colab-notebooks/16_PyTorchCNN.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "16_PyTorchCNN.ipynb",
      "provenance": [],
      "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": [
        "<a href=\"https://colab.research.google.com/github/satyajitghana/PadhAI-Course/blob/master/16_PyTorchCNN.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5nwcQmEOJguu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import torch\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "\n",
        "import seaborn as sns\n",
        "sns.set()\n",
        "from tqdm import tqdm_notebook "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "v3RZSgGLO9iY",
        "colab_type": "text"
      },
      "source": [
        "## Data loading"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fOgFtnl2ZXsx",
        "colab_type": "code",
        "outputId": "d5ec910c-4c23-4f1d-87d6-14589d885ed9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "import torchvision\n",
        "import torchvision.transforms as transforms\n",
        "trainset = torchvision.datasets.CIFAR10(root='./data', train=True, \n",
        "                                        download=True, \n",
        "                                        transform=transforms.ToTensor())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\r0it [00:00, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data/cifar-10-python.tar.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            " 99%|█████████▉| 168697856/170498071 [00:12<00:00, 17372362.42it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting ./data/cifar-10-python.tar.gz to ./data\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tjT-1-Uba1n-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "classes = ('plane', 'car', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dpU3_h-JaFi_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "trainloader = torch.utils.data.DataLoader(trainset, batch_size=4, shuffle=True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TGkggskRaXhz",
        "colab_type": "code",
        "outputId": "bf16b42d-4446-4bc1-ca65-33006e7e98d4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 70
        }
      },
      "source": [
        "dataiter = iter(trainloader)\n",
        "images, labels = dataiter.next()\n",
        "\n",
        "print(images.shape)\n",
        "\n",
        "print(images[1].shape)\n",
        "print(labels[1].item())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "torch.Size([4, 3, 32, 32])\n",
            "torch.Size([3, 32, 32])\n",
            "3\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bQTsI03XPGId",
        "colab_type": "text"
      },
      "source": [
        "## Visualise data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RMLgzgf8PtN9",
        "colab_type": "code",
        "outputId": "1795161e-a9a3-4f6a-83a9-e6d6879e8762",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "img = images[1]\n",
        "print(type(img))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'torch.Tensor'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mXeB_gG9QIIW",
        "colab_type": "code",
        "outputId": "7895bbda-c256-46a6-e47c-8aa4e74c605a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "npimg = img.numpy()\n",
        "print(npimg.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(3, 32, 32)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "T7MfJdNAQKmy",
        "colab_type": "code",
        "outputId": "114592ab-479b-491c-e525-923cddfeb9d2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "npimg = np.transpose(npimg, (1, 2, 0))\n",
        "print(npimg.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(32, 32, 3)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Xjh0IJNIQS4G",
        "colab_type": "code",
        "outputId": "f08fadf9-07cd-4c61-c0f1-7860a54f4454",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        }
      },
      "source": [
        "# plt.figure(figsize = (1,1))\n",
        "plt.imshow(npimg)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD7CAYAAAClmULcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCwY4zXjaAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ItDj5P3xZrX3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def imshow(img):\n",
        "    npimg = img.numpy()\n",
        "    plt.imshow(np.transpose(npimg, (1, 2, 0)))\n",
        "    plt.show()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vX2OhQjVhFVT",
        "colab_type": "code",
        "outputId": "e909e2ec-0ec5-474d-d62e-8edcbc2150d8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 158
        }
      },
      "source": [
        "imshow(torchvision.utils.make_grid(images))\n",
        "print(' '.join(classes[labels[j]] for j in range(4)))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAB8CAYAAAB5R0uKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "plane cat frog frog\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KIP5NFfgQfNo",
