001\n",
      "[70,    80] loss: 0.001\n",
      "[70,    90] loss: 0.001\n",
      "[70,   100] loss: 0.001\n",
      "[70,   110] loss: 0.001\n",
      "[70,   120] loss: 0.001\n",
      "[70,   130] loss: 0.001\n",
      "[70,   140] loss: 0.001\n",
      "[70,   150] loss: 0.001\n",
      "[70,   160] loss: 0.001\n",
      "[70,   170] loss: 0.001\n",
      "[70,   180] loss: 0.001\n",
      "[71,    10] loss: 0.001\n",
      "[71,    20] loss: 0.001\n",
      "[71,    30] loss: 0.001\n",
      "[71,    40] loss: 0.001\n",
      "[71,    50] loss: 0.001\n",
      "[71,    60] loss: 0.000\n",
      "[71,    70] loss: 0.001\n",
      "[71,    80] loss: 0.001\n",
      "[71,    90] loss: 0.001\n",
      "[71,   100] loss: 0.001\n",
      "[71,   110] loss: 0.001\n",
      "[71,   120] loss: 0.001\n",
      "[71,   130] loss: 0.001\n",
      "[71,   140] loss: 0.001\n",
      "[71,   150] loss: 0.001\n",
      "[71,   160] loss: 0.001\n",
      "[71,   170] loss: 0.000\n",
      "[71,   180] loss: 0.001\n",
      "[72,    10] loss: 0.001\n",
      "[72,    20] loss: 0.001\n",
      "[72,    30] loss: 0.001\n",
      "[72,    40] loss: 0.001\n",
      "[72,    50] loss: 0.001\n",
      "[72,    60] loss: 0.001\n",
      "[72,    70] loss: 0.001\n",
      "[72,    80] loss: 0.001\n",
      "[72,    90] loss: 0.001\n",
      "[72,   100] loss: 0.001\n",
      "[72,   110] loss: 0.001\n",
      "[72,   120] loss: 0.001\n",
      "[72,   130] loss: 0.001\n",
      "[72,   140] loss: 0.001\n",
      "[72,   150] loss: 0.001\n",
      "[72,   160] loss: 0.001\n",
      "[72,   170] loss: 0.001\n",
      "[72,   180] loss: 0.001\n",
      "[73,    10] loss: 0.001\n",
      "[73,    20] loss: 0.001\n",
      "[73,    30] loss: 0.001\n",
      "[73,    40] loss: 0.001\n",
      "[73,    50] loss: 0.001\n",
      "[73,    60] loss: 0.001\n",
      "[73,    70] loss: 0.001\n",
      "[73,    80] loss: 0.001\n",
      "[73,    90] loss: 0.001\n",
      "[73,   100] loss: 0.001\n",
      "[73,   110] loss: 0.001\n",
      "[73,   120] loss: 0.001\n",
      "[73,   130] loss: 0.001\n",
      "[73,   140] loss: 0.001\n",
      "[73,   150] loss: 0.001\n",
      "[73,   160] loss: 0.001\n",
      "[73,   170] loss: 0.001\n",
      "[73,   180] loss: 0.001\n",
      "[74,    10] loss: 0.000\n",
      "[74,    20] loss: 0.001\n",
      "[74,    30] loss: 0.001\n",
      "[74,    40] loss: 0.000\n",
      "[74,    50] loss: 0.001\n",
      "[74,    60] loss: 0.001\n",
      "[74,    70] loss: 0.001\n",
      "[74,    80] loss: 0.001\n",
      "[74,    90] loss: 0.000\n",
      "[74,   100] loss: 0.001\n",
      "[74,   110] loss: 0.001\n",
      "[74,   120] loss: 0.001\n",
      "[74,   130] loss: 0.001\n",
      "[74,   140] loss: 0.001\n",
      "[74,   150] loss: 0.001\n",
      "[74,   160] loss: 0.001\n",
      "[74,   170] loss: 0.001\n",
      "[74,   180] loss: 0.001\n",
      "[75,    10] loss: 0.000\n",
      "[75,    20] loss: 0.001\n",
      "[75,    30] loss: 0.001\n",
      "[75,    40] loss: 0.001\n",
      "[75,    50] loss: 0.001\n",
      "[75,    60] loss: 0.001\n",
      "[75,    70] loss: 0.001\n",
      "[75,    80] loss: 0.000\n",
      "[75,    90] loss: 0.001\n",
      "[75,   100] loss: 0.001\n",
