fashion_mnist_classification_nn_pytorch.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fashion-MNIST Classification using Neural Network\n",
    "\n",
    "In this notebook, we'll build a neural network to classify Fashion-MNIST images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "from torchvision import datasets, transforms\n",
    "import matplotlib.pyplot as plt\n",
    "from collections import OrderedDict\n",
    "\n",
    "# Download training and testing data\n",
    "transform = transforms.Compose([transforms.ToTensor(),\n",
    "                                transforms.Normalize((0.5,), (0.5,))])\n",
    "train_ds = datasets.FashionMNIST('F_MNIST_data', download=True, train=True, transform=transform)\n",
    "test_ds = datasets.FashionMNIST('F_MNIST_data', download=True, train=False, transform=transform)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(12000, 48000)"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# split train set into training (80%) and validation set (20%)\n",
    "train_num = len(train_ds)\n",
    "indices = list(range(train_num))\n",
    "np.random.shuffle(indices)\n",
    "split = int(np.floor(0.2 * train_num))\n",
    "val_idx, train_idx = indices[:split], indices[split:]\n",
    "len(val_idx), len(train_idx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [],
   "source": [
    "# prepare dataloaders\n",
    "train_sampler = torch.utils.data.sampler.SubsetRandomSampler(train_idx)\n",
    "train_dl = torch.utils.data.DataLoader(train_ds, batch_size=64, sampler=train_sampler)\n",
    "val_sampler = torch.utils.data.sampler.SubsetRandomSampler(val_idx)\n",
    "val_dl = torch.utils.data.DataLoader(train_ds, batch_size=64, sampler=val_sampler)\n",
    "test_dl = torch.utils.data.DataLoader(test_ds, batch_size=64, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([1, 28, 28]) torch.Size([64])\n",
      "Pullover\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 248,
       "width": 251
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "image, label = next(iter(train_dl))\n",
    "print(image[0].shape, label.shape)\n",
    "desc = ['T-shirt/top','Trouser','Pullover','Dress','Coat','Sandal','Shirt','Sneaker','Bag','Ankle Boot']\n",
    "print(desc[label[0].item()])\n",
    "plt.imshow(image[0].numpy().squeeze(), cmap='gray');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build the network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "def network():\n",
    "    model = nn.Sequential(OrderedDict([('fc1', nn.Linear(784, 128)),\n",
    "                                       ('relu1', nn.ReLU()),\n",
    "                                       ('drop1', nn.Dropout(0.25)),                                       \n",
    "                                       ('fc2', nn.Linear(128, 64)),\n",
    "                                       ('relu2', nn.ReLU()),\n",
    "                                       ('drop1', nn.Dropout(0.25)),                                       \n",
    "                                       ('output', nn.Linear(64, 10)),\n",
    "                                       ('logsoftmax', nn.LogSoftmax(dim=1))]))\n",
    "    # Use GPU if available\n",
    "    device = 'cuda' if torch.cuda.is_available() else 'cpu'\n",
    "    model = model.to(device)\n",
    "\n",
    "    # define the criterion and optimizer\n",
    "    loss_fn = nn.NLLLoss()\n",
    "    optimizer = optim.Adam(model.parameters(), lr=0.003)\n",
    "\n",
    "    return model, loss_fn, optimizer, device                      "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sequential(\n",
      "  (fc1): Linear(in_features=784, out_features=128, bias=True)\n",
      "  (relu1): ReLU()\n",
      "  (drop1): Dropout(p=0.25, inplace=False)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (relu2): ReLU()\n",
      "  (output): Linear(in_features=64, out_features=10, bias=True)\n",
      "  (logsoftmax): LogSoftmax(dim=1)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "model, loss_fn, optimizer, device = network()\n",
    "print(model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train the network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_validate(model, loss_fn, optimizer, trainloader, testloader, device, n_epochs=25):\n",
