colab-notebooks/17_LargeCNNs.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "17_LargeCNNs.ipynb",
      "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": [
        "<a href=\"https://colab.research.google.com/github/satyajitghana/PadhAI-Course/blob/master/17_LargeCNNs.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6KNf9CD2ytqH",
        "colab_type": "text"
      },
      "source": [
        "## Outline"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BDKI47gsW1Ii",
        "colab_type": "text"
      },
      "source": [
        "1. Loading datasets - Transforming images\n",
        "2. VGG-16 with modification to network head\n",
        "3. Using pre-trained models\n",
        "4. Storing intermediate models\n",
        "5. Resnet\n",
        "6. Inception v3\n",
        "7. Exercises"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rbZpqiiiylAg",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "\n",
        "import seaborn as sns\n",
        "sns.set()\n",
        "\n",
        "import torch\n",
        "import torchvision\n",
        "import torchvision.transforms as transforms\n",
        "import torch.nn as nn\n",
        "import torch.optim as optim"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1NBprN3Ry2NE",
        "colab_type": "code",
        "outputId": "c0840302-5c18-4386-a0c1-d91e0b4ce624",
        "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": "markdown",
      "metadata": {
        "id": "hGzpz3oE1Ikz",
        "colab_type": "text"
      },
      "source": [
        "## Dataset, transforms, and visualisation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y9c57fNA5Wsi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "transform_train = transforms.Compose([\n",
        "    transforms.RandomResizedCrop(224), \n",
        "    transforms.ToTensor(),\n",
        "    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)), # std dev. = 0.5; mean = 0.5\n",
        "    ])\n",
        "\n",
        "transform_test = transforms.Compose([\n",
        "    transforms.RandomResizedCrop(224), \n",
        "    transforms.ToTensor(),\n",
        "    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),\n",
        "    ])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fvsHorMPzISb",
        "colab_type": "code",
        "outputId": "1520bcb6-87bd-4c79-8d5f-446f9e20796d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 87
        }
      },
      "source": [
        "trainset = torchvision.datasets.CIFAR10(root='./data', train=True, \n",
        "                                        download=True, \n",
        "                                        transform=transform_train)\n",
        "testset = torchvision.datasets.CIFAR10(root='./data', train=False, \n",
        "                                        download=True, \n",
        "                                        transform=transform_test)"
      ],
      "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": [
            "170500096it [00:06, 27202199.62it/s]                               \n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting ./data/cifar-10-python.tar.gz to ./data\n",
            "Files already downloaded and verified\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V0dOpKtiAG4s",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "num_classes = 10"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xTNj3LQY4eTS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "batch_size = 4"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KfLwRIXH08tg",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KEOz-75x1NGu",
        "colab_type": "code",
        "outputId": "668ea69e-cf1e-4a6f-fe80-82e29471eb77",
        "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, 224, 224])\n",
            "torch.Size([3, 224, 224])\n",
            "0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5oBeIwYC1N3c",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def imshow(img, title):\n",
        "    npimg = img.numpy() / 2 + 0.5\n",
        "    plt.figure(figsize=(batch_size, 1))\n",
        "    plt.axis('off')\n",
        "    plt.imshow(np.transpose(npimg, (1, 2, 0)))\n",
        "    plt.title(title)\n",
        "    plt.show()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cfk-SYLY1Sbl",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def show_batch_images(dataloader):\n",
        "    images, labels = next(iter(dataloader))\n",
        "    img = torchvision.utils.make_grid(images)\n",
        "    imshow(img, title=[str(x.item()) for x in labels])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l_Zi_s3p1htN",
        "colab_type": "code",
        "outputId": "f94d89cf-b0cf-4ccc-ddfd-4237ef5c31c2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 353
        }
      },
      "source": [
        "for i in range(4):\n",
        "    show_batch_images(trainloader)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOYAAABUCAYAAACWaEvFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nfwFPexduHFz7sAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 288x72 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOYAAABUCAYAAACWaEvFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n4PocmJ/X5/U3uD4H5uf1ef0Nrv8L+OFGIC8q6SMAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 288x72 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOYAAABUCAYAAACWaEvFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n+Dr+Csf/BZUbBJ0fQ3eJAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 288x72 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOYAAABUCAYAAACWaEvFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 288x72 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1E-pnTG97DDz",
        "colab_type": "text"
      },
      "source": [
        "## Creating VGG-16"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_QMSp0gEJLow",
        "colab_type": "text"
      },
      "source": [
        "https://pytorch.org/docs/master/_modules/torchvision/models/vgg.html"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ra33PbJS28P3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from torchvision import models"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wvte5SSA7G7r",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "vgg = models.vgg16_bn()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7w-p73Tz9aZ_",
        "colab_type": "code",
        "outputId": "421dc032-ea06-49e7-9a4a-85153cf5127c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "print(vgg)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "VGG(\n",
            "  (features): Sequential(\n",
            "    (0): Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (2): ReLU(inplace=True)\n",
            "    (3): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (4): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (5): ReLU(inplace=True)\n",
            "    (6): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "    (7): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (8): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (9): ReLU(inplace=True)\n",
            "    (10): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (11): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (12): ReLU(inplace=True)\n",
            "    (13): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "    (14): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (15): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (16): ReLU(inplace=True)\n",
            "    (17): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (18): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (19): ReLU(inplace=True)\n",
            "    (20): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (21): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (22): ReLU(inplace=True)\n",