      "[75,   110] loss: 0.001\n",
      "[75,   120] loss: 0.001\n",
      "[75,   130] loss: 0.001\n",
      "[75,   140] loss: 0.000\n",
      "[75,   150] loss: 0.001\n",
      "[75,   160] loss: 0.000\n",
      "[75,   170] loss: 0.000\n",
      "[75,   180] loss: 0.001\n",
      "[76,    10] loss: 0.001\n",
      "[76,    20] loss: 0.001\n",
      "[76,    30] loss: 0.001\n",
      "[76,    40] loss: 0.001\n",
      "[76,    50] loss: 0.001\n",
      "[76,    60] loss: 0.001\n",
      "[76,    70] loss: 0.001\n",
      "[76,    80] loss: 0.001\n",
      "[76,    90] loss: 0.001\n",
      "[76,   100] loss: 0.000\n",
      "[76,   110] loss: 0.001\n",
      "[76,   120] loss: 0.001\n",
      "[76,   130] loss: 0.001\n",
      "[76,   140] loss: 0.000\n",
      "[76,   150] loss: 0.000\n",
      "[76,   160] loss: 0.000\n",
      "[76,   170] loss: 0.001\n",
      "[76,   180] loss: 0.001\n",
      "[77,    10] loss: 0.001\n",
      "[77,    20] loss: 0.001\n",
      "[77,    30] loss: 0.001\n",
      "[77,    40] loss: 0.000\n",
      "[77,    50] loss: 0.001\n",
      "[77,    60] loss: 0.001\n",
      "[77,    70] loss: 0.001\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[77,    80] loss: 0.001\n",
      "[77,    90] loss: 0.001\n",
      "[77,   100] loss: 0.000\n",
      "[77,   110] loss: 0.001\n",
      "[77,   120] loss: 0.001\n",
      "[77,   130] loss: 0.000\n",
      "[77,   140] loss: 0.001\n",
      "[77,   150] loss: 0.000\n",
      "[77,   160] loss: 0.001\n",
      "[77,   170] loss: 0.001\n",
      "[77,   180] loss: 0.001\n",
      "[78,    10] loss: 0.000\n",
      "[78,    20] loss: 0.001\n",
      "[78,    30] loss: 0.000\n",
      "[78,    40] loss: 0.001\n",
      "[78,    50] loss: 0.000\n",
      "[78,    60] loss: 0.000\n",
      "[78,    70] loss: 0.001\n",
      "[78,    80] loss: 0.000\n",
      "[78,    90] loss: 0.001\n",
      "[78,   100] loss: 0.000\n",
      "[78,   110] loss: 0.001\n",
      "[78,   120] loss: 0.001\n",
      "[78,   130] loss: 0.001\n",
      "[78,   140] loss: 0.001\n",
      "[78,   150] loss: 0.001\n",
      "[78,   160] loss: 0.001\n",
      "[78,   170] loss: 0.000\n",
      "[78,   180] loss: 0.000\n",
      "[79,    10] loss: 0.000\n",
      "[79,    20] loss: 0.001\n",
      "[79,    30] loss: 0.000\n",
      "[79,    40] loss: 0.000\n",
      "[79,    50] loss: 0.000\n",
      "[79,    60] loss: 0.000\n",
      "[79,    70] loss: 0.000\n",
      "[79,    80] loss: 0.000\n",
      "[79,    90] loss: 0.000\n",
      "[79,   100] loss: 0.000\n",
      "[79,   110] loss: 0.000\n",
      "[79,   120] loss: 0.001\n",
      "[79,   130] loss: 0.000\n",
      "[79,   140] loss: 0.000\n",
      "[79,   150] loss: 0.001\n",
      "[79,   160] loss: 0.000\n",
      "[79,   170] loss: 0.001\n",
      "[79,   180] loss: 0.000\n",
      "[80,    10] loss: 0.000\n",
      "[80,    20] loss: 0.000\n",
      "[80,    30] loss: 0.000\n",
      "[80,    40] loss: 0.000\n",
      "[80,    50] loss: 0.001\n",
      "[80,    60] loss: 0.001\n",
      "[80,    70] loss: 0.001\n",
      "[80,    80] loss: 0.000\n",
      "[80,    90] loss: 0.000\n",
      "[80,   100] loss: 0.001\n",
      "[80,   110] loss: 0.000\n",
      "[80,   120] loss: 0.001\n",