    "    train_losses = []\n",
    "    test_losses = []\n",
    "    for epoch in range(n_epochs):\n",
    "        # Set mode to training - Dropouts will be used here\n",
    "        model.train()\n",
    "        train_epoch_loss = 0\n",
    "        for images, labels in trainloader:\n",
    "            images, labels = images.to(device), labels.to(device)\n",
    "            # flatten the images to batch_size x 784\n",
    "            images = images.view(images.shape[0], -1)\n",
    "            # forward pass\n",
    "            outputs = model(images)\n",
    "            # backpropogation\n",
    "            train_batch_loss = loss_fn(outputs, labels)\n",
    "            optimizer.zero_grad()\n",
    "            train_batch_loss.backward()\n",
    "            # Weight updates\n",
    "            optimizer.step()\n",
    "            train_epoch_loss += train_batch_loss.item()\n",
    "        else:\n",
    "            # One epoch of training complete\n",
    "            # calculate average training epoch loss\n",
    "            train_epoch_loss = train_epoch_loss/len(trainloader)\n",
    "\n",
    "            # Now Validate on testset\n",
    "            with torch.no_grad():\n",
    "                test_epoch_acc = 0\n",
    "                test_epoch_loss = 0\n",
    "                # Set mode to eval - Dropouts will NOT be used here\n",
    "                model.eval()\n",
    "                for images, labels in testloader:\n",
    "                    images, labels = images.to(device), labels.to(device)                    \n",
    "                    # flatten images to batch_size x 784\n",
    "                    images = images.view(images.shape[0], -1)\n",
    "                    # make predictions \n",
    "                    test_outputs = model(images)\n",
    "                    # calculate test loss\n",
    "                    test_batch_loss = loss_fn(test_outputs, labels)\n",
    "                    test_epoch_loss += test_batch_loss\n",
    "                    \n",
    "                    # get probabilities, extract the class associated with highest probability\n",
    "                    proba = torch.exp(test_outputs)\n",
    "                    _, pred_labels = proba.topk(1, dim=1)\n",
    "                    \n",
    "                    # compare actual labels and predicted labels\n",
    "                    result = pred_labels == labels.view(pred_labels.shape)\n",
    "                    batch_acc = torch.mean(result.type(torch.FloatTensor))\n",
    "                    test_epoch_acc += batch_acc.item()\n",
    "                else:\n",
    "                    # One epoch of training and validation done\n",
    "                    # calculate average testing epoch loss\n",
    "                    test_epoch_loss = test_epoch_loss/len(testloader)\n",
    "                    # calculate accuracy as correct_pred/total_samples\n",
    "                    test_epoch_acc = test_epoch_acc/len(testloader)\n",
    "                    # save epoch losses for plotting\n",
    "                    train_losses.append(train_epoch_loss)\n",
    "                    test_losses.append(test_epoch_loss)\n",
    "                    # print stats for this epoch\n",
    "                    print(f'Epoch: {epoch} -> train_loss: {train_epoch_loss:.19f}, val_loss: {test_epoch_loss:.19f}, ',\n",
    "                          f'val_acc: {test_epoch_acc*100:.2f}%')\n",
    "    # Finally plot losses\n",
    "    plt.plot(train_losses, label='train-loss')\n",
    "    plt.plot(test_losses, label='val-loss')\n",
    "    plt.legend()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0 -> train_loss: 0.5692988239328066191, val_loss: 0.4430967867374420166,  val_acc: 83.27%\n",
      "Epoch: 1 -> train_loss: 0.4549953033725420704, val_loss: 0.4073663055896759033,  val_acc: 84.97%\n",
      "Epoch: 2 -> train_loss: 0.4211076561212539482, val_loss: 0.3762904107570648193,  val_acc: 86.11%\n",
      "Epoch: 3 -> train_loss: 0.4046927829583485958, val_loss: 0.3766101896762847900,  val_acc: 86.46%\n",
