fashion_mnist.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import torch\n",
    "import torchvision\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as func\n",
    "import torchvision.transforms as transforms\n",
    "from torch.autograd import Variable\n",
    "from torchvision import models\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "from sklearn.metrics import confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define the path of the trai  and test data\n",
    "train_path = 'fashion-mnist_train.csv'\n",
    "test_path = 'fashion-mnist_test.csv'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load the dataset \n",
    "class fashion_mnist(Dataset):\n",
    "    def __init__(self,path,transform=None):\n",
    "        self.data = pd.read_csv(path)\n",
    "        self.transform = transform\n",
    "        img, label = [], []\n",
    "        for i in self.data.values:\n",
    "            img.append(i[1:])\n",
    "            label.append(i[0])\n",
    "        self.label = np.asarray(label)\n",
    "        self.img = np.asarray(img).reshape(-1,28,28).astype('float32')\n",
    "        \n",
    "    def __getitem__(self,index):\n",
    "        image, label = self.img[index], self.label[index]\n",
    "        \n",
    "        if self.transform is not None:\n",
    "            image = self.transform(image)\n",
    "            \n",
    "        return image,label\n",
    "            \n",
    "    def __len__(self):\n",
    "        return len(self.img)\n",
    "            \n",
    "        return image, label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load train and test data\n",
    "transform = transforms.Compose([\n",
    "   transforms.ToTensor() \n",
    "])\n",
    "\n",
    "train_data = fashion_mnist(train_path,transform=transform)\n",
    "test_data = fashion_mnist(test_path,transform=transform)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define the hyperparameters and load the dataloader\n",
    "batch_size = 60\n",
    "#n_iters = 5000\n",
    "learning_rate = 0.01\n",
    "momentum = 0.3\n",
    "#num_epochs = int(n_iters/(train_data.__len__()/batch_size))\n",
    "\n",
    "# dataloader\n",
    "train_loader = DataLoader(train_data,batch_size=batch_size,shuffle=True)\n",
    "test_loader = DataLoader(test_data,batch_size=batch_size,shuffle =True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# the tags in the datasets is given via a numbering system, below is the dict of the number and is corresponding tag\n",
    "tags = {0 :'T-shirt/top',\n",
    "                  1 :'Trouser',\n",
    "                  2 :'Pullover',\n",
    "                  3 :'Dress',\n",
    "                  4 :'Coat',\n",
    "                  5 :'Sandal',\n",
    "                  6 :'Shirt',\n",
    "                  7 :'Sneaker',\n",
    "                  8 :'Bag',\n",
    "                  9 :'Ankle boot'}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# iterate through the dataloader\n",
    "_,(img,label) = next(enumerate(train_loader))\n",
    "fig = plt.figure(figsize=(7,7))\n",
    "for i in range(6):\n",
    "    plt.subplot(3,2,i+1)\n",
    "    plt.imshow(img[i][0])\n",
    "    plt.axis('off')\n",
    "    plt.title('{}'.format(tags[int(label[i].numpy())]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "# build the convolution neural network\n",
    "class fashion_model(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(fashion_model,self).__init__()\n",
    "        \n",
    "        # start with the bulind the model with convolution layers\n",