        "colab_type": "text"
      },
      "source": [
        "## Single Convolutional Layer"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rehaPwlqaapX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import torch.nn as nn\n",
        "\n",
        "class FirstCNN(nn.Module):\n",
        "    def __init__(self): \n",
        "        super(FirstCNN, self).__init__()\n",
        "        self.conv1 = nn.Conv2d(3, 16, 3, padding=(1,1), stride=(2,2)) # 3=depth of input, 16=depth of output, padding=(1,1), stride=(2,2))\n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self.conv1(x)\n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gj33mJuCg8w-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "net = FirstCNN()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BnAWY2-qg9jR",
        "colab_type": "code",
        "outputId": "b67905d7-a239-4fbb-8e66-e60a7435ad29",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "out = net(images)\n",
        "out.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([4, 16, 16, 16])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ynP7midTjfIn",
        "colab_type": "code",
        "outputId": "d0ba8a96-c2b3-41c1-90c0-f73c1f33879d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "for param in net.parameters():\n",
        "    print(param.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "torch.Size([16, 3, 3, 3])\n",
            "torch.Size([16])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aBYgoxzEia-D",
        "colab_type": "code",
        "outputId": "0b012611-a53d-4177-9000-9068b2149da7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "out1 = out[0, 0, :, :].detach().numpy()\n",
        "print(out1.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(16, 16)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FG8Mqfw5i7hP",
        "colab_type": "code",
        "outputId": "ebe6621f-da8f-419d-ec85-7b6bdaebf7f9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        }
      },
      "source": [
        "plt.imshow(out[0, 0, :, :].detach().numpy())\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD7CAYAAAClmULcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niuEnUhTDT6Qohp9IUQw/kaIYfiJF/Q99vdq1liHl6QAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Y6nSsRqRQ3sJ",
        "colab_type": "text"
      },
      "source": [
        "## Deep Convolutional Network"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YtwIDsUUjH9h",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class FirstCNN_v2(nn.Module):\n",
        "    def __init__(self): \n",
        "        super(FirstCNN_v2, self).__init__()\n",
        "        self.model = nn.Sequential(\n",
        "            nn.Conv2d(3, 8, 3),   # (N, 3, 32, 32) -> (N, 8, 30, 30)\n",
        "            nn.Conv2d(8, 16, 3)   # (N, 8, 30, 30) -> (N, 16, 28, 28)\n",
        "        )\n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self.model(x)\n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1TlJtTMoliU3",
        "colab_type": "code",
        "outputId": "8f450d94-0429-4157-97a5-7d468e35c470",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "net = FirstCNN_v2()\n",
        "out = net(images)\n",
        "out.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([4, 16, 28, 28])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jpPuOr1TlilH",
        "colab_type": "code",
        "outputId": "670560e4-703c-4479-c7ff-0708002f0fea",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        }
      },
      "source": [
        "plt.imshow(out[0, 0, :, :].detach().numpy())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7ff1f1993ac8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD7CAYAAAClmULcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nsedpV2YAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5GnYimnTmFyi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class FirstCNN_v3(nn.Module):\n",
        "    def __init__(self): \n",
        "        super(FirstCNN_v3, self).__init__()\n",
        "        self.model = nn.Sequential(\n",
        "            nn.Conv2d(3, 6, 5),          # (N, 3, 32, 32) -> (N, 6, 28, 28)\n",
        "            nn.AvgPool2d(2, stride=2),   # (N, 6, 28, 28) -> (N, 6, 14, 14)\n",
        "            nn.Conv2d(6, 16, 5),         # (N, 6, 14, 14) -> (N, 16, 10, 10)\n",
        "            nn.AvgPool2d(2, stride=2)    # (N, 16, 10, 10) -> (N, 16, 5, 5)\n",
        "        )\n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self.model(x)\n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ldq1anrdmwox",
        "colab_type": "code",
        "outputId": "0acb35be-7204-40f3-9d6b-2ae49fdff8f3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "net = FirstCNN_v3()\n",
        "out = net(images)\n",