      "Epoch: 4 -> train_loss: 0.3892695648074150361, val_loss: 0.3516367673873901367,  val_acc: 87.34%\n",
      "Epoch: 5 -> train_loss: 0.3776730005542437429, val_loss: 0.3703932464122772217,  val_acc: 86.71%\n",
      "Epoch: 6 -> train_loss: 0.3724320774773756670, val_loss: 0.3486533761024475098,  val_acc: 87.37%\n",
      "Epoch: 7 -> train_loss: 0.3617128521899382054, val_loss: 0.3633341491222381592,  val_acc: 87.08%\n",
      "Epoch: 8 -> train_loss: 0.3569285793701807430, val_loss: 0.3384532332420349121,  val_acc: 87.58%\n",
      "Epoch: 9 -> train_loss: 0.3474347092111905178, val_loss: 0.3369818329811096191,  val_acc: 87.88%\n",
      "Epoch: 10 -> train_loss: 0.3383618565599123551, val_loss: 0.3385397493839263916,  val_acc: 87.97%\n",
      "Epoch: 11 -> train_loss: 0.3360619519948959133, val_loss: 0.3329190611839294434,  val_acc: 87.97%\n",
      "Epoch: 12 -> train_loss: 0.3340372469524542365, val_loss: 0.3389228284358978271,  val_acc: 88.00%\n",
      "Epoch: 13 -> train_loss: 0.3269288233717282388, val_loss: 0.3449897170066833496,  val_acc: 87.77%\n",
      "Epoch: 14 -> train_loss: 0.3249198505381742930, val_loss: 0.3381777107715606689,  val_acc: 87.75%\n",
      "Epoch: 15 -> train_loss: 0.3214456404149532154, val_loss: 0.3334679007530212402,  val_acc: 87.97%\n",
      "Epoch: 16 -> train_loss: 0.3177473722100258025, val_loss: 0.3333184421062469482,  val_acc: 87.75%\n",
      "Epoch: 17 -> train_loss: 0.3098344492912292747, val_loss: 0.3366726040840148926,  val_acc: 88.12%\n",
      "Epoch: 18 -> train_loss: 0.3078885127405325828, val_loss: 0.3318108916282653809,  val_acc: 88.41%\n",
      "Epoch: 19 -> train_loss: 0.3075984100500742668, val_loss: 0.3197179138660430908,  val_acc: 88.56%\n",
      "Epoch: 20 -> train_loss: 0.3024848873019218565, val_loss: 0.3329302072525024414,  val_acc: 88.16%\n",
      "Epoch: 21 -> train_loss: 0.3023494395713011151, val_loss: 0.3277964591979980469,  val_acc: 88.35%\n",
      "Epoch: 22 -> train_loss: 0.2963562680284182460, val_loss: 0.3371585309505462646,  val_acc: 88.25%\n",
      "Epoch: 23 -> train_loss: 0.2981447652975718343, val_loss: 0.3276244103908538818,  val_acc: 88.55%\n",
      "Epoch: 24 -> train_loss: 0.2937244445780913260, val_loss: 0.3280715346336364746,  val_acc: 88.67%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 248,
       "width": 382
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Train and validate\n",
    "train_validate(model, loss_fn, optimizer, train_dl, val_dl, device)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Predict a single image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 936x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 424,
       "width": 897
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format = 'retina'\n",
    "\n",
    "# Test out the network!\n",
    "dataiter = iter(test_dl)\n",
    "images, labels = dataiter.next()\n",
    "images, labels = images.to(device), labels.to(device)\n",
    "index = 49\n",
    "img, label = images[index], labels[index]\n",
    "# Convert 2D image to 1D vector\n",
    "img = img.view(img.shape[0], -1)\n",
    "\n",
    "# Calculate the class probabilities (softmax) for img\n",
    "proba = torch.exp(model(img))\n",
    "\n",
    "# Plot the image and probabilities\n",
    "desc = ['T-shirt/top','Trouser','Pullover','Dress','Coat','Sandal','Shirt','Sneaker','Bag','Ankle Boot']\n",
    "fig, (ax1, ax2) =  plt.subplots(figsize=(13, 6), nrows=1, ncols=2)\n",
    "ax1.axis('off')\n",
    "ax1.imshow(images[index].cpu().numpy().squeeze())\n",
    "ax1.set_title(desc[label.item()])\n",
    "ax2.bar(range(10), proba.detach().cpu().numpy().squeeze())\n",
    "ax2.set_xticks(range(10))\n",
    "ax2.set_xticklabels(desc, size='small')\n",
    "ax2.set_title('Predicted Probabilities')\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Validate on test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy: 87.42%\n"
     ]
    }
   ],
   "source": [
    "# Validate\n",
    "with torch.no_grad():\n",
    "    batch_acc = []\n",
    "    model.eval()\n",
    "    for images, labels in test_dl:\n",
    "        images, labels = images.to(device), labels.to(device)\n",
    "        # flatten images to batch_size x 784\n",