      "[80,   130] loss: 0.001\n",
      "[80,   140] loss: 0.000\n",
      "[80,   150] loss: 0.001\n",
      "[80,   160] loss: 0.001\n",
      "[80,   170] loss: 0.001\n",
      "[80,   180] loss: 0.001\n",
      "[81,    10] loss: 0.000\n",
      "[81,    20] loss: 0.001\n",
      "[81,    30] loss: 0.000\n",
      "[81,    40] loss: 0.001\n",
      "[81,    50] loss: 0.001\n",
      "[81,    60] loss: 0.001\n",
      "[81,    70] loss: 0.001\n",
      "[81,    80] loss: 0.001\n",
      "[81,    90] loss: 0.000\n",
      "[81,   100] loss: 0.000\n",
      "[81,   110] loss: 0.000\n",
      "[81,   120] loss: 0.000\n",
      "[81,   130] loss: 0.001\n",
      "[81,   140] loss: 0.001\n",
      "[81,   150] loss: 0.000\n",
      "[81,   160] loss: 0.000\n",
      "[81,   170] loss: 0.000\n",
      "[81,   180] loss: 0.001\n",
      "[82,    10] loss: 0.000\n",
      "[82,    20] loss: 0.000\n",
      "[82,    30] loss: 0.000\n",
      "[82,    40] loss: 0.000\n",
      "[82,    50] loss: 0.001\n",
      "[82,    60] loss: 0.000\n",
      "[82,    70] loss: 0.000\n",
      "[82,    80] loss: 0.000\n",
      "[82,    90] loss: 0.001\n",
      "[82,   100] loss: 0.001\n",
      "[82,   110] loss: 0.001\n",
      "[82,   120] loss: 0.001\n",
      "[82,   130] loss: 0.001\n",
      "[82,   140] loss: 0.001\n",
      "[82,   150] loss: 0.000\n",
      "[82,   160] loss: 0.001\n",
      "[82,   170] loss: 0.000\n",
      "[82,   180] loss: 0.000\n",
      "[83,    10] loss: 0.000\n",
      "[83,    20] loss: 0.000\n",
      "[83,    30] loss: 0.000\n",
      "[83,    40] loss: 0.001\n",
      "[83,    50] loss: 0.000\n",
      "[83,    60] loss: 0.000\n",
      "[83,    70] loss: 0.000\n",
      "[83,    80] loss: 0.001\n",
      "[83,    90] loss: 0.001\n",
      "[83,   100] loss: 0.001\n",
      "[83,   110] loss: 0.001\n",
      "[83,   120] loss: 0.001\n",
      "[83,   130] loss: 0.001\n",
      "[83,   140] loss: 0.001\n",
      "[83,   150] loss: 0.001\n",
      "[83,   160] loss: 0.000\n",
      "[83,   170] loss: 0.000\n",
      "[83,   180] loss: 0.000\n",
      "[84,    10] loss: 0.001\n",
      "[84,    20] loss: 0.000\n",
      "[84,    30] loss: 0.000\n",
      "[84,    40] loss: 0.000\n",
      "[84,    50] loss: 0.000\n",
      "[84,    60] loss: 0.000\n",
      "[84,    70] loss: 0.001\n",
      "[84,    80] loss: 0.000\n",
      "[84,    90] loss: 0.000\n",
      "[84,   100] loss: 0.000\n",
      "[84,   110] loss: 0.000\n",
      "[84,   120] loss: 0.000\n",
      "[84,   130] loss: 0.000\n",
      "[84,   140] loss: 0.000\n",
      "[84,   150] loss: 0.000\n",
      "[84,   160] loss: 0.000\n",
      "[84,   170] loss: 0.000\n",
      "[84,   180] loss: 0.001\n",
      "[85,    10] loss: 0.000\n",
      "[85,    20] loss: 0.000\n",
      "[85,    30] loss: 0.000\n",
      "[85,    40] loss: 0.000\n",
      "[85,    50] loss: 0.000\n",
      "[85,    60] loss: 0.000\n",
      "[85,    70] loss: 0.000\n",
      "[85,    80] loss: 0.000\n",
      "[85,    90] loss: 0.000\n",
      "[85,   100] loss: 0.000\n",
      "[85,   110] loss: 0.001\n",
      "[85,   120] loss: 0.000\n",
      "[85,   130] loss: 0.000\n",
      "[85,   140] loss: 0.000\n",
      "[85,   150] loss: 0.000\n",