    "        self.layer1 = nn.Sequential(\n",
    "            nn.Conv2d(in_channels=1,out_channels=16,kernel_size =5, stride=1,padding=2),\n",
    "            nn.BatchNorm2d(16),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(kernel_size=2,stride=2)\n",
    "        )\n",
    "        self.layer2 = nn.Sequential(\n",
    "            nn.Conv2d(in_channels=16, out_channels=32,kernel_size=5,stride=1,padding=2),\n",
    "            nn.BatchNorm2d(32),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(kernel_size=2,stride=2)\n",
    "        )\n",
    "        \n",
    "        self.fc1 = nn.Linear(7*7*32,1000)\n",
    "        self.fc2 = nn.Linear(1000,10)\n",
    "        \n",
    "    def forward(self,x):\n",
    "        out = self.layer1(x)\n",
    "        out = self.layer2(out)\n",
    "        out = out.reshape(out.size(0),-1)\n",
    "        out = nn.ReLU()(out)\n",
    "        out = self.fc1(out)\n",
    "        out = nn.ReLU()(out)\n",
    "        out = self.fc2(out)\n",
    "        return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = fashion_model()\n",
    "error = nn.CrossEntropyLoss()\n",
    "optimizer = optim.Adam(model.parameters(),lr=learning_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fashion_model(\n",
      "  (layer1): Sequential(\n",
      "    (0): Conv2d(1, 16, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))\n",
      "    (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (2): ReLU()\n",
      "    (3): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): Conv2d(16, 32, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))\n",
      "    (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    (2): ReLU()\n",
      "    (3): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (fc1): Linear(in_features=1568, out_features=1000, bias=True)\n",
      "  (fc2): Linear(in_features=1000, out_features=10, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "print(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(num_epochs):\n",
    "    \n",
    "    model.train()\n",
    "    correct = 0\n",
    "    total = 0\n",
    "    train_error=0\n",
    "\n",
    "    for batch_id, (img,label) in enumerate(train_loader):\n",
    "        img, label = Variable(img), Variable(label)\n",
    "        optimizer.zero_grad()\n",
    "        output = model(img)\n",
    "        predict = torch.max(output.data,1)[1]\n",
    "        correct += (predict==label).sum()\n",
    "        total += len(label)\n",
    "            \n",
    "        # compute the loss\n",
    "        loss = error(output, label)\n",
    "        train_error = loss.item()\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "            \n",
    "        if (batch_id+1)%100 ==0:\n",
    "            print('Train Epoch: {} [{}/{} ({:.0f}%)]\\tLoss: {:.6f} \\t Accuracy: {}%'\n",
    "                    .format(epoch,(batch_id+1)*len(label),len(train_loader.dataset),\n",
    "                        100.0*(batch_id+1)/len(train_loader),loss.item(),100*correct/total))\n",
    "                \n",
    "            \n",
    "    \n",
    "    train_loss.append(train_error/int(len(train_loader)))\n",
    "    train_acc.append(100*correct/total)\n",
    "          \n",
    "        \n",
    "    # append the loss, accuracy and number of iterations\n",
    "          "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "def test():\n",
    "    model.eval()\n",
    "    loss = 0\n",
    "    correct = 0\n",
    "    total = 0\n",
    "    \n",