    "        images = images.view(images.shape[0], -1)\n",
    "        # make predictions and get probabilities\n",
    "        proba = torch.exp(model(images))\n",
    "        # extract the class associted with highest probability\n",
    "        _, pred_labels = proba.topk(1, dim=1)\n",
    "        # compare actual labels and predicted labels\n",
    "        result = pred_labels == labels.view(pred_labels.shape)\n",
    "        acc = torch.mean(result.type(torch.FloatTensor))\n",
    "        batch_acc.append(acc.item())\n",
    "    else:\n",
    "        print(f'Test Accuracy: {torch.mean(torch.tensor(batch_acc))*100:.2f}%')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## More powerful model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Redefine network with dropout layers in between\n",
    "def network():\n",
    "    model = nn.Sequential(OrderedDict([('fc1', nn.Linear(784, 392)),\n",
    "                                       ('relu1', nn.ReLU()),\n",
    "                                       ('drop1', nn.Dropout(0.25)),\n",
    "                                       ('fc12', nn.Linear(392, 196)),\n",
    "                                       ('relu2', nn.ReLU()),\n",
    "                                       ('drop2', nn.Dropout(0.25)),\n",
    "                                       ('fc3', nn.Linear(196, 98)),\n",
    "                                       ('relu3', nn.ReLU()),\n",
    "                                       ('drop3', nn.Dropout(0.25)),                                       \n",
    "                                       ('fc4', nn.Linear(98, 49)),\n",
    "                                       ('relu4', nn.ReLU()),\n",
    "                                       ('output', nn.Linear(49, 10)),\n",
    "                                       ('logsoftmax', nn.LogSoftmax(dim=1))]))\n",
    "    \n",
    "    # Use GPU if available\n",
    "    device = 'cuda' if torch.cuda.is_available() else 'cpu'\n",
    "    model = model.to(device)\n",
    "\n",
    "    # define the criterion and optimizer\n",
    "    loss_fn = nn.NLLLoss()\n",
    "    optimizer = optim.Adam(model.parameters(), lr=0.0007)\n",
    "\n",
    "    return model, loss_fn, optimizer, device       "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Sequential(\n",
       "  (fc1): Linear(in_features=784, out_features=392, bias=True)\n",
       "  (relu1): ReLU()\n",
       "  (drop1): Dropout(p=0.25, inplace=False)\n",
       "  (fc12): Linear(in_features=392, out_features=196, bias=True)\n",
       "  (relu2): ReLU()\n",
       "  (drop2): Dropout(p=0.25, inplace=False)\n",
       "  (fc3): Linear(in_features=196, out_features=98, bias=True)\n",
       "  (relu3): ReLU()\n",
       "  (drop3): Dropout(p=0.25, inplace=False)\n",
       "  (fc4): Linear(in_features=98, out_features=49, bias=True)\n",
       "  (relu4): ReLU()\n",
       "  (output): Linear(in_features=49, out_features=10, bias=True)\n",
       "  (logsoftmax): LogSoftmax(dim=1)\n",
       ")"
      ]
     },
     "execution_count": 132,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model, loss_fn, optimizer, device = network()\n",
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0 -> train_loss: 0.6715717662970225321, val_loss: 0.4384306371212005615,  val_acc: 83.58%\n",
      "Epoch: 1 -> train_loss: 0.4610257453123728366, val_loss: 0.4046924412250518799,  val_acc: 84.90%\n",
      "Epoch: 2 -> train_loss: 0.4132076732317606638, val_loss: 0.3710882067680358887,  val_acc: 86.29%\n",
      "Epoch: 3 -> train_loss: 0.3874667741258939202, val_loss: 0.3711968660354614258,  val_acc: 86.47%\n",
      "Epoch: 4 -> train_loss: 0.3702941476901372431, val_loss: 0.3610889613628387451,  val_acc: 87.08%\n",
      "Epoch: 5 -> train_loss: 0.3529314926067987912, val_loss: 0.3292031884193420410,  val_acc: 88.07%\n",
      "Epoch: 6 -> train_loss: 0.3384442613820234924, val_loss: 0.3356113731861114502,  val_acc: 87.88%\n",
      "Epoch: 7 -> train_loss: 0.3299360174735387341, val_loss: 0.3304152488708496094,  val_acc: 88.25%\n",
      "Epoch: 8 -> train_loss: 0.3183699831465879870, val_loss: 0.3215932250022888184,  val_acc: 88.32%\n",