      "[85,   160] loss: 0.001\n",
      "[85,   170] loss: 0.000\n",
      "[85,   180] loss: 0.001\n",
      "[86,    10] loss: 0.001\n",
      "[86,    20] loss: 0.000\n",
      "[86,    30] loss: 0.000\n",
      "[86,    40] loss: 0.000\n",
      "[86,    50] loss: 0.001\n",
      "[86,    60] loss: 0.000\n",
      "[86,    70] loss: 0.000\n",
      "[86,    80] loss: 0.001\n",
      "[86,    90] loss: 0.000\n",
      "[86,   100] loss: 0.000\n",
      "[86,   110] loss: 0.001\n",
      "[86,   120] loss: 0.000\n",
      "[86,   130] loss: 0.000\n",
      "[86,   140] loss: 0.000\n",
      "[86,   150] loss: 0.001\n",
      "[86,   160] loss: 0.000\n",
      "[86,   170] loss: 0.000\n",
      "[86,   180] loss: 0.000\n",
      "[87,    10] loss: 0.000\n",
      "[87,    20] loss: 0.000\n",
      "[87,    30] loss: 0.000\n",
      "[87,    40] loss: 0.000\n",
      "[87,    50] loss: 0.000\n",
      "[87,    60] loss: 0.001\n",
      "[87,    70] loss: 0.001\n",
      "[87,    80] loss: 0.000\n",
      "[87,    90] loss: 0.000\n",
      "[87,   100] loss: 0.000\n",
      "[87,   110] loss: 0.000\n",
      "[87,   120] loss: 0.000\n",
      "[87,   130] loss: 0.000\n",
      "[87,   140] loss: 0.000\n",
      "[87,   150] loss: 0.001\n",
      "[87,   160] loss: 0.000\n",
      "[87,   170] loss: 0.000\n",
      "[87,   180] loss: 0.000\n",
      "[88,    10] loss: 0.000\n",
      "[88,    20] loss: 0.000\n",
      "[88,    30] loss: 0.000\n",
      "[88,    40] loss: 0.000\n",
      "[88,    50] loss: 0.000\n",
      "[88,    60] loss: 0.000\n",
      "[88,    70] loss: 0.000\n",
      "[88,    80] loss: 0.000\n",
      "[88,    90] loss: 0.000\n",
      "[88,   100] loss: 0.000\n",
      "[88,   110] loss: 0.000\n",
      "[88,   120] loss: 0.000\n",
      "[88,   130] loss: 0.000\n",
      "[88,   140] loss: 0.000\n",
      "[88,   150] loss: 0.000\n",
      "[88,   160] loss: 0.001\n",
      "[88,   170] loss: 0.001\n",
      "[88,   180] loss: 0.001\n",
      "[89,    10] loss: 0.000\n",
      "[89,    20] loss: 0.000\n",
      "[89,    30] loss: 0.000\n",
      "[89,    40] loss: 0.000\n",
      "[89,    50] loss: 0.000\n",
      "[89,    60] loss: 0.000\n",
      "[89,    70] loss: 0.000\n",
      "[89,    80] loss: 0.000\n",
      "[89,    90] loss: 0.000\n",
      "[89,   100] loss: 0.000\n",
      "[89,   110] loss: 0.000\n",
      "[89,   120] loss: 0.000\n",
      "[89,   130] loss: 0.000\n",
      "[89,   140] loss: 0.000\n",
      "[89,   150] loss: 0.001\n",
      "[89,   160] loss: 0.000\n",
      "[89,   170] loss: 0.001\n",
      "[89,   180] loss: 0.000\n",
      "[90,    10] loss: 0.000\n",
      "[90,    20] loss: 0.000\n",
      "[90,    30] loss: 0.000\n",
      "[90,    40] loss: 0.000\n",
      "[90,    50] loss: 0.000\n",
      "[90,    60] loss: 0.000\n",
      "[90,    70] loss: 0.000\n",
      "[90,    80] loss: 0.000\n",
      "[90,    90] loss: 0.000\n",
      "[90,   100] loss: 0.000\n",
      "[90,   110] loss: 0.000\n",
      "[90,   120] loss: 0.000\n",
      "[90,   130] loss: 0.000\n",
      "[90,   140] loss: 0.000\n",
      "[90,   150] loss: 0.000\n",
      "[90,   160] loss: 0.000\n",
      "[90,   170] loss: 0.000\n",
      "[90,   180] loss: 0.000\n",