            "    (23): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "    (24): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (25): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (26): ReLU(inplace=True)\n",
            "    (27): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (28): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (29): ReLU(inplace=True)\n",
            "    (30): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (31): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (32): ReLU(inplace=True)\n",
            "    (33): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "    (34): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (35): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (36): ReLU(inplace=True)\n",
            "    (37): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (38): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (39): ReLU(inplace=True)\n",
            "    (40): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (41): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (42): ReLU(inplace=True)\n",
            "    (43): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "  )\n",
            "  (avgpool): AdaptiveAvgPool2d(output_size=(7, 7))\n",
            "  (classifier): Sequential(\n",
            "    (0): Linear(in_features=25088, out_features=4096, bias=True)\n",
            "    (1): ReLU(inplace=True)\n",
            "    (2): Dropout(p=0.5, inplace=False)\n",
            "    (3): Linear(in_features=4096, out_features=4096, bias=True)\n",
            "    (4): ReLU(inplace=True)\n",
            "    (5): Dropout(p=0.5, inplace=False)\n",
            "    (6): Linear(in_features=4096, out_features=1000, bias=True)\n",
            "  )\n",
            ")\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "N1bQdxQz-Dil",
        "colab_type": "code",
        "outputId": "84de56c5-4171-40c0-b555-7e06c9da3ea9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "print(vgg.features[0])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CX9PohSB-1Dx",
        "colab_type": "code",
        "outputId": "53c212d0-1e06-43c9-a9b5-4b0a315ed827",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "print(vgg.classifier[6])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Linear(in_features=4096, out_features=1000, bias=True)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "koDRbWi8_ApT",
        "colab_type": "code",
        "outputId": "0134ff33-67d9-4a47-d604-76a210cf01c1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "final_in_features = vgg.classifier[6].in_features\n",
        "mod_classifier = list(vgg.classifier.children())[:-1]\n",
        "mod_classifier.extend([nn.Linear(final_in_features, num_classes)])\n",
        "print(mod_classifier)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[Linear(in_features=25088, out_features=4096, bias=True), ReLU(inplace=True), Dropout(p=0.5, inplace=False), Linear(in_features=4096, out_features=4096, bias=True), ReLU(inplace=True), Dropout(p=0.5, inplace=False), Linear(in_features=4096, out_features=10, bias=True)]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_vHjXi1j_glv",
        "colab_type": "code",
        "outputId": "3c38d14c-cf27-44cb-dd90-58b57220b2d6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "vgg.classifier = nn.Sequential(*mod_classifier)\n",
        "print(vgg)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "VGG(\n",
            "  (features): Sequential(\n",
            "    (0): Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (2): ReLU(inplace=True)\n",
            "    (3): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (4): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (5): ReLU(inplace=True)\n",
            "    (6): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "    (7): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (8): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (9): ReLU(inplace=True)\n",
            "    (10): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (11): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (12): ReLU(inplace=True)\n",
            "    (13): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "    (14): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (15): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (16): ReLU(inplace=True)\n",
            "    (17): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (18): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (19): ReLU(inplace=True)\n",
            "    (20): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (21): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (22): ReLU(inplace=True)\n",
            "    (23): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "    (24): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (25): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (26): ReLU(inplace=True)\n",
            "    (27): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (28): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (29): ReLU(inplace=True)\n",
            "    (30): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (31): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (32): ReLU(inplace=True)\n",
            "    (33): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "    (34): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (35): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (36): ReLU(inplace=True)\n",
            "    (37): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (38): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (39): ReLU(inplace=True)\n",
            "    (40): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
            "    (41): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    (42): ReLU(inplace=True)\n",
            "    (43): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
            "  )\n",
            "  (avgpool): AdaptiveAvgPool2d(output_size=(7, 7))\n",
            "  (classifier): Sequential(\n",
            "    (0): Linear(in_features=25088, out_features=4096, bias=True)\n",
            "    (1): ReLU(inplace=True)\n",
            "    (2): Dropout(p=0.5, inplace=False)\n",
            "    (3): Linear(in_features=4096, out_features=4096, bias=True)\n",
            "    (4): ReLU(inplace=True)\n",
            "    (5): Dropout(p=0.5, inplace=False)\n",
            "    (6): Linear(in_features=4096, out_features=10, bias=True)\n",
            "  )\n",
            ")\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DVR1FiuJ-BzX",
        "colab_type": "text"
      },
      "source": [
        "### Train CIFAR10"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pKnlGE1q7JtN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "batch_size = 16\n",
        "\n",
        "trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True)\n",
        "testloader = torch.utils.data.DataLoader(testset, batch_size=batch_size, shuffle=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MtRHmDs_BvZG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def evaluation(dataloader, model):\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 = model(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": "htVdEliECDsz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "vgg = vgg.to(device)\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "opt = optim.SGD(vgg.parameters(), lr=0.05)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xSWnZJxjBbwW",
        "colab_type": "code",
        "outputId": "8952f06b-f097-41cb-f1d3-e17bc302f068",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 849
        }
      },
      "source": [
        "loss_epoch_arr = []\n",
        "max_epochs = 1\n",
        "\n",
        "n_iters = np.ceil(50000/batch_size)\n",
        "\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 = vgg(inputs)\n",
        "        loss = loss_fn(outputs, labels)\n",
