finetune-resnext.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 102,
      "metadata": {
        "id": "W4i8tF1LP7Bw"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import cv2\n",
        "import numpy as np\n",
        "from torch.utils.data import Dataset, DataLoader\n",
        "import torchvision.transforms as transforms\n",
        "import torch\n",
        "import tqdm\n",
        "import torch.nn as nn"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 103,
      "metadata": {
        "id": "aJyGePRgP7Bx"
      },
      "outputs": [],
      "source": [
        "transform = transforms.Compose([\n",
        "    transforms.ToPILImage(),\n",
        "    transforms.Resize((256, 256)),\n",
        "    transforms.ToTensor(),\n",
        "    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),\n",
        "])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 104,
      "metadata": {
        "id": "7TclxM3CP7Bx"
      },
      "outputs": [],
      "source": [
        "# check device\n",
        "device = 'cuda' if torch.cuda.is_available() else 'cpu'"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 105,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2pFM7gHVP7Bx",
        "outputId": "db6cd16e-83bb-4d5a-d752-b128a880cd99"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "torch.Size([999, 3, 256, 256]) torch.Size([999])\n"
          ]
        }
      ],
      "source": [
        "\n",
        "classes = os.listdir('Data/images_original')\n",
        "classes.sort()\n",
        "class_to_idx = {cls_name: i for i, cls_name in enumerate(classes)}\n",
        "idx_to_class = {i: cls_name for i, cls_name in enumerate(classes)}\n",
        "\n",
        "images = []\n",
        "labels = []\n",
        "for i, cls_name in enumerate(classes):\n",
        "    path = os.path.join('Data/images_original', cls_name)\n",
        "    for img_file in os.listdir(path):\n",
        "        img_path = os.path.join(path, img_file)\n",
        "        img = cv2.imread(img_path)\n",
        "        img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
        "        img = transform(img)\n",
        "        images.append(img)\n",
        "        labels.append(i)\n",
        "images = torch.stack(images).to(device)\n",
        "labels = torch.tensor(labels).to(device)\n",
        "print(images.shape, labels.shape)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 106,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ABtfMMt3P7By",
        "outputId": "2aa620cd-78dd-4346-d8b3-9ca4ac318455"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "torch.Size([799, 3, 256, 256]) torch.Size([200, 3, 256, 256]) torch.Size([799]) torch.Size([200])\n"
          ]
        }
      ],
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "train_images, test_images, train_labels, test_labels = train_test_split(images, labels, shuffle=True, test_size=0.2, random_state=42)\n",
        "print(train_images.shape, test_images.shape, train_labels.shape, test_labels.shape)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 107,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 507
        },
        "id": "VKGXX7pDP7By",
        "outputId": "2f940549-abab-478b-b437-87fc85384cd6"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:matplotlib.image:Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'jazz')"
            ]
          },
          "execution_count": 107,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# print a image from the dataset\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.imshow(train_images[110].cpu().numpy().transpose(1, 2, 0))\n",
        "plt.title(idx_to_class[train_labels[110].item()])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 108,
      "metadata": {
        "id": "3o1wKwGNP7By"
      },
      "outputs": [],
      "source": [
        "\n",
        "class MusicDataset(Dataset):\n",
        "    def __init__(self, images, labels):\n",
        "        self.images = images\n",
        "        self.labels = labels\n",
        "\n",
        "    def __len__(self):\n",
        "        return len(self.images)\n",
        "\n",
        "    def __getitem__(self, idx):\n",
        "        return self.images[idx], self.labels[idx]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 109,
      "metadata": {
        "id": "-xKBU_LUP7Bz"
      },
      "outputs": [],
      "source": [
        "dataset_train = MusicDataset(train_images, train_labels)\n",
        "DataLoader_train = DataLoader(dataset_train, batch_size=50, shuffle=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 110,
      "metadata": {
        "id": "0YTm3i37P7Bz"
      },
      "outputs": [],
      "source": [
        "import torch.nn as nn\n",
        "\n",
        "class Identity(nn.Module):\n",
        "    def __init__(self):\n",
        "        super(Identity, self).__init__()\n",
        "\n",
        "    def forward(self, x):\n",
        "        return x"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 125,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0QOtQI7vQBPU",
        "outputId": "834ee545-e504-4fb0-d583-f3100c4c6a98"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "Using cache found in /root/.cache/torch/hub/facebookresearch_WSL-Images_main\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "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): Bottleneck(\n",
              "      (conv1): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "      (downsample): Sequential(\n",
              "        (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      )\n",
              "    )\n",
              "    (1): Bottleneck(\n",
              "      (conv1): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (2): Bottleneck(\n",
