Image_Classifier-CIFAR10.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Image_Classifier.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "[View in Colaboratory](https://colab.research.google.com/github/gshashank84/learn-pytorch/blob/master/Image_Classifier.ipynb)"
      ]
    },
    {
      "metadata": {
        "id": "TveSZw38vXTD",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Image Classifier"
      ]
    },
    {
      "metadata": {
        "id": "lXZUv2EBvXTI",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### Dataset -> CIFAR10"
      ]
    },
    {
      "metadata": {
        "id": "NQzknABpvXTL",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- Import Dataset\n",
        "- Design model\n",
        "- Specify Hyperparameters\n",
        "- Forward Prop\n",
        "- Compute Loss\n",
        "- Backward Prop\n",
        "- Update Gradients"
      ]
    },
    {
      "metadata": {
        "id": "yuxuvIfcvzNJ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1201
        },
        "outputId": "2ff363e2-246d-4845-a490-823da6ee1629"
      },
      "cell_type": "code",
      "source": [
        "!pip install torch\n",
        "!pip install torchvision\n",
        "!pip install livelossplot"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting torch\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/49/0e/e382bcf1a6ae8225f50b99cc26effa2d4cc6d66975ccf3fa9590efcbedce/torch-0.4.1-cp36-cp36m-manylinux1_x86_64.whl (519.5MB)\n",
            "\u001b[K    100% |████████████████████████████████| 519.5MB 31kB/s \n",
            "tcmalloc: large alloc 1073750016 bytes == 0x5a49a000 @  0x7f03d64671c4 0x46d6a4 0x5fcbcc 0x4c494d 0x54f3c4 0x553aaf 0x54e4c8 0x54f4f6 0x553aaf 0x54efc1 0x54f24d 0x553aaf 0x54efc1 0x54f24d 0x553aaf 0x54efc1 0x54f24d 0x551ee0 0x54e4c8 0x54f4f6 0x553aaf 0x54efc1 0x54f24d 0x551ee0 0x54efc1 0x54f24d 0x551ee0 0x54e4c8 0x54f4f6 0x553aaf 0x54e4c8\n",
            "\u001b[?25hInstalling collected packages: torch\n",
            "Successfully installed torch-0.4.1\n",
            "Collecting torchvision\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/ca/0d/f00b2885711e08bd71242ebe7b96561e6f6d01fdb4b9dcf4d37e2e13c5e1/torchvision-0.2.1-py2.py3-none-any.whl (54kB)\n",
            "\u001b[K    100% |████████████████████████████████| 61kB 5.4MB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from torchvision) (1.14.6)\n",
            "Requirement already satisfied: torch in /usr/local/lib/python3.6/dist-packages (from torchvision) (0.4.1)\n",
            "Collecting pillow>=4.1.1 (from torchvision)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/62/94/5430ebaa83f91cc7a9f687ff5238e26164a779cca2ef9903232268b0a318/Pillow-5.3.0-cp36-cp36m-manylinux1_x86_64.whl (2.0MB)\n",
            "\u001b[K    100% |████████████████████████████████| 2.0MB 9.5MB/s \n",
            "\u001b[?25hRequirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from torchvision) (1.11.0)\n",
            "Installing collected packages: pillow, torchvision\n",
            "  Found existing installation: Pillow 4.0.0\n",
            "    Uninstalling Pillow-4.0.0:\n",
            "      Successfully uninstalled Pillow-4.0.0\n",
            "Successfully installed pillow-5.3.0 torchvision-0.2.1\n",
            "Collecting livelossplot\n",
            "  Downloading https://files.pythonhosted.org/packages/0c/01/3e34559dbdb98580d5e8f62bc3853975c2a2079834bfba2c1bdffe680804/livelossplot-0.2.0.tar.gz\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.6/dist-packages (from livelossplot) (2.1.2)\n",
            "Requirement already satisfied: notebook in /usr/local/lib/python3.6/dist-packages (from livelossplot) (5.2.2)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->livelossplot) (0.10.0)\n",
            "Requirement already satisfied: pytz in /usr/local/lib/python3.6/dist-packages (from matplotlib->livelossplot) (2018.5)\n",
            "Requirement already satisfied: six>=1.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->livelossplot) (1.11.0)\n",
            "Requirement already satisfied: numpy>=1.7.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->livelossplot) (1.14.6)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->livelossplot) (2.5.3)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->livelossplot) (2.2.1)\n",