        "        loss.backward()\n",
        "        opt.step()\n",
        "        \n",
        "        del inputs, labels, outputs\n",
        "        torch.cuda.empty_cache()\n",
        "        \n",
        "        if i % 100 == 0:\n",
        "            print('Iteration: %d/%d, Loss: %0.2f' % (i, n_iters, loss.item()))\n",
        "        \n",
        "         loss_epoch_arr.append(loss.item())\n",
        "        \n",
        "    print('Epoch: %d/%d, Test acc: %0.2f, Train acc: %0.2f' % (\n",
        "        epoch, max_epochs, \n",
        "        evaluation(testloader, vgg), evaluation(trainloader, vgg)))\n",
        "    \n",
        "plt.plot(loss_epoch_arr)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Iteration: 0/3125, Loss: 2.50\n",
            "Iteration: 100/3125, Loss: 2.47\n",
            "Iteration: 200/3125, Loss: 2.22\n",
            "Iteration: 300/3125, Loss: 2.36\n",
            "Iteration: 400/3125, Loss: 2.22\n",
            "Iteration: 500/3125, Loss: 2.45\n",
            "Iteration: 600/3125, Loss: 2.11\n",
            "Iteration: 700/3125, Loss: 2.28\n",
            "Iteration: 800/3125, Loss: 2.24\n",
            "Iteration: 900/3125, Loss: 2.55\n",
            "Iteration: 1000/3125, Loss: 2.16\n",
            "Iteration: 1100/3125, Loss: 2.26\n",
            "Iteration: 1200/3125, Loss: 2.20\n",
            "Iteration: 1300/3125, Loss: 2.15\n",
            "Iteration: 1400/3125, Loss: 2.08\n",
            "Iteration: 1500/3125, Loss: 2.45\n",
            "Iteration: 1600/3125, Loss: 2.19\n",
            "Iteration: 1700/3125, Loss: 2.02\n",
            "Iteration: 1800/3125, Loss: 2.33\n",
            "Iteration: 1900/3125, Loss: 2.18\n",
            "Iteration: 2000/3125, Loss: 2.28\n",
            "Iteration: 2100/3125, Loss: 2.28\n",
            "Iteration: 2200/3125, Loss: 1.78\n",
            "Iteration: 2300/3125, Loss: 2.04\n",
            "Iteration: 2400/3125, Loss: 2.37\n",
            "Iteration: 2500/3125, Loss: 2.56\n",
            "Iteration: 2600/3125, Loss: 2.01\n",
            "Iteration: 2700/3125, Loss: 2.16\n",
            "Iteration: 2800/3125, Loss: 1.97\n",
            "Iteration: 2900/3125, Loss: 2.01\n",
            "Iteration: 3000/3125, Loss: 1.93\n",
            "Iteration: 3100/3125, Loss: 1.94\n",
            "Epoch: 0/1, Test acc: 23.02, Train acc: 23.22\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAD7CAYAAABuSzNOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Bh4gREC-IGFB",
        "colab_type": "text"
      },
      "source": [
        "### Freeze layers of Convolutional Operations"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3aD1lt1qcJOU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "batch_size = 16\n",
        "\n",
        "trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True)\n",
        "testloader = torch.utils.data.DataLoader(testset, batch_size=batch_size, shuffle=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FzniE5sKKlnY",
        "colab_type": "code",
        "outputId": "3cdbef93-c1ba-4f09-cb4c-ba4b6ac62cea",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "vgg = models.vgg16_bn(pretrained=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading: \"https://download.pytorch.org/models/vgg16_bn-6c64b313.pth\" to /root/.cache/torch/checkpoints/vgg16_bn-6c64b313.pth\n",
            "100%|██████████| 528M/528M [00:19<00:00, 28.7MB/s]\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MzaHEHaczpWS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for param in vgg.parameters():\n",
        "    param.requires_grad = False"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YFAhAvahzrRh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "final_in_features = vgg.classifier[6].in_features\n",
        "vgg.classifier[6] = nn.Linear(final_in_features, num_classes)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uqXCJEiK0Z2V",
        "colab_type": "code",
        "outputId": "d75af2e2-ff6f-4304-ada4-3db7cfe55628",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "for param in vgg.parameters():\n",
        "    if param.requires_grad:\n",
        "        print(param.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "torch.Size([10, 4096])\n",
            "torch.Size([10])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lH-99npmPRUk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "vgg = vgg.to(device)\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "opt = optim.SGD(vgg.parameters(), lr=0.05)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5HuWwxaxIMTV",
        "colab_type": "code",
        "outputId": "47b30398-5d44-4e14-a0a3-470069e3b5a8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 849
        }
      },
      "source": [
        "%%time\n",
        "loss_epoch_arr = []\n",
        "max_epochs = 1\n",
        "\n",
        "n_iters = np.ceil(50000/batch_size)\n",
        "\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 = vgg(inputs)\n",
        "        loss = loss_fn(outputs, labels)\n",
        "        loss.backward()\n",
        "        opt.step()\n",
        "        \n",
        "        if i % 100 == 0:\n",
        "            print('Iteration: %d/%d, Loss: %0.2f' % (i, n_iters, loss.item()))\n",
        "            \n",
        "        del inputs, labels, outputs\n",
        "        torch.cuda.empty_cache()\n",
        "        \n",
        "        loss_epoch_arr.append(loss.item())\n",
        "        \n",
        "    print('Epoch: %d/%d, Test acc: %0.2f, Train acc: %0.2f' % (\n",
        "        epoch, max_epochs, \n",
        "        evaluation(testloader, vgg), evaluation(trainloader, vgg)))\n",
        "    \n",
        "    \n",
        "plt.plot(loss_epoch_arr)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Iteration: 0/3125, Loss: 2.30\n",
            "Iteration: 100/3125, Loss: 1.51\n",
            "Iteration: 200/3125, Loss: 2.13\n",
            "Iteration: 300/3125, Loss: 1.48\n",
            "Iteration: 400/3125, Loss: 2.30\n",
            "Iteration: 500/3125, Loss: 2.07\n",
            "Iteration: 600/3125, Loss: 2.18\n",
            "Iteration: 700/3125, Loss: 1.81\n",
            "Iteration: 800/3125, Loss: 1.58\n",
            "Iteration: 900/3125, Loss: 1.60\n",
            "Iteration: 1000/3125, Loss: 1.92\n",
            "Iteration: 1100/3125, Loss: 1.90\n",
            "Iteration: 1200/3125, Loss: 1.53\n",
            "Iteration: 1300/3125, Loss: 2.82\n",
            "Iteration: 1400/3125, Loss: 2.70\n",
            "Iteration: 1500/3125, Loss: 2.16\n",
            "Iteration: 1600/3125, Loss: 1.71\n",
            "Iteration: 1700/3125, Loss: 3.21\n",
            "Iteration: 1800/3125, Loss: 1.35\n",
            "Iteration: 1900/3125, Loss: 1.33\n",
            "Iteration: 2000/3125, Loss: 3.75\n",
            "Iteration: 2100/3125, Loss: 2.09\n",
            "Iteration: 2200/3125, Loss: 3.01\n",
            "Iteration: 2300/3125, Loss: 2.91\n",
            "Iteration: 2400/3125, Loss: 3.17\n",
            "Iteration: 2500/3125, Loss: 1.73\n",
            "Iteration: 2600/3125, Loss: 3.01\n",
            "Iteration: 2700/3125, Loss: 1.28\n",
            "Iteration: 2800/3125, Loss: 1.40\n",
            "Iteration: 2900/3125, Loss: 2.06\n",
            "Iteration: 3000/3125, Loss: 2.71\n",
            "Iteration: 3100/3125, Loss: 1.61\n",
            "Epoch: 0/1, Test acc: 45.11, Train acc: 46.12\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAD7CAYAAABt0P8jAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAcAYwg8AxvwfGkZWIv+6EeIAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AeacSwlg5p2r",
        "colab_type": "text"
      },
      "source": [
        "### With model copies"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4-dZMeUTpAxC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import copy"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Wf0qM_-PlXl-",