      "[91,    10] loss: 0.001\n",
      "[91,    20] loss: 0.000\n",
      "[91,    30] loss: 0.001\n",
      "[91,    40] loss: 0.000\n",
      "[91,    50] loss: 0.001\n",
      "[91,    60] loss: 0.000\n",
      "[91,    70] loss: 0.001\n",
      "[91,    80] loss: 0.000\n",
      "[91,    90] loss: 0.000\n",
      "[91,   100] loss: 0.000\n",
      "[91,   110] loss: 0.000\n",
      "[91,   120] loss: 0.000\n",
      "[91,   130] loss: 0.000\n",
      "[91,   140] loss: 0.000\n",
      "[91,   150] loss: 0.000\n",
      "[91,   160] loss: 0.000\n",
      "[91,   170] loss: 0.000\n",
      "[91,   180] loss: 0.000\n",
      "[92,    10] loss: 0.000\n",
      "[92,    20] loss: 0.000\n",
      "[92,    30] loss: 0.001\n",
      "[92,    40] loss: 0.000\n",
      "[92,    50] loss: 0.000\n",
      "[92,    60] loss: 0.000\n",
      "[92,    70] loss: 0.000\n",
      "[92,    80] loss: 0.000\n",
      "[92,    90] loss: 0.000\n",
      "[92,   100] loss: 0.000\n",
      "[92,   110] loss: 0.000\n",
      "[92,   120] loss: 0.000\n",
      "[92,   130] loss: 0.000\n",
      "[92,   140] loss: 0.000\n",
      "[92,   150] loss: 0.000\n",
      "[92,   160] loss: 0.000\n",
      "[92,   170] loss: 0.000\n",
      "[92,   180] loss: 0.000\n",
      "[93,    10] loss: 0.000\n",
      "[93,    20] loss: 0.000\n",
      "[93,    30] loss: 0.000\n",
      "[93,    40] loss: 0.000\n",
      "[93,    50] loss: 0.000\n",
      "[93,    60] loss: 0.000\n",
      "[93,    70] loss: 0.000\n",
      "[93,    80] loss: 0.000\n",
      "[93,    90] loss: 0.000\n",
      "[93,   100] loss: 0.000\n",
      "[93,   110] loss: 0.000\n",
      "[93,   120] loss: 0.000\n",
      "[93,   130] loss: 0.000\n",
      "[93,   140] loss: 0.000\n",
      "[93,   150] loss: 0.001\n",
      "[93,   160] loss: 0.000\n",
      "[93,   170] loss: 0.000\n",
      "[93,   180] loss: 0.000\n",
      "[94,    10] loss: 0.000\n",
      "[94,    20] loss: 0.000\n",
      "[94,    30] loss: 0.001\n",
      "[94,    40] loss: 0.000\n",
      "[94,    50] loss: 0.001\n",
      "[94,    60] loss: 0.000\n",
      "[94,    70] loss: 0.001\n",
      "[94,    80] loss: 0.000\n",
      "[94,    90] loss: 0.000\n",
      "[94,   100] loss: 0.000\n",
      "[94,   110] loss: 0.000\n",
      "[94,   120] loss: 0.000\n",
      "[94,   130] loss: 0.000\n",
      "[94,   140] loss: 0.001\n",
      "[94,   150] loss: 0.001\n",
      "[94,   160] loss: 0.001\n",
      "[94,   170] loss: 0.000\n",
      "[94,   180] loss: 0.000\n",
      "[95,    10] loss: 0.000\n",
      "[95,    20] loss: 0.000\n",
      "[95,    30] loss: 0.000\n",
      "[95,    40] loss: 0.000\n",
      "[95,    50] loss: 0.000\n",
      "[95,    60] loss: 0.000\n",
      "[95,    70] loss: 0.000\n",
      "[95,    80] loss: 0.000\n",
      "[95,    90] loss: 0.000\n",
      "[95,   100] loss: 0.000\n",
      "[95,   110] loss: 0.000\n",
      "[95,   120] loss: 0.000\n",
      "[95,   130] loss: 0.000\n",
      "[95,   140] loss: 0.000\n",
      "[95,   150] loss: 0.000\n",
      "[95,   160] loss: 0.000\n",
      "[95,   170] loss: 0.000\n",
      "[95,   180] loss: 0.000\n",
      "[96,    10] loss: 0.000\n",
      "[96,    20] loss: 0.000\n",
      "[96,    30] loss: 0.000\n",