            "Requirement already satisfied: tornado>=4 in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (4.5.3)\n",
            "Requirement already satisfied: terminado>=0.3.3; sys_platform != \"win32\" in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (0.8.1)\n",
            "Requirement already satisfied: jupyter-core in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (4.4.0)\n",
            "Requirement already satisfied: ipython-genutils in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (0.2.0)\n",
            "Requirement already satisfied: ipykernel in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (4.6.1)\n",
            "Requirement already satisfied: nbformat in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (4.4.0)\n",
            "Requirement already satisfied: traitlets>=4.2.1 in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (4.3.2)\n",
            "Requirement already satisfied: jinja2 in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (2.10)\n",
            "Requirement already satisfied: nbconvert in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (5.4.0)\n",
            "Requirement already satisfied: jupyter-client in /usr/local/lib/python3.6/dist-packages (from notebook->livelossplot) (5.2.3)\n",
            "Requirement already satisfied: ptyprocess; os_name != \"nt\" in /usr/local/lib/python3.6/dist-packages (from terminado>=0.3.3; sys_platform != \"win32\"->notebook->livelossplot) (0.6.0)\n",
            "Requirement already satisfied: ipython>=4.0.0 in /usr/local/lib/python3.6/dist-packages (from ipykernel->notebook->livelossplot) (5.5.0)\n",
            "Requirement already satisfied: jsonschema!=2.5.0,>=2.4 in /usr/local/lib/python3.6/dist-packages (from nbformat->notebook->livelossplot) (2.6.0)\n",
            "Requirement already satisfied: decorator in /usr/local/lib/python3.6/dist-packages (from traitlets>=4.2.1->notebook->livelossplot) (4.3.0)\n",
            "Requirement already satisfied: MarkupSafe>=0.23 in /usr/local/lib/python3.6/dist-packages (from jinja2->notebook->livelossplot) (1.0)\n",
            "Requirement already satisfied: pandocfilters>=1.4.1 in /usr/local/lib/python3.6/dist-packages (from nbconvert->notebook->livelossplot) (1.4.2)\n",
            "Requirement already satisfied: mistune>=0.8.1 in /usr/local/lib/python3.6/dist-packages (from nbconvert->notebook->livelossplot) (0.8.3)\n",
            "Requirement already satisfied: bleach in /usr/local/lib/python3.6/dist-packages (from nbconvert->notebook->livelossplot) (2.1.4)\n",
            "Requirement already satisfied: testpath in /usr/local/lib/python3.6/dist-packages (from nbconvert->notebook->livelossplot) (0.4.1)\n",
            "Requirement already satisfied: defusedxml in /usr/local/lib/python3.6/dist-packages (from nbconvert->notebook->livelossplot) (0.5.0)\n",
            "Requirement already satisfied: entrypoints>=0.2.2 in /usr/local/lib/python3.6/dist-packages (from nbconvert->notebook->livelossplot) (0.2.3)\n",
            "Requirement already satisfied: pygments in /usr/local/lib/python3.6/dist-packages (from nbconvert->notebook->livelossplot) (2.1.3)\n",
            "Requirement already satisfied: pyzmq>=13 in /usr/local/lib/python3.6/dist-packages (from jupyter-client->notebook->livelossplot) (16.0.4)\n",
            "Requirement already satisfied: pickleshare in /usr/local/lib/python3.6/dist-packages (from ipython>=4.0.0->ipykernel->notebook->livelossplot) (0.7.5)\n",
            "Requirement already satisfied: prompt-toolkit<2.0.0,>=1.0.4 in /usr/local/lib/python3.6/dist-packages (from ipython>=4.0.0->ipykernel->notebook->livelossplot) (1.0.15)\n",
            "Requirement already satisfied: pexpect; sys_platform != \"win32\" in /usr/local/lib/python3.6/dist-packages (from ipython>=4.0.0->ipykernel->notebook->livelossplot) (4.6.0)\n",
            "Requirement already satisfied: simplegeneric>0.8 in /usr/local/lib/python3.6/dist-packages (from ipython>=4.0.0->ipykernel->notebook->livelossplot) (0.8.1)\n",
            "Requirement already satisfied: setuptools>=18.5 in /usr/local/lib/python3.6/dist-packages (from ipython>=4.0.0->ipykernel->notebook->livelossplot) (39.1.0)\n",