        "colab_type": "code",
        "outputId": "84710505-b2f0-4e21-ff15-9d1eb9d686c6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 361
        }
      },
      "source": [
        "%%time\n",
        "\n",
        "loss_epoch_arr = []\n",
        "max_epochs = 1\n",
        "\n",
        "min_loss = 1000\n",
        "\n",
        "n_iters = np.ceil(50000/batch_size)\n",
        "\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 = vgg(inputs)\n",
        "        loss = loss_fn(outputs, labels)\n",
        "        loss.backward()\n",
        "        opt.step()\n",
        "        \n",
        "        if min_loss > loss.item():\n",
        "            min_loss = loss.item()\n",
        "            best_model = copy.deepcopy(vgg.state_dict())\n",
        "            print('Min loss %0.2f' % min_loss)\n",
        "        \n",
        "        if i % 100 == 0:\n",
        "            print('Iteration: %d/%d, Loss: %0.2f' % (i, n_iters, loss.item()))\n",
        "            \n",
        "        del inputs, labels, outputs\n",
        "        torch.cuda.empty_cache()\n",
        "        \n",
        "        loss_epoch_arr.append(loss.item())\n",
        "        \n",
        "plt.plot(loss_epoch_arr)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Min loss 2.00\n",
            "Iteration: 0/3125, Loss: 2.00\n",
            "Min loss 1.80\n",
            "Min loss 0.96\n",
            "Min loss 0.66\n",
            "Iteration: 100/3125, Loss: 1.85\n",
            "Iteration: 200/3125, Loss: 2.39\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-31-549acea3c91d>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     20\u001b[0m         \u001b[0mopt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 22\u001b[0;31m         \u001b[0;32mif\u001b[0m \u001b[0mmin_loss\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0mloss\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     23\u001b[0m             \u001b[0mmin_loss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mloss\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     24\u001b[0m             \u001b[0mbest_model\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcopy\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdeepcopy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mvgg\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstate_dict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_2GAOmBytwZ9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "vgg.load_state_dict(best_model)\n",
        "print(evaluation(trainloader, vgg), evaluation(testloader, vgg))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "83zNw_MLFBhd",
        "colab_type": "text"
      },
      "source": [
        "## ResNet Model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RJHZ2ibFGYfG",
        "colab_type": "text"
      },
      "source": [
        "https://pytorch.org/docs/master/_modules/torchvision/models/resnet.html"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gYZJSHEKFDfS",
        "colab_type": "code",
        "outputId": "71296df0-0ba4-4ab3-d4cb-963f7ee50c98",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "resnet = models.resnet18(pretrained=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading: \"https://download.pytorch.org/models/resnet18-5c106cde.pth\" to /root/.cache/torch/checkpoints/resnet18-5c106cde.pth\n",
            "100%|██████████| 44.7M/44.7M [00:01<00:00, 28.3MB/s]\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KMUoM7ToFFeK",
        "colab_type": "code",
        "outputId": "9c3bbd2f-136a-44be-aa24-f41cdd1679bc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "print(resnet)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "ResNet(\n",
            "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
            "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "  (relu): ReLU(inplace=True)\n",
            "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
            "  (layer1): Sequential(\n",
            "    (0): BasicBlock(\n",
            "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (relu): ReLU(inplace=True)\n",
            "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (1): BasicBlock(\n",
            "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (relu): ReLU(inplace=True)\n",
            "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (layer2): Sequential(\n",
            "    (0): BasicBlock(\n",
            "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
            "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (relu): ReLU(inplace=True)\n",
            "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (downsample): Sequential(\n",
            "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
            "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      )\n",
            "    )\n",
            "    (1): BasicBlock(\n",
            "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (relu): ReLU(inplace=True)\n",
            "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (layer3): Sequential(\n",
            "    (0): BasicBlock(\n",
            "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
            "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (relu): ReLU(inplace=True)\n",
            "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (downsample): Sequential(\n",
            "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
            "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      )\n",
            "    )\n",
            "    (1): BasicBlock(\n",
            "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (relu): ReLU(inplace=True)\n",
            "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (layer4): Sequential(\n",
            "    (0): BasicBlock(\n",
            "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
            "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (relu): ReLU(inplace=True)\n",
            "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (downsample): Sequential(\n",
            "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
            "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      )\n",
            "    )\n",
            "    (1): BasicBlock(\n",
            "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "      (relu): ReLU(inplace=True)\n",
            "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
            "  (fc): Linear(in_features=512, out_features=1000, bias=True)\n",
            ")\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "t1KOHDdtFoK5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for param in resnet.parameters():\n",
        "    param.requires_grad = False"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LrL2ZlhrGwFL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "in_features = resnet.fc.in_features\n",
        "resnet.fc = nn.Linear(in_features, num_classes)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "53REdyaaG8ap",
        "colab_type": "code",
        "outputId": "98ca5918-b972-4a98-d3cd-8a36f094c51c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "for param in resnet.parameters():\n",
        "    if param.requires_grad:\n",
        "        print(param.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "torch.Size([10, 512])\n",
            "torch.Size([10])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vQTZ3X_pG_tT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "resnet = resnet.to(device)\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "opt = optim.SGD(resnet.parameters(), lr=0.01)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rklYsln-Hcpj",
        "colab_type": "code",
        "outputId": "1630c4af-63d4-47ea-b469-582f20ea6583",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "%%time\n",