              "      (conv1): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "  )\n",
              "  (layer2): Sequential(\n",
              "    (0): Bottleneck(\n",
              "      (conv1): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=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): Bottleneck(\n",
              "      (conv1): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (2): Bottleneck(\n",
              "      (conv1): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (3): Bottleneck(\n",
              "      (conv1): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "  )\n",
              "  (layer3): Sequential(\n",
              "    (0): Bottleneck(\n",
              "      (conv1): Conv2d(512, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "      (downsample): Sequential(\n",
              "        (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
              "        (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      )\n",
              "    )\n",
              "    (1): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (2): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (3): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (4): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (5): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (6): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (7): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (8): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (9): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (10): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (11): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (12): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (13): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (14): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (15): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (16): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (17): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (18): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (19): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (20): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (21): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (22): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "  )\n",
              "  (layer4): Sequential(\n",
              "    (0): Bottleneck(\n",
              "      (conv1): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(2048, 2048, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(2048, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "      (downsample): Sequential(\n",
              "        (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
              "        (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      )\n",
              "    )\n",
              "    (1): Bottleneck(\n",
              "      (conv1): Conv2d(2048, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(2048, 2048, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(2048, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "    (2): Bottleneck(\n",
              "      (conv1): Conv2d(2048, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv2): Conv2d(2048, 2048, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
              "      (bn2): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (conv3): Conv2d(2048, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
              "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
              "      (relu): ReLU(inplace=True)\n",
              "    )\n",
              "  )\n",
              "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
              "  (fc): Linear(in_features=2048, out_features=1000, bias=True)\n",
              ")"
            ]
          },
          "execution_count": 125,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "model = torch.hub.load('facebookresearch/WSL-Images', 'resnext101_32x8d_wsl')\n",
        "model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 126,
      "metadata": {
        "id": "Rj70EL_cP7Bz"
      },
      "outputs": [],
      "source": [
        "for param in model.parameters():\n",
        "    param.requires_grad = False\n",
        "\n",
        "\n",
        "model.fc = nn.Sequential(\n",
        "    nn.Linear(2048, 10, bias=True)\n",
        ")\n",
        "\n",
        "model = model.to(device)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 127,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-Hll5WQbP7Bz",
        "outputId": "c21eeb31-e006-411b-a84c-cf0792deac52"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.33it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 0, loss_train: 0.11478875577449799, loss_val: 1.881117820739746, accuracy_val: 0.365\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.31it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1, loss_train: 0.10236429423093796, loss_val: 1.623960018157959, accuracy_val: 0.43\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.31it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 2, loss_train: 0.08984575420618057, loss_val: 1.5840507745742798, accuracy_val: 0.43\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.33it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 3, loss_train: 0.0701548308134079, loss_val: 1.4702978134155273, accuracy_val: 0.51\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 4, loss_train: 0.063980832695961, loss_val: 1.4157978296279907, accuracy_val: 0.52\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 5, loss_train: 0.071120984852314, loss_val: 1.388016700744629, accuracy_val: 0.535\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 