            "Requirement already satisfied: html5lib!=1.0b1,!=1.0b2,!=1.0b3,!=1.0b4,!=1.0b5,!=1.0b6,!=1.0b7,!=1.0b8,>=0.99999999pre in /usr/local/lib/python3.6/dist-packages (from bleach->nbconvert->notebook->livelossplot) (1.0.1)\n",
            "Requirement already satisfied: wcwidth in /usr/local/lib/python3.6/dist-packages (from prompt-toolkit<2.0.0,>=1.0.4->ipython>=4.0.0->ipykernel->notebook->livelossplot) (0.1.7)\n",
            "Requirement already satisfied: webencodings in /usr/local/lib/python3.6/dist-packages (from html5lib!=1.0b1,!=1.0b2,!=1.0b3,!=1.0b4,!=1.0b5,!=1.0b6,!=1.0b7,!=1.0b8,>=0.99999999pre->bleach->nbconvert->notebook->livelossplot) (0.5.1)\n",
            "Building wheels for collected packages: livelossplot\n",
            "  Running setup.py bdist_wheel for livelossplot ... \u001b[?25l-\b \b\\\b \bdone\n",
            "\u001b[?25h  Stored in directory: /root/.cache/pip/wheels/67/8b/e9/3990164e3b2a421145b9d9d76fc8aa889225d495b043faeb73\n",
            "Successfully built livelossplot\n",
            "Installing collected packages: livelossplot\n",
            "Successfully installed livelossplot-0.2.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "8Bd7ux_jvXTP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "6652876d-f75a-45c9-c56e-3b368bc0d09c"
      },
      "cell_type": "code",
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import torchvision\n",
        "import torch.optim as optim\n",
        "import torchvision.transforms as transforms\n",
        "from livelossplot import PlotLosses\n",
        "import matplotlib.pyplot\n",
        "% matplotlib inline"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "vQeu_vb1vXVN",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "6b07914f-302b-4b6c-93b5-5b0c759a5f2d"
      },
      "cell_type": "code",
      "source": [
        "torch.cuda.is_available()"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "True"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "metadata": {
        "id": "2iOv2kYuvXTf",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Model"
      ]
    },
    {
      "metadata": {
        "id": "ly5vlgS3vXTl",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "class Net(nn.Module):\n",
        "    def __init__(self,input_channel,hidden_channel,out_channel, input_size, hidden_size, num_classes):\n",
        "        super(Net, self).__init__()\n",
        "        self.conv2d = nn.Conv2d(input_channel,hidden_channel,\n",
        "                                kernel_size=3,\n",
        "                                stride=1,padding=1)\n",
        "        self.conv2d_p = nn.Conv2d(hidden_channel, out_channel,\n",
        "                                 kernel_size= 5, stride=1)\n",
        "        self.fc1 = nn.Linear(input_size, hidden_size)\n",
        "        self.fc2 = nn.Linear(hidden_size, num_classes)\n",
        "        self.relu = nn.ReLU()\n",
        "        self.BatchNorm2d = nn.BatchNorm2d(hidden_channel)  # n_Channel\n",
        "        self.BatchNorm2d_p = nn.BatchNorm2d(out_channel)\n",
        "        self.sigmoid = nn.Sigmoid()\n",
        "        self.MaxPool2d = nn.MaxPool2d(kernel_size=3,stride=1)\n",
        "        self.Dropout2d = nn.Dropout2d(p=0.5)\n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self.conv2d(x)\n",
        "        x = self.BatchNorm2d(x)\n",
        "        x = self.MaxPool2d(x)\n",
        "        x = self.relu(x)\n",
        "        x = self.conv2d_p(x)\n",
        "        x = self.BatchNorm2d_p(x)\n",
        "        x = F.relu(F.max_pool2d(x,3))\n",
        "        x = x.reshape(x.size(0),-1) \n",
        "        x = self.fc1(x)\n",
        "        x = self.sigmoid(x)\n",
        "        x = self.Dropout2d(x)\n",
        "        x = self.fc2(x)\n",
        "        return x\n",
        "        "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "nlGS1NBtvXTu",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### Dataset load"
      ]
    },
    {
      "metadata": {
        "id": "Mndqeti9vXTz",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 72
        },
        "outputId": "42e14ea6-a91b-439a-c2c4-0a1b4df2ef94"
      },
      "cell_type": "code",