        "\n",
        "loss_epoch_arr = []\n",
        "max_epochs = 4\n",
        "\n",
        "min_loss = 1000\n",
        "\n",
        "n_iters = np.ceil(50000/batch_size)\n",
        "\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 = resnet(inputs)\n",
        "        loss = loss_fn(outputs, labels)\n",
        "        loss.backward()\n",
        "        opt.step()\n",
        "        \n",
        "        if min_loss > loss.item():\n",
        "            min_loss = loss.item()\n",
        "            best_model = copy.deepcopy(resnet.state_dict())\n",
        "            print('Min loss %0.2f' % min_loss)\n",
        "        \n",
        "        if i % 100 == 0:\n",
        "            print('Iteration: %d/%d, Loss: %0.2f' % (i, n_iters, loss.item()))\n",
        "            \n",
        "        del inputs, labels, outputs\n",
        "        torch.cuda.empty_cache()\n",
        "        \n",
        "        loss_epoch_arr.append(loss.item())\n",
        "        \n",
        "    print('Epoch: %d/%d, Test acc: %0.2f, Train acc: %0.2f' % (\n",
        "        epoch, max_epochs, \n",
        "        evaluation(testloader, resnet), evaluation(trainloader, resnet)))\n",
        "    \n",
        "    \n",
        "plt.plot(loss_epoch_arr)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Min loss 2.38\n",
            "Iteration: 0/3125, Loss: 2.38\n",
            "Min loss 2.28\n",
            "Min loss 2.25\n",
            "Min loss 2.23\n",
            "Min loss 2.09\n",
            "Min loss 2.05\n",
            "Min loss 1.93\n",
            "Min loss 1.81\n",
            "Min loss 1.80\n",
            "Min loss 1.77\n",
            "Min loss 1.68\n",
            "Min loss 1.59\n",
            "Min loss 1.58\n",
            "Iteration: 100/3125, Loss: 1.58\n",
            "Min loss 1.32\n",
            "Min loss 1.31\n",
            "Min loss 1.30\n",
            "Iteration: 200/3125, Loss: 1.67\n",
            "Min loss 1.28\n",
            "Min loss 1.24\n",
            "Min loss 1.06\n",
            "Iteration: 300/3125, Loss: 1.73\n",
            "Min loss 1.03\n",
            "Min loss 1.02\n",
            "Iteration: 400/3125, Loss: 1.42\n",
            "Min loss 0.95\n",
            "Min loss 0.93\n",
            "Iteration: 500/3125, Loss: 1.48\n",
            "Iteration: 600/3125, Loss: 1.31\n",
            "Min loss 0.90\n",
            "Min loss 0.86\n",
            "Min loss 0.70\n",
            "Iteration: 700/3125, Loss: 1.06\n",
            "Iteration: 800/3125, Loss: 1.67\n",
            "Iteration: 900/3125, Loss: 0.93\n",
            "Iteration: 1000/3125, Loss: 1.29\n",
            "Iteration: 1100/3125, Loss: 1.82\n",
            "Min loss 0.68\n",
            "Iteration: 1200/3125, Loss: 1.16\n",
            "Min loss 0.67\n",
            "Iteration: 1300/3125, Loss: 1.04\n",
            "Iteration: 1400/3125, Loss: 1.41\n",
            "Iteration: 1500/3125, Loss: 1.16\n",
            "Min loss 0.59\n",
            "Iteration: 1600/3125, Loss: 1.66\n",
            "Iteration: 1700/3125, Loss: 2.02\n",
            "Iteration: 1800/3125, Loss: 1.58\n",
            "Iteration: 1900/3125, Loss: 1.05\n",
            "Min loss 0.56\n",
            "Iteration: 2000/3125, Loss: 2.20\n",
            "Iteration: 2100/3125, Loss: 1.06\n",
            "Iteration: 2200/3125, Loss: 1.50\n",
            "Iteration: 2300/3125, Loss: 1.46\n",
            "Iteration: 2400/3125, Loss: 1.20\n",
            "Iteration: 2500/3125, Loss: 1.77\n",
            "Iteration: 2600/3125, Loss: 1.56\n",
            "Min loss 0.52\n",
            "Iteration: 2700/3125, Loss: 1.09\n",
            "Iteration: 2800/3125, Loss: 1.31\n",
            "Iteration: 2900/3125, Loss: 1.66\n",
            "Iteration: 3000/3125, Loss: 1.51\n",
            "Iteration: 3100/3125, Loss: 1.32\n",
            "Epoch: 0/4, Test acc: 55.45, Train acc: 55.53\n",
            "Iteration: 0/3125, Loss: 1.18\n",
            "Iteration: 100/3125, Loss: 1.48\n",
            "Iteration: 200/3125, Loss: 1.34\n",
            "Min loss 0.48\n",
            "Iteration: 300/3125, Loss: 1.12\n",
            "Iteration: 400/3125, Loss: 1.19\n",
            "Iteration: 500/3125, Loss: 0.80\n",
            "Iteration: 600/3125, Loss: 0.85\n",
            "Iteration: 700/3125, Loss: 1.28\n",
            "Iteration: 800/3125, Loss: 0.88\n",
            "Iteration: 900/3125, Loss: 1.65\n",
            "Iteration: 1000/3125, Loss: 1.08\n",
            "Iteration: 1100/3125, Loss: 0.97\n",
            "Min loss 0.47\n",
            "Iteration: 1200/3125, Loss: 0.83\n",
            "Iteration: 1300/3125, Loss: 1.38\n",
            "Iteration: 1400/3125, Loss: 1.29\n",
            "Iteration: 1500/3125, Loss: 1.31\n",
            "Iteration: 1600/3125, Loss: 1.52\n",
            "Iteration: 1700/3125, Loss: 1.31\n",
            "Iteration: 1800/3125, Loss: 1.56\n",
            "Iteration: 1900/3125, Loss: 0.90\n",
            "Iteration: 2000/3125, Loss: 1.52\n",
            "Iteration: 2100/3125, Loss: 0.92\n",
            "Min loss 0.45\n",
            "Iteration: 2200/3125, Loss: 1.19\n",
            "Iteration: 2300/3125, Loss: 1.59\n",
            "Iteration: 2400/3125, Loss: 1.25\n",
            "Iteration: 2500/3125, Loss: 1.52\n",
            "Iteration: 2600/3125, Loss: 1.11\n",
            "Iteration: 2700/3125, Loss: 1.29\n",
            "Iteration: 2800/3125, Loss: 0.94\n",
            "Iteration: 2900/3125, Loss: 1.18\n",
            "Iteration: 3000/3125, Loss: 1.11\n",
            "Iteration: 3100/3125, Loss: 1.24\n",
            "Epoch: 1/4, Test acc: 56.01, Train acc: 56.70\n",
            "Iteration: 0/3125, Loss: 1.25\n",
            "Iteration: 100/3125, Loss: 1.02\n",
            "Iteration: 200/3125, Loss: 1.53\n",
            "Iteration: 300/3125, Loss: 0.87\n",
            "Iteration: 400/3125, Loss: 1.08\n",
            "Iteration: 500/3125, Loss: 1.20\n",
            "Iteration: 600/3125, Loss: 1.43\n",
            "Iteration: 700/3125, Loss: 0.98\n",
            "Iteration: 800/3125, Loss: 1.47\n",
            "Iteration: 900/3125, Loss: 1.45\n",
            "Iteration: 1000/3125, Loss: 1.23\n",
            "Iteration: 1100/3125, Loss: 1.06\n",
            "Iteration: 1200/3125, Loss: 1.77\n",
            "Iteration: 1300/3125, Loss: 0.79\n",
            "Iteration: 1400/3125, Loss: 1.39\n",
            "Iteration: 1500/3125, Loss: 1.07\n",
            "Iteration: 1600/3125, Loss: 1.42\n",
            "Iteration: 1700/3125, Loss: 1.69\n",
            "Iteration: 1800/3125, Loss: 1.23\n",
            "Iteration: 1900/3125, Loss: 0.98\n",
            "Iteration: 2000/3125, Loss: 1.65\n",
            "Iteration: 2100/3125, Loss: 1.84\n",
            "Iteration: 2200/3125, Loss: 1.05\n",
            "Iteration: 2300/3125, Loss: 1.21\n",
            "Iteration: 2400/3125, Loss: 0.98\n",
            "Iteration: 2500/3125, Loss: 1.09\n",
            "Iteration: 2600/3125, Loss: 1.48\n",
            "Iteration: 2700/3125, Loss: 0.88\n",
            "Iteration: 2800/3125, Loss: 1.07\n",
            "Iteration: 2900/3125, Loss: 1.11\n",
            "Min loss 0.41\n",
            "Iteration: 3000/3125, Loss: 1.52\n",
            "Iteration: 3100/3125, Loss: 1.21\n",
            "Epoch: 2/4, Test acc: 57.05, Train acc: 57.47\n",
            "Iteration: 0/3125, Loss: 1.25\n",
            "Iteration: 100/3125, Loss: 1.22\n",
            "Iteration: 200/3125, Loss: 1.48\n",
            "Iteration: 300/3125, Loss: 1.18\n",
            "Iteration: 400/3125, Loss: 0.99\n",
            "Iteration: 500/3125, Loss: 0.94\n",
            "Iteration: 600/3125, Loss: 1.21\n",
            "Iteration: 700/3125, Loss: 2.00\n",
            "Iteration: 800/3125, Loss: 1.30\n",
            "Iteration: 900/3125, Loss: 1.34\n",