6, loss_train: 0.05186191946268082, loss_val: 1.3330879211425781, accuracy_val: 0.55\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.35it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 7, loss_train: 0.07001405209302902, loss_val: 1.3291757106781006, accuracy_val: 0.545\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.36it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 8, loss_train: 0.052440617233514786, loss_val: 1.3430793285369873, accuracy_val: 0.54\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.35it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 9, loss_train: 0.05039546266198158, loss_val: 1.3121562004089355, accuracy_val: 0.565\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.36it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 10, loss_train: 0.05532374978065491, loss_val: 1.2949752807617188, accuracy_val: 0.59\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.37it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 11, loss_train: 0.041720181703567505, loss_val: 1.2665438652038574, accuracy_val: 0.565\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.36it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 12, loss_train: 0.056451015174388885, loss_val: 1.2961713075637817, accuracy_val: 0.58\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.35it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 13, loss_train: 0.042470533400774, loss_val: 1.2772961854934692, accuracy_val: 0.595\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.35it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 14, loss_train: 0.03477953001856804, loss_val: 1.2607040405273438, accuracy_val: 0.6\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 15, loss_train: 0.04371544346213341, loss_val: 1.2763592004776, accuracy_val: 0.605\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.33it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 16, loss_train: 0.03188919648528099, loss_val: 1.2411961555480957, accuracy_val: 0.6\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.33it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 17, loss_train: 0.03942263126373291, loss_val: 1.225676417350769, accuracy_val: 0.57\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 18, loss_train: 0.0313294492661953, loss_val: 1.2657263278961182, accuracy_val: 0.575\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.33it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 19, loss_train: 0.03801688551902771, loss_val: 1.265925645828247, accuracy_val: 0.555\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 20, loss_train: 0.04068251699209213, loss_val: 1.3039536476135254, accuracy_val: 0.565\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.33it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 21, loss_train: 0.04244745895266533, loss_val: 1.3284536600112915, accuracy_val: 0.55\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 22, loss_train: 0.029833542183041573, loss_val: 1.2515047788619995, accuracy_val: 0.59\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 23, loss_train: 0.027373570948839188, loss_val: 1.263946533203125, accuracy_val: 0.565\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 24, loss_train: 0.026955127716064453, loss_val: 1.2406082153320312, accuracy_val: 0.625\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 25, loss_train: 0.02348511479794979, loss_val: 1.2290540933609009, accuracy_val: 0.605\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.35it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 26, loss_train: 0.022513793781399727, loss_val: 1.2188948392868042, accuracy_val: 0.58\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.36it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 27, loss_train: 0.021640747785568237, loss_val: 1.2370779514312744, accuracy_val: 0.575\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 28, loss_train: 0.02623559720814228, loss_val: 1.2319504022598267, accuracy_val: 0.61\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 29, loss_train: 0.019909586757421494, loss_val: 1.2323893308639526, accuracy_val: 0.585\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 30, loss_train: 0.021964065730571747, loss_val: 1.2564337253570557, accuracy_val: 0.59\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 31, loss_train: 0.014908966608345509, loss_val: 1.2490363121032715, accuracy_val: 0.61\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.33it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 32, loss_train: 0.022232089191675186, loss_val: 1.2567737102508545, accuracy_val: 0.6\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 33, loss_train: 0.016659043729305267, loss_val: 1.2478070259094238, accuracy_val: 0.59\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 34, loss_train: 0.020477300509810448, loss_val: 1.2099888324737549, accuracy_val: 0.615\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 35, loss_train: 0.017279377207159996, loss_val: 1.280646800994873, accuracy_val: 0.595\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 36, loss_train: 0.018698308616876602, loss_val: 1.2106356620788574, accuracy_val: 0.615\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 37, loss_train: 0.01820881851017475, loss_val: 1.263532280921936, accuracy_val: 0.615\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.33it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 38, loss_train: 0.013737893663346767, loss_val: 1.2671102285385132, accuracy_val: 0.585\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 39, loss_train: 0.018050305545330048, loss_val: 1.321760654449463, accuracy_val: 0.595\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 40, loss_train: 0.014775656163692474, loss_val: 