VisionTransformer.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 129,
   "id": "6c8ff150-15e6-4fea-bdcc-acbcbac5be81",
   "metadata": {
    "id": "npmDe_CGKoPO"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "import torchvision.transforms as transforms\n",
    "import torchvision.models as models\n",
    "from PIL import Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "id": "3a18fe77-63ed-45b2-bcd3-611ea2354a73",
   "metadata": {},
   "outputs": [],
   "source": [
    "gpu_id = 0  # Change this to the index of the desired GPU\n",
    "torch.cuda.set_device(gpu_id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "id": "b64e675e-96e4-42e1-8ef2-bec0fe125071",
   "metadata": {
    "id": "rGAyI9xdKxTB"
   },
   "outputs": [],
   "source": [
    "DATADIR = \"RediMinds/Data_Cleaned/\"\n",
    "categories = [\"Anaesthesia_machine\",\"baby_incubator\",\"Bone_saws\",\"C_arm\",\"colonoscope\",\"Curved_Mayo_Scissor\",\"difibrillator\",\"Electrocautery_devices\",\"fetal_doppler\",\"forceps\",\"Heart_Lung_Machine\",\"IABP\",\"IMRT\",\"infusion_pump\",\"Laryngoscopes\",\"mayfield_clamp\",\"Needle_Biopsy_Device\",\"phacoemulsification\",\"Radiofrequency_Ablation_Device\",\"Scalpel\",\"Straight_Dissection_Clamp\",\"Straight_Mayo_Scissor\",\"Suction_Machine\",\"ventilator\",\"x_ray\"]\n",
    "input_size = (224, 224)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "id": "5ce2df18-f8a9-4324-acf9-490a8ad68509",
   "metadata": {
    "id": "d0mpzTyLK1l2"
   },
   "outputs": [],
   "source": [
    "class CustomDataset(Dataset):\n",
    "    def __init__(self, data_dir, categories, input_size, transform=None):\n",
    "        self.data_dir = data_dir\n",
    "        self.categories = categories\n",
    "        self.input_size = input_size\n",
    "        self.transform = transform\n",
    "        self.data = []\n",
    "        self.labels = []\n",
    "        self.load_data()\n",
    "\n",
    "    def load_data(self):\n",
    "        for cat in self.categories:\n",
    "            path = os.path.join(self.data_dir, cat)\n",
    "            class_num = self.categories.index(cat)\n",
    "            for img_name in os.listdir(path):\n",
    "                try:\n",
    "                    img_path = os.path.join(path, img_name)\n",
    "                    img = Image.open(img_path).convert(\"RGB\")\n",
    "                    img = img.resize(self.input_size)\n",
    "                    if self.transform:\n",
    "                        img = self.transform(img)\n",
    "                    self.data.append(img)\n",
    "                    self.labels.append(class_num)\n",
    "                except Exception as e:\n",
    "                    pass\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.data)\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        img = self.data[idx]\n",
    "        label = self.labels[idx]\n",
    "        return img, label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "id": "5ede01ba-380e-4c72-83ac-c417e22fa1cb",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "4y-YObECLeC2",
    "outputId": "37531a35-eaa5-45a9-a0d6-b7a5e3d683ad"
   },
   "outputs": [],
   "source": [
    "transform = transforms.Compose([\n",
    "    transforms.RandomResizedCrop(input_size, scale=(0.8, 1.0)),\n",
    "    transforms.RandomHorizontalFlip(),\n",
    "    # transforms.CenterCrop(10),\n",
    "    # transforms.RandomRotation(15),\n",
    "    transforms.ColorJitter(brightness=0.2, contrast = 0.25, saturation = 0.2, hue=0.1),\n",
    "    transforms.ToTensor(),\n",
    "    transforms.Normalize((0.5, 0.5, 0.5), (0.2, 0.2, 0.2))\n",
    "])\n",
    "\n",
    "dataset = CustomDataset(DATADIR, categories, input_size, transform=transform)\n",
    "\n",
    "train_size = int(0.70 * len(dataset))\n",
    "validation_size = int(0.15 * len(dataset))\n",
    "test_size = len(dataset) - train_size - validation_size\n",