            "Iteration: 1000/3125, Loss: 0.89\n",
            "Iteration: 1100/3125, Loss: 1.36\n",
            "Iteration: 1200/3125, Loss: 1.37\n",
            "Iteration: 1300/3125, Loss: 1.49\n",
            "Iteration: 1400/3125, Loss: 0.85\n",
            "Iteration: 1500/3125, Loss: 1.83\n",
            "Iteration: 1600/3125, Loss: 1.22\n",
            "Iteration: 1700/3125, Loss: 1.62\n",
            "Iteration: 1800/3125, Loss: 1.10\n",
            "Iteration: 1900/3125, Loss: 1.06\n",
            "Iteration: 2000/3125, Loss: 0.75\n",
            "Iteration: 2100/3125, Loss: 1.14\n",
            "Iteration: 2200/3125, Loss: 1.27\n",
            "Iteration: 2300/3125, Loss: 1.29\n",
            "Iteration: 2400/3125, Loss: 0.94\n",
            "Iteration: 2500/3125, Loss: 1.99\n",
            "Iteration: 2600/3125, Loss: 0.98\n",
            "Iteration: 2700/3125, Loss: 1.53\n",
            "Iteration: 2800/3125, Loss: 1.15\n",
            "Iteration: 2900/3125, Loss: 1.89\n",
            "Iteration: 3000/3125, Loss: 1.13\n",
            "Iteration: 3100/3125, Loss: 1.10\n",
            "Epoch: 3/4, Test acc: 56.08, Train acc: 55.81\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAD7CAYAAABgzo9kAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 13min 30s, sys: 1min 29s, total: 15min\n",
            "Wall time: 15min 1s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SbetTwcrIMnQ",
        "colab_type": "code",
        "outputId": "dcc75099-9acf-4167-a38b-a52e5c3d5c86",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "resnet.load_state_dict(best_model)\n",
        "print(evaluation(trainloader, resnet), evaluation(testloader, resnet))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "57.272 57.25\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "asFT3q7vdDbd",
        "colab_type": "text"
      },
      "source": [
        "## Inception Model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k-gBu2zCGWE8",
        "colab_type": "text"
      },
      "source": [
        "https://pytorch.org/docs/master/_modules/torchvision/models/inception.html"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rpAr08aGIOle",
        "colab_type": "code",
        "outputId": "b80727ae-2a3f-43e6-ced2-b7ee424c11ef",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        }
      },
      "source": [
        "inception = models.inception_v3(pretrained=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading: \"https://download.pytorch.org/models/inception_v3_google-1a9a5a14.pth\" to /root/.cache/torch/checkpoints/inception_v3_google-1a9a5a14.pth\n",
            "100%|██████████| 104M/104M [00:02<00:00, 44.2MB/s] \n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TM_ACRUHdMfi",
        "colab_type": "code",
        "outputId": "b176f93d-a11c-4542-f2ca-9a81c07b22f4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "print(inception)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Inception3(\n",
            "  (Conv2d_1a_3x3): BasicConv2d(\n",
            "    (conv): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), bias=False)\n",
            "    (bn): BatchNorm2d(32, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "  )\n",
            "  (Conv2d_2a_3x3): BasicConv2d(\n",
            "    (conv): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), bias=False)\n",
            "    (bn): BatchNorm2d(32, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "  )\n",
            "  (Conv2d_2b_3x3): BasicConv2d(\n",
            "    (conv): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "    (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "  )\n",
            "  (Conv2d_3b_1x1): BasicConv2d(\n",
            "    (conv): Conv2d(64, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "    (bn): BatchNorm2d(80, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "  )\n",
            "  (Conv2d_4a_3x3): BasicConv2d(\n",
            "    (conv): Conv2d(80, 192, kernel_size=(3, 3), stride=(1, 1), bias=False)\n",
            "    (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "  )\n",
            "  (Mixed_5b): InceptionA(\n",
            "    (branch1x1): BasicConv2d(\n",
            "      (conv): Conv2d(192, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch5x5_1): BasicConv2d(\n",
            "      (conv): Conv2d(192, 48, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(48, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch5x5_2): BasicConv2d(\n",
            "      (conv): Conv2d(48, 64, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(192, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(64, 96, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_3): BasicConv2d(\n",
            "      (conv): Conv2d(96, 96, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch_pool): BasicConv2d(\n",
            "      (conv): Conv2d(192, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(32, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (Mixed_5c): InceptionA(\n",
            "    (branch1x1): BasicConv2d(\n",
            "      (conv): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch5x5_1): BasicConv2d(\n",
            "      (conv): Conv2d(256, 48, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(48, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch5x5_2): BasicConv2d(\n",
            "      (conv): Conv2d(48, 64, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(64, 96, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_3): BasicConv2d(\n",
            "      (conv): Conv2d(96, 96, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch_pool): BasicConv2d(\n",
            "      (conv): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (Mixed_5d): InceptionA(\n",
            "    (branch1x1): BasicConv2d(\n",
            "      (conv): Conv2d(288, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch5x5_1): BasicConv2d(\n",
            "      (conv): Conv2d(288, 48, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(48, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch5x5_2): BasicConv2d(\n",
            "      (conv): Conv2d(48, 64, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(288, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(64, 96, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_3): BasicConv2d(\n",
            "      (conv): Conv2d(96, 96, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch_pool): BasicConv2d(\n",
            "      (conv): Conv2d(288, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (Mixed_6a): InceptionB(\n",
            "    (branch3x3): BasicConv2d(\n",
            "      (conv): Conv2d(288, 384, kernel_size=(3, 3), stride=(2, 2), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(288, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(64, 96, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_3): BasicConv2d(\n",
            "      (conv): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), bias=False)\n",
            "      (bn): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (Mixed_6b): InceptionC(\n",
            "    (branch1x1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_2): BasicConv2d(\n",
            "      (conv): Conv2d(128, 128, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_3): BasicConv2d(\n",
            "      (conv): Conv2d(128, 192, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(128, 128, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_3): BasicConv2d(\n",
            "      (conv): Conv2d(128, 128, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_4): BasicConv2d(\n",
            "      (conv): Conv2d(128, 128, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_5): BasicConv2d(\n",
            "      (conv): Conv2d(128, 192, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch_pool): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (Mixed_6c): InceptionC(\n",