    "    for i, (img,label) in enumerate(test_loader):\n",
    "        img, label = Variable(img), Variable(label)\n",
    "        \n",
    "        output = model(img)\n",
    "        predict = torch.max(output.data,1)[1]\n",
    "        loss_error = error(output,label)\n",
    "        loss += loss_error.item()\n",
    "        total += len(label)\n",
    "        correct += (predict==label).sum()\n",
    "        \n",
    "    loss /= len(test_loader.dataset)\n",
    "    test_loss.append(loss)\n",
    "    test_acc.append(100*correct/total)\n",
    "    print('\\nAverage Validation loss: {:.5f}\\t Accuracy: {}'.format(loss,100*correct/total))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 0 [6000/60000 (10%)]\tLoss: 0.678437 \t Accuracy: 63%\n",
      "Train Epoch: 0 [12000/60000 (20%)]\tLoss: 0.595381 \t Accuracy: 70%\n",
      "Train Epoch: 0 [18000/60000 (30%)]\tLoss: 0.502638 \t Accuracy: 74%\n",
      "Train Epoch: 0 [24000/60000 (40%)]\tLoss: 0.486310 \t Accuracy: 76%\n",
      "Train Epoch: 0 [30000/60000 (50%)]\tLoss: 0.349632 \t Accuracy: 78%\n",
      "Train Epoch: 0 [36000/60000 (60%)]\tLoss: 0.690727 \t Accuracy: 79%\n",
      "Train Epoch: 0 [42000/60000 (70%)]\tLoss: 0.458854 \t Accuracy: 80%\n",
      "Train Epoch: 0 [48000/60000 (80%)]\tLoss: 0.448559 \t Accuracy: 81%\n",
      "Train Epoch: 0 [54000/60000 (90%)]\tLoss: 0.463590 \t Accuracy: 81%\n",
      "Train Epoch: 0 [60000/60000 (100%)]\tLoss: 0.388342 \t Accuracy: 82%\n",
      "\n",
      "Average Validation loss: 0.00608\t Accuracy: 86\n",
      "Train Epoch: 1 [6000/60000 (10%)]\tLoss: 0.511132 \t Accuracy: 88%\n",
      "Train Epoch: 1 [12000/60000 (20%)]\tLoss: 0.351023 \t Accuracy: 88%\n",
      "Train Epoch: 1 [18000/60000 (30%)]\tLoss: 0.325936 \t Accuracy: 88%\n",
      "Train Epoch: 1 [24000/60000 (40%)]\tLoss: 0.376291 \t Accuracy: 88%\n",
      "Train Epoch: 1 [30000/60000 (50%)]\tLoss: 0.146977 \t Accuracy: 88%\n",
      "Train Epoch: 1 [36000/60000 (60%)]\tLoss: 0.190094 \t Accuracy: 88%\n",
      "Train Epoch: 1 [42000/60000 (70%)]\tLoss: 0.414773 \t Accuracy: 88%\n",
      "Train Epoch: 1 [48000/60000 (80%)]\tLoss: 0.327020 \t Accuracy: 88%\n",
      "Train Epoch: 1 [54000/60000 (90%)]\tLoss: 0.332643 \t Accuracy: 88%\n",
      "Train Epoch: 1 [60000/60000 (100%)]\tLoss: 0.332369 \t Accuracy: 88%\n",
      "\n",
      "Average Validation loss: 0.00469\t Accuracy: 89\n",
      "Train Epoch: 2 [6000/60000 (10%)]\tLoss: 0.166166 \t Accuracy: 90%\n",
      "Train Epoch: 2 [12000/60000 (20%)]\tLoss: 0.237915 \t Accuracy: 90%\n",
      "Train Epoch: 2 [18000/60000 (30%)]\tLoss: 0.163561 \t Accuracy: 90%\n",
      "Train Epoch: 2 [24000/60000 (40%)]\tLoss: 0.148741 \t Accuracy: 90%\n",
      "Train Epoch: 2 [30000/60000 (50%)]\tLoss: 0.187764 \t Accuracy: 90%\n",
      "Train Epoch: 2 [36000/60000 (60%)]\tLoss: 0.258362 \t Accuracy: 89%\n",
      "Train Epoch: 2 [42000/60000 (70%)]\tLoss: 0.387667 \t Accuracy: 90%\n",
      "Train Epoch: 2 [48000/60000 (80%)]\tLoss: 0.178120 \t Accuracy: 90%\n",
      "Train Epoch: 2 [54000/60000 (90%)]\tLoss: 0.436027 \t Accuracy: 90%\n",
      "Train Epoch: 2 [60000/60000 (100%)]\tLoss: 0.291990 \t Accuracy: 90%\n",
      "\n",
      "Average Validation loss: 0.00450\t Accuracy: 90\n",
      "Train Epoch: 3 [6000/60000 (10%)]\tLoss: 0.230587 \t Accuracy: 91%\n",
      "Train Epoch: 3 [12000/60000 (20%)]\tLoss: 0.514711 \t Accuracy: 91%\n",
      "Train Epoch: 3 [18000/60000 (30%)]\tLoss: 0.180741 \t Accuracy: 91%\n",
      "Train Epoch: 3 [24000/60000 (40%)]\tLoss: 0.261741 \t Accuracy: 91%\n",