      "Epoch: 9 -> train_loss: 0.3066120248536268744, val_loss: 0.3331522047519683838,  val_acc: 88.21%\n",
      "Epoch: 10 -> train_loss: 0.3030157501995563440, val_loss: 0.3165770173072814941,  val_acc: 88.86%\n",
      "Epoch: 11 -> train_loss: 0.2961013345122337492, val_loss: 0.3231025636196136475,  val_acc: 88.11%\n",
      "Epoch: 12 -> train_loss: 0.2876731425921122409, val_loss: 0.3123144209384918213,  val_acc: 88.69%\n",
      "Epoch: 13 -> train_loss: 0.2815732069909572810, val_loss: 0.3170706033706665039,  val_acc: 88.46%\n",
      "Epoch: 14 -> train_loss: 0.2739904375871022313, val_loss: 0.3109590411186218262,  val_acc: 88.96%\n",
      "Epoch: 15 -> train_loss: 0.2682814720571041200, val_loss: 0.3042424619197845459,  val_acc: 89.29%\n",
      "Epoch: 16 -> train_loss: 0.2654960344235102188, val_loss: 0.3168744444847106934,  val_acc: 89.11%\n",
      "Epoch: 17 -> train_loss: 0.2633762069841226183, val_loss: 0.3101049065589904785,  val_acc: 88.79%\n",
      "Epoch: 18 -> train_loss: 0.2575348454515139496, val_loss: 0.3040629029273986816,  val_acc: 89.22%\n",
      "Epoch: 19 -> train_loss: 0.2502811358968416999, val_loss: 0.3050162196159362793,  val_acc: 89.39%\n",
      "Epoch: 20 -> train_loss: 0.2457498750587304437, val_loss: 0.3026749193668365479,  val_acc: 89.45%\n",
      "Epoch: 21 -> train_loss: 0.2395246799389521286, val_loss: 0.3004771769046783447,  val_acc: 89.69%\n",
      "Epoch: 22 -> train_loss: 0.2380762482732534380, val_loss: 0.3205380141735076904,  val_acc: 89.53%\n",
      "Epoch: 23 -> train_loss: 0.2313262286484241426, val_loss: 0.3075141012668609619,  val_acc: 89.69%\n",
      "Epoch: 24 -> train_loss: 0.2325393498738606846, val_loss: 0.2975164949893951416,  val_acc: 89.84%\n",
      "Epoch: 25 -> train_loss: 0.2314718961914380424, val_loss: 0.2991222441196441650,  val_acc: 89.60%\n",
      "Epoch: 26 -> train_loss: 0.2241996667136748611, val_loss: 0.3062023818492889404,  val_acc: 89.69%\n",
      "Epoch: 27 -> train_loss: 0.2192408299843470187, val_loss: 0.3058206140995025635,  val_acc: 89.56%\n",
      "Epoch: 28 -> train_loss: 0.2181850755562385058, val_loss: 0.3072552680969238281,  val_acc: 89.46%\n",
      "Epoch: 29 -> train_loss: 0.2161911162187655777, val_loss: 0.3013016283512115479,  val_acc: 90.00%\n",
      "Epoch: 30 -> train_loss: 0.2105504951179027473, val_loss: 0.3058974742889404297,  val_acc: 89.90%\n",
      "Epoch: 31 -> train_loss: 0.2090706681162118885, val_loss: 0.2904951870441436768,  val_acc: 89.88%\n",
      "Epoch: 32 -> train_loss: 0.2068803765972455355, val_loss: 0.3066459596157073975,  val_acc: 89.62%\n",
      "Epoch: 33 -> train_loss: 0.2020631300210952719, val_loss: 0.3165600001811981201,  val_acc: 89.64%\n",
      "Epoch: 34 -> train_loss: 0.1992565853993097935, val_loss: 0.3035983443260192871,  val_acc: 89.69%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 248,
       "width": 372
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Train and validate again with new architecture\n",
    "train_validate(model, loss_fn, optimizer, train_dl, val_dl, device, n_epochs=35)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Validate on test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 88.65%\n"
     ]
    }
   ],
   "source": [
    "# Validate\n",
    "with torch.no_grad():\n",
    "    model.eval()\n",
    "    batch_acc = []\n",
    "    for images, labels in test_dl:\n",
    "        images, labels = images.to(device), labels.to(device)\n",
    "        # flatten images to batch_size x 784\n",
    "        images = images.view(images.shape[0], -1)\n",
    "        # make predictions and get probabilities\n",
    "        proba = torch.exp(model(images))\n",
    "        # extract the class associted with highest probability\n",
    "        _, pred_labels = proba.topk(1, dim=1)\n",
    "        # compare actual labels and predicted labels\n",
    "        result = pred_labels == labels.view(pred_labels.shape)\n",
    "        acc = torch.mean(result.type(torch.FloatTensor))\n",
    "        batch_acc.append(acc.item())\n",
    "    else:\n",
    "        print(f'Accuracy: {torch.mean(torch.tensor(batch_acc))*100:.2f}%')"
   ]
  }
 ],
 "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}