            "    (branch1x1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_2): BasicConv2d(\n",
            "      (conv): Conv2d(160, 160, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_3): BasicConv2d(\n",
            "      (conv): Conv2d(160, 192, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(160, 160, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_3): BasicConv2d(\n",
            "      (conv): Conv2d(160, 160, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_4): BasicConv2d(\n",
            "      (conv): Conv2d(160, 160, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_5): BasicConv2d(\n",
            "      (conv): Conv2d(160, 192, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch_pool): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (Mixed_6d): InceptionC(\n",
            "    (branch1x1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_2): BasicConv2d(\n",
            "      (conv): Conv2d(160, 160, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_3): BasicConv2d(\n",
            "      (conv): Conv2d(160, 192, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(160, 160, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_3): BasicConv2d(\n",
            "      (conv): Conv2d(160, 160, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_4): BasicConv2d(\n",
            "      (conv): Conv2d(160, 160, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_5): BasicConv2d(\n",
            "      (conv): Conv2d(160, 192, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch_pool): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (Mixed_6e): InceptionC(\n",
            "    (branch1x1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_2): BasicConv2d(\n",
            "      (conv): Conv2d(192, 192, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7_3): BasicConv2d(\n",
            "      (conv): Conv2d(192, 192, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(192, 192, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_3): BasicConv2d(\n",
            "      (conv): Conv2d(192, 192, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_4): BasicConv2d(\n",
            "      (conv): Conv2d(192, 192, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7dbl_5): BasicConv2d(\n",
            "      (conv): Conv2d(192, 192, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch_pool): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (AuxLogits): InceptionAux(\n",
            "    (conv0): BasicConv2d(\n",
            "      (conv): Conv2d(768, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (conv1): BasicConv2d(\n",
            "      (conv): Conv2d(128, 768, kernel_size=(5, 5), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (fc): Linear(in_features=768, out_features=1000, bias=True)\n",
            "  )\n",
            "  (Mixed_7a): InceptionD(\n",
            "    (branch3x3_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3_2): BasicConv2d(\n",
            "      (conv): Conv2d(192, 320, kernel_size=(3, 3), stride=(2, 2), bias=False)\n",
            "      (bn): BatchNorm2d(320, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7x3_1): BasicConv2d(\n",
            "      (conv): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7x3_2): BasicConv2d(\n",
            "      (conv): Conv2d(192, 192, kernel_size=(1, 7), stride=(1, 1), padding=(0, 3), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7x3_3): BasicConv2d(\n",
            "      (conv): Conv2d(192, 192, kernel_size=(7, 1), stride=(1, 1), padding=(3, 0), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch7x7x3_4): BasicConv2d(\n",
            "      (conv): Conv2d(192, 192, kernel_size=(3, 3), stride=(2, 2), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (Mixed_7b): InceptionE(\n",
            "    (branch1x1): BasicConv2d(\n",
            "      (conv): Conv2d(1280, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(320, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3_1): BasicConv2d(\n",
            "      (conv): Conv2d(1280, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3_2a): BasicConv2d(\n",
            "      (conv): Conv2d(384, 384, kernel_size=(1, 3), stride=(1, 1), padding=(0, 1), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3_2b): BasicConv2d(\n",
            "      (conv): Conv2d(384, 384, kernel_size=(3, 1), stride=(1, 1), padding=(1, 0), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(1280, 448, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(448, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(448, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_3a): BasicConv2d(\n",
            "      (conv): Conv2d(384, 384, kernel_size=(1, 3), stride=(1, 1), padding=(0, 1), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_3b): BasicConv2d(\n",
            "      (conv): Conv2d(384, 384, kernel_size=(3, 1), stride=(1, 1), padding=(1, 0), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch_pool): BasicConv2d(\n",
            "      (conv): Conv2d(1280, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (Mixed_7c): InceptionE(\n",
            "    (branch1x1): BasicConv2d(\n",
            "      (conv): Conv2d(2048, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(320, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3_1): BasicConv2d(\n",
            "      (conv): Conv2d(2048, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3_2a): BasicConv2d(\n",
            "      (conv): Conv2d(384, 384, kernel_size=(1, 3), stride=(1, 1), padding=(0, 1), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3_2b): BasicConv2d(\n",
            "      (conv): Conv2d(384, 384, kernel_size=(3, 1), stride=(1, 1), padding=(1, 0), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_1): BasicConv2d(\n",
            "      (conv): Conv2d(2048, 448, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(448, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_2): BasicConv2d(\n",
            "      (conv): Conv2d(448, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_3a): BasicConv2d(\n",
            "      (conv): Conv2d(384, 384, kernel_size=(1, 3), stride=(1, 1), padding=(0, 1), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch3x3dbl_3b): BasicConv2d(\n",
            "      (conv): Conv2d(384, 384, kernel_size=(3, 1), stride=(1, 1), padding=(1, 0), bias=False)\n",
            "      (bn): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "    (branch_pool): BasicConv2d(\n",
            "      (conv): Conv2d(2048, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
            "      (bn): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n",
            "    )\n",
            "  )\n",
            "  (fc): Linear(in_features=2048, out_features=1000, bias=True)\n",
            ")\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Nbu0JobP6ea9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for param in inception.parameters():\n",
        "    param.requires_grad = False"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j31it2qX5Nfs",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "aux_in_features = inception.AuxLogits.fc.in_features\n",
        "inception.AuxLogits.fc = nn.Linear(aux_in_features, num_classes)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7-M6nGOh6Xg2",
        "colab_type": "code",
        "outputId": "62978c0e-446e-4540-daa0-c7c7c20c0502",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "for param in inception.parameters():\n",
        "    if param.requires_grad:\n",
        "        print(param.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "torch.Size([10, 768])\n",
            "torch.Size([10])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pOI1Au2A6l9X",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "in_features = inception.fc.in_features\n",