      "Train Epoch: 3 [30000/60000 (50%)]\tLoss: 0.121591 \t Accuracy: 91%\n",
      "Train Epoch: 3 [36000/60000 (60%)]\tLoss: 0.521753 \t Accuracy: 91%\n",
      "Train Epoch: 3 [42000/60000 (70%)]\tLoss: 0.120659 \t Accuracy: 91%\n",
      "Train Epoch: 3 [48000/60000 (80%)]\tLoss: 0.315639 \t Accuracy: 91%\n",
      "Train Epoch: 3 [54000/60000 (90%)]\tLoss: 0.185204 \t Accuracy: 91%\n",
      "Train Epoch: 3 [60000/60000 (100%)]\tLoss: 0.186985 \t Accuracy: 91%\n",
      "\n",
      "Average Validation loss: 0.00434\t Accuracy: 90\n",
      "Train Epoch: 4 [6000/60000 (10%)]\tLoss: 0.159852 \t Accuracy: 92%\n",
      "Train Epoch: 4 [12000/60000 (20%)]\tLoss: 0.104220 \t Accuracy: 92%\n",
      "Train Epoch: 4 [18000/60000 (30%)]\tLoss: 0.289373 \t Accuracy: 92%\n",
      "Train Epoch: 4 [24000/60000 (40%)]\tLoss: 0.265900 \t Accuracy: 91%\n",
      "Train Epoch: 4 [30000/60000 (50%)]\tLoss: 0.125543 \t Accuracy: 91%\n",
      "Train Epoch: 4 [36000/60000 (60%)]\tLoss: 0.340564 \t Accuracy: 91%\n",
      "Train Epoch: 4 [42000/60000 (70%)]\tLoss: 0.325535 \t Accuracy: 91%\n",
      "Train Epoch: 4 [48000/60000 (80%)]\tLoss: 0.165920 \t Accuracy: 91%\n",
      "Train Epoch: 4 [54000/60000 (90%)]\tLoss: 0.296033 \t Accuracy: 91%\n",
      "Train Epoch: 4 [60000/60000 (100%)]\tLoss: 0.376983 \t Accuracy: 91%\n",
      "\n",
      "Average Validation loss: 0.00417\t Accuracy: 90\n",
      "Train Epoch: 5 [6000/60000 (10%)]\tLoss: 0.325795 \t Accuracy: 92%\n",
      "Train Epoch: 5 [12000/60000 (20%)]\tLoss: 0.089896 \t Accuracy: 92%\n",
      "Train Epoch: 5 [18000/60000 (30%)]\tLoss: 0.272921 \t Accuracy: 92%\n",
      "Train Epoch: 5 [24000/60000 (40%)]\tLoss: 0.255745 \t Accuracy: 92%\n",
      "Train Epoch: 5 [30000/60000 (50%)]\tLoss: 0.162977 \t Accuracy: 92%\n",
      "Train Epoch: 5 [36000/60000 (60%)]\tLoss: 0.204284 \t Accuracy: 92%\n",
      "Train Epoch: 5 [42000/60000 (70%)]\tLoss: 0.170659 \t Accuracy: 92%\n",
      "Train Epoch: 5 [48000/60000 (80%)]\tLoss: 0.157017 \t Accuracy: 91%\n",
      "Train Epoch: 5 [54000/60000 (90%)]\tLoss: 0.121519 \t Accuracy: 91%\n",
      "Train Epoch: 5 [60000/60000 (100%)]\tLoss: 0.113946 \t Accuracy: 91%\n",
      "\n",
      "Average Validation loss: 0.00427\t Accuracy: 90\n",
      "Train Epoch: 6 [6000/60000 (10%)]\tLoss: 0.116774 \t Accuracy: 92%\n",
      "Train Epoch: 6 [12000/60000 (20%)]\tLoss: 0.114645 \t Accuracy: 92%\n",
      "Train Epoch: 6 [18000/60000 (30%)]\tLoss: 0.215252 \t Accuracy: 92%\n",
      "Train Epoch: 6 [24000/60000 (40%)]\tLoss: 0.183527 \t Accuracy: 92%\n",
      "Train Epoch: 6 [30000/60000 (50%)]\tLoss: 0.116037 \t Accuracy: 92%\n",
      "Train Epoch: 6 [36000/60000 (60%)]\tLoss: 0.113504 \t Accuracy: 92%\n",
      "Train Epoch: 6 [42000/60000 (70%)]\tLoss: 0.174883 \t Accuracy: 92%\n",
      "Train Epoch: 6 [48000/60000 (80%)]\tLoss: 0.125918 \t Accuracy: 92%\n",
      "Train Epoch: 6 [54000/60000 (90%)]\tLoss: 0.161334 \t Accuracy: 92%\n",
      "Train Epoch: 6 [60000/60000 (100%)]\tLoss: 0.235268 \t Accuracy: 92%\n",
      "\n",
      "Average Validation loss: 0.00422\t Accuracy: 91\n",
      "Train Epoch: 7 [6000/60000 (10%)]\tLoss: 0.170990 \t Accuracy: 92%\n",
      "Train Epoch: 7 [12000/60000 (20%)]\tLoss: 0.322544 \t Accuracy: 92%\n",
      "Train Epoch: 7 [18000/60000 (30%)]\tLoss: 0.095232 \t Accuracy: 92%\n",
      "Train Epoch: 7 [24000/60000 (40%)]\tLoss: 0.175236 \t Accuracy: 92%\n",
      "Train Epoch: 7 [30000/60000 (50%)]\tLoss: 0.178418 \t Accuracy: 92%\n",