1.2316690683364868, accuracy_val: 0.61\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 41, loss_train: 0.012327042408287525, loss_val: 1.2682759761810303, accuracy_val: 0.605\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 42, loss_train: 0.01849776692688465, loss_val: 1.242411494255066, accuracy_val: 0.6\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 43, loss_train: 0.014365823939442635, loss_val: 1.2163028717041016, accuracy_val: 0.615\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 44, loss_train: 0.012473024427890778, loss_val: 1.297045111656189, accuracy_val: 0.62\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 45, loss_train: 0.010219295509159565, loss_val: 1.2619719505310059, accuracy_val: 0.625\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 46, loss_train: 0.01081455685198307, loss_val: 1.2433280944824219, accuracy_val: 0.6\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 47, loss_train: 0.012277667410671711, loss_val: 1.2631555795669556, accuracy_val: 0.59\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 48, loss_train: 0.01071406714618206, loss_val: 1.27071213722229, accuracy_val: 0.59\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 16/16 [00:04<00:00,  3.34it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 49, loss_train: 0.010659119114279747, loss_val: 1.263744831085205, accuracy_val: 0.595\n",
            "Early stopping at epoch 49\n"
          ]
        }
      ],
      "source": [
        "# 定义损失函数和优化器\n",
        "criterion = nn.CrossEntropyLoss()\n",
        "optimizer = torch.optim.Adam(model.parameters())\n",
        "\n",
        "# 训练模型\n",
        "num_epochs = 200\n",
        "patience = 15  # Number of epochs to wait for improvement before stopping\n",
        "best_val_loss = float('inf')  # Initialize best validation loss\n",
        "patience_counter = 0  # Counter to keep track of the number of epochs without improvement\n",
        "log = {}\n",
        "\n",
        "\n",
        "for epoch in range(num_epochs):\n",
        "    model.train()\n",
        "    run_loss = 0.0\n",
        "    for images, labels in tqdm.tqdm(DataLoader_train):\n",
        "        optimizer.zero_grad()\n",
        "        outputs = model(images)\n",
        "        loss = criterion(outputs, labels)\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "        run_loss += loss.item()\n",
        "\n",
        "    model.eval()\n",
        "    with torch.no_grad():\n",
        "        test_outputs = model(test_images)\n",
        "        test_loss = criterion(test_outputs, test_labels)\n",
        "        pred = torch.argmax(test_outputs, dim=1)\n",
        "        acc = (pred == test_labels).sum().item() / len(test_labels)\n",
        "\n",
        "        # save the loss and accuracy\n",
        "        log[epoch] = {'loss_train': loss.item(), 'loss_val': test_loss.item(), 'accuracy_val': acc}\n",
        "        print(f'Epoch {epoch}, loss_train: {loss.item() / len(DataLoader_train)}, loss_val: {test_loss.item()}, accuracy_val: {acc}')\n",
        "\n",
        "    if test_loss < best_val_loss:\n",
        "        best_val_loss = test_loss\n",
        "        patience_counter = 0\n",
        "        torch.save(model.state_dict(), 'find-tuned-resnext.pth')\n",
        "    else:\n",
        "        patience_counter += 1\n",
        "        if patience_counter >= patience:\n",
        "            print(f'Early stopping at epoch {epoch}')\n",
        "            break"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 128,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 430
        },
        "id": "kFlKYw6FP7Bz",
        "outputId": "aeee7b28-108f-46c9-c0c2-a1a19ac7f5c8"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# draw the loss and accuracy\n",
        "import matplotlib.pyplot as plt\n",
        "plt.figure()\n",
        "plt.plot([log[k]['loss_train'] for k in log], label='train loss')\n",
        "plt.plot([log[k]['loss_val'] for k in log], label='val loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 130,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tju0xNNqRsbv",
        "outputId": "dfc8ca17-dc5e-428d-f241-eb5e28fa9f17"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.71      0.48      0.57        21\n",
            "           1       0.75      1.00      0.86        12\n",
            "           2       0.57      0.54      0.55        24\n",
            "           3       0.45      0.45      0.45        22\n",
            "           4       0.65      0.87      0.74        15\n",
            "           5       0.67      0.81      0.73        27\n",
            "           6       0.82      0.78      0.80        18\n",
            "           7       0.59      0.68      0.63        19\n",
            "           8       0.58      0.50      0.54        22\n",
            "           9       0.36      0.25      0.29        20\n",
            "\n",
            "    accuracy                           0.61       200\n",
            "   macro avg       0.62      0.64      0.62       200\n",
            "weighted avg       0.61      0.61      0.60       200\n",
            "\n"
          ]
        }
      ],
      "source": [
        "# load the model\n",
        "model.load_state_dict(torch.load('find-tuned-resnext.pth'))\n",
        "\n",
        "# test the model\n",
        "model.eval()\n",
        "from sklearn.metrics import classification_report\n",
        "\n",
        "print(classification_report(test_labels.detach().cpu().numpy(), torch.argmax(model(test_images), dim=1).detach().cpu().numpy()))"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.8.18"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}