        "inception.fc = nn.Linear(in_features, num_classes)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sF5pPlUS7I-7",
        "colab_type": "code",
        "outputId": "aa795347-a946-409a-c659-8a6de2316752",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        }
      },
      "source": [
        "for param in inception.parameters():\n",
        "    if param.requires_grad:\n",
        "        print(param.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "torch.Size([10, 768])\n",
            "torch.Size([10])\n",
            "torch.Size([10, 2048])\n",
            "torch.Size([10])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vMSMZa_j7JXG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "transform_train = transforms.Compose([\n",
        "    transforms.RandomResizedCrop(299), \n",
        "    transforms.ToTensor(),\n",
        "    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),\n",
        "    ])\n",
        "\n",
        "transform_test = transforms.Compose([\n",
        "    transforms.RandomResizedCrop(299), \n",
        "    transforms.ToTensor(),\n",
        "    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),\n",
        "    ])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kYVy8C0H7vL7",
        "colab_type": "code",
        "outputId": "3e2d295a-8701-4828-eda5-65a8525ee851",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "trainset = torchvision.datasets.CIFAR10(root='./data', train=True, \n",
        "                                        download=True, \n",
        "                                        transform=transform_train)\n",
        "testset = torchvision.datasets.CIFAR10(root='./data', train=False, \n",
        "                                        download=True, \n",
        "                                        transform=transform_test)"
      ],
      "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": "tkUGBe3x7zld",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "batch_size=16\n",
        "\n",
        "trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True)\n",
        "testloader = torch.utils.data.DataLoader(testset, batch_size=batch_size, shuffle=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jeKDmSV98QfZ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "inception = inception.to(device)\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "opt = optim.SGD(inception.parameters(), lr=0.01)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ou_Q_JwK_fqn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def evaluation_inception(dataloader, model):\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, aux_outputs = model(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": "VqmnOJqP75AK",
        "colab_type": "code",
        "outputId": "7fd0770d-5fa7-4785-e86a-77b189d65b86",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "%%time\n",
        "\n",
        "loss_epoch_arr = []\n",
        "max_epochs = 1\n",
        "\n",
        "min_loss = 1000\n",
        "\n",
        "n_iters = np.ceil(50000/batch_size)\n",
        "\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, aux_outputs = inception(inputs)\n",
        "        loss = loss_fn(outputs, labels) + 0.3 * loss_fn(aux_outputs, labels)\n",
        "        loss.backward()\n",
        "        opt.step()\n",
        "        \n",
        "        if min_loss > loss.item():\n",
        "            min_loss = loss.item()\n",
        "            best_model = copy.deepcopy(inception.state_dict())\n",
        "            print('Min loss %0.2f' % min_loss)\n",
        "        \n",
        "        if i % 100 == 0:\n",
        "            print('Iteration: %d/%d, Loss: %0.2f' % (i, n_iters, loss.item()))\n",
        "            \n",
        "        del inputs, labels, outputs\n",
        "        torch.cuda.empty_cache()\n",
        "        \n",
        "        loss_epoch_arr.append(loss.item())\n",
        "        \n",
        "    print('Epoch: %d/%d, Test acc: %0.2f, Train acc: %0.2f' % (\n",
        "        epoch, max_epochs, \n",
        "        evaluation_inception(testloader, inception), \n",
        "        evaluation_inception(trainloader, inception)))\n",
        "    \n",
        "    \n",
        "plt.plot(loss_epoch_arr)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Min loss 2.93\n",
            "Iteration: 0/3125, Loss: 2.93\n",
            "Min loss 2.87\n",
            "Min loss 2.87\n",
            "Min loss 2.82\n",
            "Min loss 2.76\n",
            "Min loss 2.76\n",
            "Min loss 2.72\n",
            "Min loss 2.67\n",
            "Min loss 2.62\n",
            "Min loss 2.60\n",
            "Min loss 2.57\n",
            "Min loss 2.56\n",
            "Min loss 2.51\n",
            "Iteration: 100/3125, Loss: 2.51\n",
            "Min loss 2.45\n",
            "Min loss 2.33\n",
            "Min loss 2.29\n",
            "Iteration: 200/3125, Loss: 2.40\n",
            "Min loss 2.23\n",
            "Min loss 2.17\n",
            "Min loss 2.16\n",
            "Iteration: 300/3125, Loss: 2.16\n",
            "Min loss 2.10\n",
            "Min loss 2.07\n",
            "Min loss 2.03\n",
            "Min loss 1.97\n",
            "Min loss 1.96\n",
            "Iteration: 400/3125, Loss: 1.96\n",
            "Min loss 1.83\n",
            "Iteration: 500/3125, Loss: 1.87\n",
            "Min loss 1.82\n",
            "Min loss 1.78\n",
            "Min loss 1.77\n",
            "Iteration: 600/3125, Loss: 2.30\n",
            "Min loss 1.72\n",
            "Iteration: 700/3125, Loss: 2.39\n",
            "Min loss 1.65\n",
            "Min loss 1.65\n",
            "Min loss 1.60\n",
            "Iteration: 800/3125, Loss: 2.43\n",
            "Min loss 1.49\n",
            "Iteration: 900/3125, Loss: 2.40\n",
            "Min loss 1.34\n",
            "Iteration: 1000/3125, Loss: 1.74\n",
            "Min loss 1.31\n",
            "Iteration: 1100/3125, Loss: 2.39\n",
            "Min loss 1.30\n",
            "Iteration: 1200/3125, Loss: 2.99\n",
            "Iteration: 1300/3125, Loss: 2.48\n",
            "Iteration: 1400/3125, Loss: 2.34\n",
            "Min loss 1.17\n",
            "Iteration: 1500/3125, Loss: 2.04\n",
            "Iteration: 1600/3125, Loss: 1.57\n",
            "Iteration: 1700/3125, Loss: 2.12\n",
            "Iteration: 1800/3125, Loss: 2.14\n",
            "Iteration: 1900/3125, Loss: 1.59\n",
            "Iteration: 2000/3125, Loss: 1.90\n",
            "Iteration: 2100/3125, Loss: 1.65\n",
            "Iteration: 2200/3125, Loss: 2.24\n",
            "Iteration: 2300/3125, Loss: 2.23\n",
            "Iteration: 2400/3125, Loss: 1.82\n",
            "Min loss 1.11\n",
            "Iteration: 2500/3125, Loss: 1.17\n",
            "Iteration: 2600/3125, Loss: 1.86\n",
            "Iteration: 2700/3125, Loss: 2.01\n",
            "Min loss 1.04\n",
            "Iteration: 2800/3125, Loss: 2.20\n",
            "Iteration: 2900/3125, Loss: 1.83\n",
            "Iteration: 3000/3125, Loss: 2.15\n",
            "Iteration: 3100/3125, Loss: 2.15\n",
            "Epoch: 0/1, Test acc: 47.04, Train acc: 47.34\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAD7CAYAAABgzo9kAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 8min 4s, sys: 1min 39s, total: 9min 43s\n",
            "Wall time: 9min 44s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YfcUcxaL8T_U",
        "colab_type": "code",
        "outputId": "1f5b2659-7db1-4a49-cc69-73d9770dfd23",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "inception.load_state_dict(best_model)\n",
        "print(evaluation_inception(trainloader, inception), evaluation_inception(testloader, inception))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "47.13 47.45\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "on1EjhmKmbyp",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}