      "[96,    40] loss: 0.000\n",
      "[96,    50] loss: 0.000\n",
      "[96,    60] loss: 0.000\n",
      "[96,    70] loss: 0.000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[96,    80] loss: 0.000\n",
      "[96,    90] loss: 0.000\n",
      "[96,   100] loss: 0.000\n",
      "[96,   110] loss: 0.000\n",
      "[96,   120] loss: 0.000\n",
      "[96,   130] loss: 0.001\n",
      "[96,   140] loss: 0.000\n",
      "[96,   150] loss: 0.000\n",
      "[96,   160] loss: 0.000\n",
      "[96,   170] loss: 0.000\n",
      "[96,   180] loss: 0.000\n",
      "[97,    10] loss: 0.000\n",
      "[97,    20] loss: 0.000\n",
      "[97,    30] loss: 0.000\n",
      "[97,    40] loss: 0.000\n",
      "[97,    50] loss: 0.000\n",
      "[97,    60] loss: 0.000\n",
      "[97,    70] loss: 0.000\n",
      "[97,    80] loss: 0.000\n",
      "[97,    90] loss: 0.001\n",
      "[97,   100] loss: 0.000\n",
      "[97,   110] loss: 0.000\n",
      "[97,   120] loss: 0.000\n",
      "[97,   130] loss: 0.000\n",
      "[97,   140] loss: 0.000\n",
      "[97,   150] loss: 0.000\n",
      "[97,   160] loss: 0.000\n",
      "[97,   170] loss: 0.000\n",
      "[97,   180] loss: 0.000\n",
      "[98,    10] loss: 0.000\n",
      "[98,    20] loss: 0.000\n",
      "[98,    30] loss: 0.000\n",
      "[98,    40] loss: 0.000\n",
      "[98,    50] loss: 0.000\n",
      "[98,    60] loss: 0.000\n",
      "[98,    70] loss: 0.000\n",
      "[98,    80] loss: 0.000\n",
      "[98,    90] loss: 0.000\n",
      "[98,   100] loss: 0.000\n",
      "[98,   110] loss: 0.000\n",
      "[98,   120] loss: 0.000\n",
      "[98,   130] loss: 0.000\n",
      "[98,   140] loss: 0.000\n",
      "[98,   150] loss: 0.000\n",
      "[98,   160] loss: 0.000\n",
      "[98,   170] loss: 0.000\n",
      "[98,   180] loss: 0.000\n",
      "[99,    10] loss: 0.000\n",
      "[99,    20] loss: 0.000\n",
      "[99,    30] loss: 0.000\n",
      "[99,    40] loss: 0.000\n",
      "[99,    50] loss: 0.000\n",
      "[99,    60] loss: 0.000\n",
      "[99,    70] loss: 0.000\n",
      "[99,    80] loss: 0.000\n",
      "[99,    90] loss: 0.000\n",
      "[99,   100] loss: 0.000\n",
      "[99,   110] loss: 0.000\n",
      "[99,   120] loss: 0.000\n",
      "[99,   130] loss: 0.000\n",
      "[99,   140] loss: 0.000\n",
      "[99,   150] loss: 0.000\n",
      "[99,   160] loss: 0.000\n",
      "[99,   170] loss: 0.000\n",
      "[99,   180] loss: 0.000\n",
      "[100,    10] loss: 0.000\n",
      "[100,    20] loss: 0.000\n",
      "[100,    30] loss: 0.000\n",
      "[100,    40] loss: 0.000\n",
      "[100,    50] loss: 0.000\n",
      "[100,    60] loss: 0.000\n",
      "[100,    70] loss: 0.000\n",
      "[100,    80] loss: 0.000\n",
      "[100,    90] loss: 0.000\n",
      "[100,   100] loss: 0.000\n",
      "[100,   110] loss: 0.000\n",
      "[100,   120] loss: 0.000\n",
      "[100,   130] loss: 0.000\n",
      "[100,   140] loss: 0.000\n",
      "[100,   150] loss: 0.000\n",
      "[100,   160] loss: 0.000\n",
      "[100,   170] loss: 0.000\n",
      "[100,   180] loss: 0.000\n",
      "Finished Training\n"
     ]
    }
   ],
   "source": [
    "for epoch in range(100):  # loop over the dataset multiple times\n",
    "\n",