        "out.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([4, 16, 5, 5])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tOnUd7TyVunB",
        "colab_type": "code",
        "outputId": "1914ab24-7318-4d88-fe83-8846c045f187",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        }
      },
      "source": [
        "plt.imshow(out[0, 0, :, :].detach().numpy())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7ff1f163d0b8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPgAAAD7CAYAAABHRVmlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vymNQCrDQ7dM",
        "colab_type": "text"
      },
      "source": [
        "## LeNet"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "npS8lRGT15bx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class LeNet(nn.Module):\n",
        "    def __init__(self): \n",
        "        super(LeNet, self).__init__()\n",
        "        self.cnn_model = nn.Sequential(\n",
        "            nn.Conv2d(3, 6, 5),         # (N, 3, 32, 32) -> (N,  6, 28, 28)\n",
        "            nn.Tanh(),\n",
        "            nn.AvgPool2d(2, stride=2),  # (N, 6, 28, 28) -> (N,  6, 14, 14)\n",
        "            nn.Conv2d(6, 16, 5),        # (N, 6, 14, 14) -> (N, 16, 10, 10)  \n",
        "            nn.Tanh(),\n",
        "            nn.AvgPool2d(2, stride=2)   # (N,16, 10, 10) -> (N, 16, 5, 5)\n",
        "        )\n",
        "        self.fc_model = nn.Sequential(\n",
        "            nn.Linear(400,120),         # (N, 400) -> (N, 120)\n",
        "            nn.Tanh(),\n",
        "            nn.Linear(120,84),          # (N, 120) -> (N, 84)\n",
        "            nn.Tanh(),\n",
        "            nn.Linear(84,10)            # (N, 84)  -> (N, 10)\n",
        "        )\n",
        "        \n",
        "    def forward(self, x):\n",
        "        print(x.shape)\n",
        "        x = self.cnn_model(x)\n",
        "        print(x.shape)\n",
        "        x = x.view(x.size(0), -1)\n",
        "        print(x.shape)\n",
        "        x = self.fc_model(x)\n",
        "        print(x.shape)\n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hruUII0sRIO4",
        "colab_type": "code",
        "outputId": "40ae685a-2c82-4be2-95bc-c9b8ceb18d7a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 87
        }
      },
      "source": [
        "net = LeNet()\n",
        "out = net(images)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "torch.Size([4, 3, 32, 32])\n",
            "torch.Size([4, 16, 5, 5])\n",
            "torch.Size([4, 400])\n",
            "torch.Size([4, 10])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2E66qlEzRapT",
        "colab_type": "code",
        "outputId": "0ad3e77d-e682-4712-c86a-99a376ac2c89",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 175
        }
      },
      "source": [
        "print(out)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tensor([[-1.5041e-01,  6.0973e-02,  1.1077e-01,  1.6792e-01, -4.4197e-05,\n",
            "         -9.4120e-02, -4.9929e-03, -5.1180e-02, -5.0174e-02,  1.1286e-02],\n",
            "        [-1.4711e-01,  5.4986e-02,  1.1074e-01,  1.4922e-01,  7.2940e-03,\n",
            "         -9.6358e-02, -1.6733e-02, -4.4345e-02, -5.6173e-02,  3.0789e-02],\n",
            "        [-1.4184e-01,  6.4094e-02,  1.0411e-01,  1.5147e-01, -5.6122e-03,\n",
            "         -9.7452e-02, -1.3315e-02, -3.7054e-02, -5.2521e-02,  4.1820e-02],\n",
            "        [-1.4405e-01,  6.3414e-02,  1.0522e-01,  1.4196e-01,  5.6484e-04,\n",
            "         -8.6328e-02, -9.1592e-03, -2.8914e-02, -5.0737e-02,  6.8504e-02]],\n",
            "       grad_fn=<AddmmBackward>)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZiYvD0l7Rhma",
        "colab_type": "code",
        "outputId": "f8c94945-cc88-4b8e-9741-de91579d95d8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "max_values, pred_class = torch.max(out.data, 1)\n",
        "print(pred_class)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tensor([3, 3, 3, 3])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lOQGdybZRQqk",
        "colab_type": "text"
      },
      "source": [
        "## Training LeNet"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ylTZodUuV1eb",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class LeNet(nn.Module):\n",
        "    def __init__(self): \n",
        "        super(LeNet, self).__init__()\n",
        "        self.cnn_model = nn.Sequential(\n",
        "            nn.Conv2d(3, 6, 5),         # (N, 3, 32, 32) -> (N,  6, 28, 28)\n",
        "            nn.Tanh(),\n",
        "            nn.AvgPool2d(2, stride=2),  # (N, 6, 28, 28) -> (N,  6, 14, 14)\n",
        "            nn.Conv2d(6, 16, 5),        # (N, 6, 14, 14) -> (N, 16, 10, 10)  \n",
        "            nn.Tanh(),\n",
        "            nn.AvgPool2d(2, stride=2)   # (N,16, 10, 10) -> (N, 16, 5, 5)\n",