      "source": [
        "train_data = torchvision.datasets.CIFAR10(root = './data',\n",
        "                                              train = True,\n",
        "                                              transform = transforms.ToTensor(),\n",
        "                                              download = True)\n",
        "\n",
        "test_data = torchvision.datasets.CIFAR10(root = './data',\n",
        "                                             train = False,\n",
        "                                             transform = transforms.ToTensor(),\n",
        "                                             download = True)\n",
        "\n",
        "train_loader = torch.utils.data.DataLoader(dataset= train_data,\n",
        "                                          batch_size= 100,\n",
        "                                          shuffle = True)\n",
        "\n",
        "test_loader = torch.utils.data.DataLoader(dataset= test_data,\n",
        "                                         batch_size = 100,\n",
        "                                         shuffle = False)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data/cifar-10-python.tar.gz\n",
            "Files already downloaded and verified\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "5K8FNruFvXUA",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Implementation"
      ]
    },
    {
      "metadata": {
        "id": "aYdDL3DuvXUF",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "L9m5TLvIvXUM",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "input_channel = 3\n",
        "hidden_channel = 15\n",
        "out_channel = 25\n",
        "num_classes = 10\n",
        "input_size = out_channel*8*8\n",
        "hidden_size = 500\n",
        "epochs = 25\n",
        "total_step = len(train_loader)\n",
        "val_total_step = len(test_loader)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ZkAL3JDjvXUY",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "model = Net(input_channel,hidden_channel,out_channel, input_size, hidden_size, num_classes).to(device)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "1GOWxjsRvXUh",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# loss function\n",
        "criterion = nn.CrossEntropyLoss()\n",
        "\n",
        "# optimizer\n",
        "optimizer = optim.Adam(model.parameters(), lr=0.001)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "K5QoG6B1RVmM",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "liveloss = PlotLosses()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "oTVrB-6LvXUs",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 451
        },
        "outputId": "1120294e-9d69-47b0-d4f2-eb51fb3fe600"
      },
      "cell_type": "code",
      "source": [
        "for epoch in range(epochs):\n",
        "    avg_accuracy = 0\n",
        "    avg_loss = 0\n",
        "    total = 0\n",
        "    val_total = 0\n",
        "    val_avg_loss = 0\n",
        "    val_avg_accuracy = 0\n",
        "    for i, (images, labels) in enumerate(train_loader):\n",
        "        images = images.reshape(-1,input_channel,32,32).to(device)\n",
        "        labels = labels.to(device)\n",
        "        \n",
        "        #Forward Prop\n",
        "        output = model(images)\n",
        "        loss = criterion(output, labels)\n",
        "        \n",
        "        _,predicted = torch.max(output.data, 1)\n",
        "\n",
        "        total += labels.size(0)\n",
        "        avg_loss += loss.item()\n",
        "        avg_accuracy += (labels == predicted).sum().item()\n",
        "        \n",
        "        #Backward Prop\n",
        "        optimizer.zero_grad()\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "        \n",
        "        if (i+1)%(total_step) == 0:\n",
        "            print('Epoch: {}/{}, step: {}/{}, loss: {:.4f}'.format(epoch+1,\n",
        "                                                              epochs,\n",
        "                                                              i+1,\n",
        "                                                              total_step,\n",
        "                                                              loss.item()))\n",