    "    running_loss = 0.0\n",
    "    for i, data in enumerate(train_loader, 0):\n",
    "        # get the inputs; data is a list of [inputs, labels]\n",
    "        inputs, labels = data\n",
    "\n",
    "        # zero the parameter gradients\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        # forward + backward + optimize\n",
    "        outputs = net(inputs)\n",
    "        loss = criterion(outputs, labels)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        # print statistics\n",
    "        running_loss += loss.item()\n",
    "        if i % 10== 9:    # print every 2000 mini-batches\n",
    "            print('[%d, %5d] loss: %.3f' %\n",
    "                  (epoch + 1, i + 1, running_loss / 2000))\n",
    "            running_loss = 0.0\n",
    "\n",
    "print('Finished Training')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GroundTruth:  Food Food Food Food Food Food Food Food Food Food Food Food Food Food Food Food\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "dataiter = iter(test_loader)\n",
    "images, labels = dataiter.next()\n",
    "\n",
    "# print images\n",
    "imshow(torchvision.utils.make_grid(images))\n",
    "print('GroundTruth: ', ' '.join('%s' % classes[labels[j]] for j in range(16)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "outputs = net(images) #lets see what our neural network thinks of\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted:   Food  Food  Food  Food  Food  Food  Food  Food  Food  Food  Food  Food  Food  Food  Food  Food\n"
     ]
    }
   ],
   "source": [
    "_, predicted = torch.max(outputs, 1)\n",
    "\n",
    "print('Predicted: ', ' '.join('%5s' % classes[predicted[j]]\n",
    "                              for j in range(16)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of the network on the 1000 test images: 85 %\n"
     ]
    }
   ],
   "source": [
    "correct = 0\n",
    "total = 0\n",
    "with torch.no_grad():\n",
    "    for data in test_loader:\n",
    "        images, labels = data\n",
    "        outputs = net(images)\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "        total += labels.size(0)\n",
    "        correct += (predicted == labels).sum().item()\n",
    "\n",
    "print('Accuracy of the network on the 1000 test images: %d %%' % (\n",
    "    100 * correct / total))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of  Food : 100 %\n",
      "Accuracy of Non-Food :  0 %\n"
     ]
    }
   ],
   "source": [
    "class_correct = list(0. for i in range(4))\n",
    "class_total = list(0. for i in range(4))\n",
    "with torch.no_grad():\n",
    "    for data in test_loader:\n",
    "        images, labels = data\n",
    "        outputs = net(images)\n",
    "        _, predicted = torch.max(outputs, 1)\n",
    "        c = (predicted == labels).squeeze()\n",
    "        for i in range(3):\n",
    "            label = labels[i]\n",
    "            class_correct[label] += c[i].item()\n",
    "            class_total[label] += 1\n",
    "\n",
    "\n",
    "for i in range(2):\n",
    "    print('Accuracy of %5s : %2d %%' % (classes[i], 100* class_correct[i] / class_total[i]))"
   ]
  },
  {
   "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
}