        "        )\n",
        "        self.fc_model = nn.Sequential(\n",
        "            nn.Linear(400,120),         # (N, 400) -> (N, 120)\n",
        "            nn.Tanh(),\n",
        "            nn.Linear(120,84),          # (N, 120) -> (N, 84)\n",
        "            nn.Tanh(),\n",
        "            nn.Linear(84,10)            # (N, 84)  -> (N, 10)\n",
        "        )\n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self.cnn_model(x)\n",
        "        x = x.view(x.size(0), -1)\n",
        "        x = self.fc_model(x)\n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GOWc4kya8UDC",
        "colab_type": "code",
        "outputId": "91e46117-3794-4e15-929d-ea8759c7bf6d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "batch_size = 128\n",
        "trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transforms.ToTensor())\n",
        "trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True)\n",
        "testset = torchvision.datasets.CIFAR10(root='./data', train=False, download=True, transform=transforms.ToTensor())\n",
        "testloader = torch.utils.data.DataLoader(testset, batch_size=batch_size, shuffle=False)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Files already downloaded and verified\n",
            "Files already downloaded and verified\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x9ehrnaHHZoj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def evaluation(dataloader):\n",
        "    total, correct = 0, 0\n",
        "    for data in dataloader:\n",
        "        inputs, labels = data\n",
        "        outputs = net(inputs)\n",
        "        _, pred = torch.max(outputs.data, 1)\n",
        "        total += labels.size(0)\n",
        "        correct += (pred == labels).sum().item()\n",
        "    return 100 * correct / total"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3pYTvSvD2TwM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "net = LeNet()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2T6Wzz7J2xq3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import torch.optim as optim\n",
        "\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "opt = optim.Adam(net.parameters())"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a0LnxsXi6Cgt",
        "colab_type": "code",
        "outputId": "66f12cb9-4a8e-4522-e2fa-0f82d90c6a5a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 585
        }
      },
      "source": [
        "%%time\n",
        "loss_arr = []\n",
        "loss_epoch_arr = []\n",
        "max_epochs = 16\n",
        "\n",
        "for epoch in range(max_epochs):\n",
        "\n",
        "    for i, data in enumerate(trainloader, 0):\n",
        "\n",
        "        inputs, labels = data\n",
        "\n",
        "        opt.zero_grad()\n",
        "\n",
        "        outputs = net(inputs)\n",
        "        loss = loss_fn(outputs, labels)\n",
        "        loss.backward()\n",
        "        opt.step()\n",
        "        \n",
        "        loss_arr.append(loss.item())\n",
        "        \n",
        "    loss_epoch_arr.append(loss.item())\n",
        "        \n",
        "    print('Epoch: %d/%d, Test acc: %0.2f, Train acc: %0.2f' % (epoch, max_epochs, evaluation(testloader), evaluation(trainloader)))\n",
        "    \n",
        "    \n",
        "plt.plot(loss_epoch_arr)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\r170500096it [00:30, 17372362.42it/s]                               "
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch: 0/16, Test acc: 38.06, Train acc: 38.45\n",
            "Epoch: 1/16, Test acc: 42.39, Train acc: 42.48\n",
            "Epoch: 2/16, Test acc: 46.11, Train acc: 47.08\n",
            "Epoch: 3/16, Test acc: 49.62, Train acc: 50.51\n",
            "Epoch: 4/16, Test acc: 51.65, Train acc: 52.92\n",
            "Epoch: 5/16, Test acc: 52.50, Train acc: 54.87\n",
            "Epoch: 6/16, Test acc: 53.30, Train acc: 56.29\n",
            "Epoch: 7/16, Test acc: 53.87, Train acc: 58.12\n",
            "Epoch: 8/16, Test acc: 54.12, Train acc: 58.99\n",
            "Epoch: 9/16, Test acc: 54.93, Train acc: 60.31\n",
            "Epoch: 10/16, Test acc: 54.51, Train acc: 61.30\n",
            "Epoch: 11/16, Test acc: 55.37, Train acc: 62.45\n",
            "Epoch: 12/16, Test acc: 55.55, Train acc: 62.73\n",
            "Epoch: 13/16, Test acc: 56.22, Train acc: 64.03\n",
            "Epoch: 14/16, Test acc: 55.96, Train acc: 64.79\n",
            "Epoch: 15/16, Test acc: 54.99, Train acc: 64.93\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAD7CAYAAABgzo9kAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 8min 48s, sys: 9.19 s, total: 8min 57s\n",
            "Wall time: 8min 59s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ECHeB_BUTHl5",
        "colab_type": "text"
      },
      "source": [