      "Train Epoch: 7 [36000/60000 (60%)]\tLoss: 0.240939 \t Accuracy: 92%\n",
      "Train Epoch: 7 [42000/60000 (70%)]\tLoss: 0.135438 \t Accuracy: 92%\n",
      "Train Epoch: 7 [48000/60000 (80%)]\tLoss: 0.109649 \t Accuracy: 92%\n",
      "Train Epoch: 7 [54000/60000 (90%)]\tLoss: 0.247417 \t Accuracy: 92%\n",
      "Train Epoch: 7 [60000/60000 (100%)]\tLoss: 0.309840 \t Accuracy: 92%\n",
      "\n",
      "Average Validation loss: 0.00411\t Accuracy: 91\n",
      "Train Epoch: 8 [6000/60000 (10%)]\tLoss: 0.137649 \t Accuracy: 93%\n",
      "Train Epoch: 8 [12000/60000 (20%)]\tLoss: 0.106223 \t Accuracy: 93%\n",
      "Train Epoch: 8 [18000/60000 (30%)]\tLoss: 0.146944 \t Accuracy: 93%\n",
      "Train Epoch: 8 [24000/60000 (40%)]\tLoss: 0.226813 \t Accuracy: 93%\n",
      "Train Epoch: 8 [30000/60000 (50%)]\tLoss: 0.188333 \t Accuracy: 93%\n",
      "Train Epoch: 8 [36000/60000 (60%)]\tLoss: 0.191070 \t Accuracy: 93%\n",
      "Train Epoch: 8 [42000/60000 (70%)]\tLoss: 0.210472 \t Accuracy: 93%\n",
      "Train Epoch: 8 [48000/60000 (80%)]\tLoss: 0.154720 \t Accuracy: 93%\n",
      "Train Epoch: 8 [54000/60000 (90%)]\tLoss: 0.157909 \t Accuracy: 92%\n",
      "Train Epoch: 8 [60000/60000 (100%)]\tLoss: 0.153908 \t Accuracy: 92%\n",
      "\n",
      "Average Validation loss: 0.00424\t Accuracy: 91\n",
      "Train Epoch: 9 [6000/60000 (10%)]\tLoss: 0.225151 \t Accuracy: 93%\n",
      "Train Epoch: 9 [12000/60000 (20%)]\tLoss: 0.105567 \t Accuracy: 93%\n",
      "Train Epoch: 9 [18000/60000 (30%)]\tLoss: 0.134988 \t Accuracy: 93%\n",
      "Train Epoch: 9 [24000/60000 (40%)]\tLoss: 0.134708 \t Accuracy: 93%\n",
      "Train Epoch: 9 [30000/60000 (50%)]\tLoss: 0.171048 \t Accuracy: 93%\n",
      "Train Epoch: 9 [36000/60000 (60%)]\tLoss: 0.153232 \t Accuracy: 93%\n",
      "Train Epoch: 9 [42000/60000 (70%)]\tLoss: 0.105727 \t Accuracy: 93%\n",
      "Train Epoch: 9 [48000/60000 (80%)]\tLoss: 0.145457 \t Accuracy: 93%\n",
      "Train Epoch: 9 [54000/60000 (90%)]\tLoss: 0.408614 \t Accuracy: 93%\n",
      "Train Epoch: 9 [60000/60000 (100%)]\tLoss: 0.123366 \t Accuracy: 93%\n",
      "\n",
      "Average Validation loss: 0.00433\t Accuracy: 91\n"
     ]
    }
   ],
   "source": [
    "n_epochs = 10\n",
    "train_loss = []\n",
    "train_acc = []\n",
    "test_loss = []\n",
    "test_acc = []\n",
    "for epoch in range(n_epochs):\n",
    "    train(epoch)\n",
    "    test()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x220afc822e8>]"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.plot(train_loss)\n",
    "plt.plot(test_loss)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x220b33ea470>]"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.plot(train_acc)\n",
    "plt.plot(test_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save the model\n",
    "torch.save(model.state_dict(),'fashion_mnist.pth')\n",
    "torch.save(optimizer.state_dict,'optimizer.pth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# load and predict from the model\n",
    "_,(image,label) = next(enumerate(test_loader))\n",
    "with torch.no_grad():\n",
    "    output = model(image)\n",
    "fig = plt.figure(figsize=(10,10))\n",
    "for i in range(6):\n",
    "    plt.subplot(3,2,i+1)\n",
    "    plt.imshow(image[i][0])\n",
    "    pred = output.data.max(1,keepdim=True)[1][i].item()\n",
    "    plt.axis('off')\n",
    "    plt.title('Label:{} || Predicted: {}'.format(tags[int(label[i].numpy())],tags[int(pred)]))"
   ]
  },
  {
   "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": 4
}