        "        \n",
        "      \n",
        "        \n",
        "        \n",
        "  # The below code is for only validation loss and accuracy \n",
        "    \n",
        "    for val_images, val_labels in test_loader:\n",
        "      val_images = val_images.reshape(-1, input_channel,32,32).to(device)\n",
        "      val_labels = val_labels.to(device)\n",
        "      \n",
        "      val_output = model(val_images)\n",
        "      val_loss = criterion(val_output, val_labels)\n",
        "      _, val_predicted = torch.max(val_output.data, 1)\n",
        "      \n",
        "      val_total += val_labels.size(0)\n",
        "      val_avg_loss += val_loss.item()\n",
        "      val_avg_accuracy += (val_labels == val_predicted).sum().item()\n",
        "    \n",
        "    avg_loss = avg_loss/total_step\n",
        "    avg_accuracy = avg_accuracy/total\n",
        "    val_avg_loss = val_avg_loss/ val_total_step\n",
        "    val_avg_accuracy = val_avg_accuracy/ val_total\n",
        "    liveloss.update({'log loss': avg_loss,'val_log loss':val_avg_loss,\n",
        "                     'accuracy': avg_accuracy,'val_accuracy':val_avg_accuracy})"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch: 1/25, step: 500/500, loss: 1.4079\n",
            "Epoch: 2/25, step: 500/500, loss: 1.0167\n",
            "Epoch: 3/25, step: 500/500, loss: 1.1113\n",
            "Epoch: 4/25, step: 500/500, loss: 0.8793\n",
            "Epoch: 5/25, step: 500/500, loss: 0.7177\n",
            "Epoch: 6/25, step: 500/500, loss: 0.6775\n",
            "Epoch: 7/25, step: 500/500, loss: 0.7675\n",
            "Epoch: 8/25, step: 500/500, loss: 0.6890\n",
            "Epoch: 9/25, step: 500/500, loss: 0.5361\n",
            "Epoch: 10/25, step: 500/500, loss: 0.4384\n",
            "Epoch: 11/25, step: 500/500, loss: 0.7000\n",
            "Epoch: 12/25, step: 500/500, loss: 0.6408\n",
            "Epoch: 13/25, step: 500/500, loss: 0.5202\n",
            "Epoch: 14/25, step: 500/500, loss: 0.5485\n",
            "Epoch: 15/25, step: 500/500, loss: 0.4312\n",
            "Epoch: 16/25, step: 500/500, loss: 0.4233\n",
            "Epoch: 17/25, step: 500/500, loss: 0.4532\n",
            "Epoch: 18/25, step: 500/500, loss: 0.5158\n",
            "Epoch: 19/25, step: 500/500, loss: 0.4155\n",
            "Epoch: 20/25, step: 500/500, loss: 0.3234\n",
            "Epoch: 21/25, step: 500/500, loss: 0.3494\n",
            "Epoch: 22/25, step: 500/500, loss: 0.5302\n",
            "Epoch: 23/25, step: 500/500, loss: 0.3109\n",
            "Epoch: 24/25, step: 500/500, loss: 0.2486\n",
            "Epoch: 25/25, step: 500/500, loss: 0.3375\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "D48FmGJCvXVB",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "0ab605e6-3197-447a-9b08-e88b6cc21d19"
      },
      "cell_type": "code",
      "source": [
        "#Evaluating the model\n",
        "with torch.no_grad():    \n",
        "    correct = 0\n",
        "    total = 0\n",
        "    for images, labels in test_loader:\n",
        "        images = images.reshape(-1,input_channel,32,32).to(device)\n",
        "        labels = labels.to(device)\n",
        "\n",
        "        output = model(images)\n",
        "        _,predicted = torch.max(output.data, 1)\n",
        "\n",
        "        total += labels.size(0)\n",
        "        correct += (labels == predicted).sum().item()\n",
        "\n",
        "    print('Accuracy: {}%'.format(100*correct/total))"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy: 72.43%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "x-sGA6i7SEzE",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 346
        },
        "outputId": "7af6d5d7-5156-4567-bfa2-3b4ee169aaae"
      },
      "cell_type": "code",
      "source": [
        "liveloss.draw()"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA1gAAAE1CAYAAAD6akEFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f23d63589b0>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "l8caFbjux8ya",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "#72% <-- Highest Accuracy so far"
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}