        "## Move to GPU"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NxJCf_FdX6Lb",
        "colab_type": "code",
        "outputId": "31a320bf-2dec-4466-b5a0-475ef091110f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "torch.cuda.device_count()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "1"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 36
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mj0-Y6zJTJ5z",
        "colab_type": "code",
        "outputId": "a0323039-d8a0-49f9-cc7e-56f23eaf29a2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
        "print(device)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "cuda:0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n4fHbz2fTNpR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def evaluation(dataloader):\n",
        "    total, correct = 0, 0\n",
        "    for data in dataloader:\n",
        "        inputs, labels = data\n",
        "        inputs, labels = inputs.to(device), labels.to(device)\n",
        "        outputs = net(inputs)\n",
        "        _, pred = torch.max(outputs.data, 1)\n",
        "        total += labels.size(0)\n",
        "        correct += (pred == labels).sum().item()\n",
        "    return 100 * correct / total"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mnxecsfVTQH-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "net = LeNet().to(device)\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "opt = optim.Adam(net.parameters())"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i9t3Js-XTWkf",
        "colab_type": "code",
        "outputId": "723bb9a5-21f3-4f4f-ec8f-ac0b81f4ab7d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 585
        }
      },
      "source": [
        "%%time\n",
        "max_epochs = 16\n",
        "loss_epoch_arr = []\n",
        "for epoch in range(max_epochs):\n",
        "\n",
        "    for i, data in enumerate(trainloader, 0):\n",
        "\n",
        "        inputs, labels = data\n",
        "        inputs, labels = inputs.to(device), labels.to(device)\n",
        "\n",
        "        opt.zero_grad()\n",
        "\n",
        "        outputs = net(inputs)\n",
        "        loss = loss_fn(outputs, labels)\n",
        "        loss.backward()\n",
        "        opt.step()\n",
        "        loss_epoch_arr.append(loss.item())\n",
        "        \n",
        "    print('Epoch: {}/{} Loss: {} Test acc: {} Train acc: {}'.format(epoch, max_epochs, loss, evaluation(testloader), evaluation(trainloader)))\n",
        "plt.plot(loss_epoch_arr)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch: 0/16 Loss: 1.7385966777801514 Test acc: 39.14 Train acc: 39.314\n",
            "Epoch: 1/16 Loss: 1.7396268844604492 Test acc: 44.1 Train acc: 44.04\n",
            "Epoch: 2/16 Loss: 1.6666606664657593 Test acc: 47.54 Train acc: 48.112\n",
            "Epoch: 3/16 Loss: 1.4193602800369263 Test acc: 48.56 Train acc: 50.4\n",
            "Epoch: 4/16 Loss: 1.3457355499267578 Test acc: 51.71 Train acc: 53.858\n",
            "Epoch: 5/16 Loss: 1.4077545404434204 Test acc: 52.15 Train acc: 54.714\n",
            "Epoch: 6/16 Loss: 1.3515287637710571 Test acc: 52.48 Train acc: 56.13\n",
            "Epoch: 7/16 Loss: 1.3254845142364502 Test acc: 53.51 Train acc: 57.882\n",
            "Epoch: 8/16 Loss: 1.0246491432189941 Test acc: 53.48 Train acc: 57.772\n",
            "Epoch: 9/16 Loss: 1.0698554515838623 Test acc: 54.77 Train acc: 60.406\n",
            "Epoch: 10/16 Loss: 1.0937621593475342 Test acc: 54.54 Train acc: 60.244\n",
            "Epoch: 11/16 Loss: 1.1237317323684692 Test acc: 55.02 Train acc: 61.252\n",
            "Epoch: 12/16 Loss: 1.119950771331787 Test acc: 54.51 Train acc: 62.454\n",
            "Epoch: 13/16 Loss: 0.9621752500534058 Test acc: 55.98 Train acc: 63.918\n",
            "Epoch: 14/16 Loss: 0.8914324641227722 Test acc: 56.06 Train acc: 64.878\n",
            "Epoch: 15/16 Loss: 0.9728390574455261 Test acc: 55.8 Train acc: 64.71\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAD7CAYAAABgzo9kAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nqP8H4Jsg3vJh678AAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 4min 37s, sys: 1.88 s, total: 4min 38s\n",
            "Wall time: 4min 39s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1mao_iiQZVlE",
        "colab_type": "code",
        "outputId": "eb1bd74b-669f-467a-d7e8-a5d1ce0370df",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "print('Test acc: %0.2f, Train acc: %0.2f' % (evaluation(testloader), evaluation(trainloader)))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Test acc: 55.80, Train acc: 64.71\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fq5T-KsjTEUQ",
        "colab_type": "text"
      },
      "source": [
        "## Basic Visualisation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LWfeXhrzSFSc",
        "colab_type": "code",
        "outputId": "4c163293-fdae-4746-c0c3-f96eff2656b6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 141
        }
      },
      "source": [
        "imshow(torchvision.utils.make_grid(images))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAB8CAYAAAB5R0uKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DiorUKc-LnWx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "net = net.to('cpu')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "luNsBRCAKQC_",
        "colab_type": "code",
        "outputId": "4c5ffac0-9187-435e-9f32-a8f17c76f199",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "out = net(images)\n",
        "print(out.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "torch.Size([4, 10])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-38nfaIUKSeF",
        "colab_type": "code",
        "outputId": "9c883f7b-259e-4f8e-9c4d-f0be81c38be5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "out = net.cnn_model[0](images)\n",
        "out.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([4, 6, 28, 28])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 45
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gqj4_qG1KYcO",
        "colab_type": "code",
        "outputId": "f5080cb2-f050-4f0e-b01c-2441c4f3ff78",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 161
        }
      },
      "source": [
        "image_id = 3\n",
        "plt.figure(figsize = (2,2))\n",
        "imshow(images[image_id,])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAJIAAACQCAYAAADjszKFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJTWO5twi2aoAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 144x144 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "19iXHHLitP2D",
        "colab_type": "code",
        "outputId": "03de0454-751c-44a2-ad45-7eae248bb02f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 434
        }
      },
      "source": [
        "plt.figure(figsize = (6,6))\n",
        "plt.subplot(321)\n",
        "for i in range(6):\n",
        "    ax1 = plt.subplot(3, 2, i+1)\n",
        "    plt.imshow(out[image_id, i, :, :].detach().numpy(), cmap=\"binary\")\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:4: MatplotlibDeprecationWarning: Adding an axes using the same arguments as a previous axes currently reuses the earlier instance.  In a future version, a new instance will always be created and returned.  Meanwhile, this warning can be suppressed, and the future behavior ensured, by passing a unique label to each axes instance.\n",
            "  after removing the cwd from sys.path.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAATwAAAFqCAYAAACH0ubsAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ng3V4hmEkBuvwDMNIDP8f5l/OEANvJi4AAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x432 with 6 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "outputId": "cb47159a-684a-4692-a9a2-df52f3f6d8c6",
        "id": "qN3d3CsTgM4S",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "out = net.cnn_model[1](images)\n",
        "out.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([4, 3, 32, 32])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 48
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "outputId": "68c9d89f-f1ad-42ea-a4b1-a50a5802dcff",
        "id": "xvw--jMwgM4c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 161
        }
      },
      "source": [
        "image_id = 3\n",
        "plt.figure(figsize = (2,2))\n",
        "imshow(images[image_id,])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAJIAAACQCAYAAADjszKFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJTWO5twi2aoAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 144x144 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "outputId": "4919516d-f359-4c4e-e395-0bc1a20a852e",
        "id": "sEBw19EGgM4e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 319
        }
      },
      "source": [
        "plt.figure(figsize = (6,6))\n",
        "plt.subplot(321)\n",
        "for i in range(3):\n",
        "    ax1 = plt.subplot(3, 2, i+1)\n",
        "    plt.imshow(out[image_id, i, :, :].detach().numpy(), cmap=\"binary\")\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:4: MatplotlibDeprecationWarning: Adding an axes using the same arguments as a previous axes currently reuses the earlier instance.  In a future version, a new instance will always be created and returned.  Meanwhile, this warning can be suppressed, and the future behavior ensured, by passing a unique label to each axes instance.\n",
            "  after removing the cwd from sys.path.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAATwAAAD3CAYAAACAXejFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x432 with 3 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nH4ceUXcgSeK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}