    "train_dataset, validation_dataset, test_dataset = torch.utils.data.random_split(dataset, [train_size,validation_size, test_size])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "id": "b93b6748-9a1c-4739-9985-d6db8c702efc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3344 716 718\n"
     ]
    }
   ],
   "source": [
    "print(train_size,validation_size,test_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "id": "b8b9b514-05c0-4bc6-994f-24da1988fe6c",
   "metadata": {
    "id": "HDjJkmtgLjVZ"
   },
   "outputs": [],
   "source": [
    "batch_size = 32\n",
    "train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
    "validation_loader = DataLoader(validation_dataset, batch_size=batch_size)\n",
    "test_loader = DataLoader(test_dataset, batch_size=batch_size)\n",
    "\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "id": "7533de0c-9ae8-4850-9bf1-270d32a97e1a",
   "metadata": {},
   "outputs": [],
   "source": [
    "class CustomVisionTransformer(nn.Module):\n",
    "    def __init__(self, num_classes):\n",
    "        super(CustomVisionTransformer, self).__init__()\n",
    "        vistrans = models.vit_l_32(pretrained=models.ViT_L_32_Weights)\n",
    "\n",
    "        # Freeze all layers by default\n",
    "        for param in vistrans.parameters():\n",
    "            param.requires_grad = False\n",
    "\n",
    "        # Unfreeze the last few layers of the Vision Transformer\n",
    "        unfreeze_layers = ['layers.encoder_layer_22', 'layers.encoder_layer_23']\n",
    "        for name, param in vistrans.named_parameters():\n",
    "            for layer_name in unfreeze_layers:\n",
    "                if layer_name in name:\n",
    "                    param.requires_grad = True\n",
    "                    break\n",
    "                    \n",
    "        self.vistrans = vistrans\n",
    "        delattr(self.vistrans.heads, 'head')\n",
    "        \n",
    "        # CNN layers similar to the provided GoogLeNet model\n",
    "        self.cnnl = nn.Sequential(\n",
    "            nn.Conv2d(1, 512, kernel_size=3, padding=1),\n",
    "            nn.BatchNorm2d(512),\n",
    "            nn.ReLU(),\n",
    "            nn.Dropout(0.5),\n",
    "            nn.Conv2d(512, 512, kernel_size=3, padding=1),\n",
    "            nn.BatchNorm2d(512),\n",
    "            nn.ReLU(),\n",
    "            nn.Dropout(0.4),\n",
    "            # nn.Conv2d(512, 128, kernel_size=3, padding=1),\n",
    "            # nn.BatchNorm2d(128),\n",
    "            # nn.ReLU(),\n",
    "            # nn.Dropout(0.4),\n",
    "            nn.AdaptiveAvgPool2d((1, 1))\n",
    "        )\n",
    "\n",
    "        self.head = nn.Sequential(\n",
    "            # nn.Linear(128, 128),\n",
    "            # nn.ReLU(),\n",
    "            # nn.Dropout(0.4),\n",
    "            # nn.Linear(128, 64),\n",
    "            # nn.ReLU(),\n",
    "            # nn.Dropout(0.3),\n",
    "            nn.Linear(512, num_classes),\n",
    "        )\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.vistrans(x)\n",
    "        x = x[:, 0]  # Only keep the CLS token\n",
    "        x = x.view(x.size(0), -1)\n",
    "        x = x.view(x.size(0), -1, 1, 1)\n",
    "        x = self.cnnl(x)\n",
    "        x = x.view(x.size(0), -1)\n",
    "        x = self.head(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "id": "31a54ac2-9ac3-495c-b650-a4b8237c28ec",
   "metadata": {
    "editable": true,
    "id": "x03pIT-qMgF1",
    "slideshow": {
     "slide_type": ""
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "from torch.optim.lr_scheduler import StepLR\n",
    "num_classes = len(categories)\n",
    "model = CustomVisionTransformer(num_classes).to(device)\n",
    "\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = optim.Adam(model.parameters(), lr=0.005, weight_decay=1e-5)\n",
    "scheduler = StepLR(optimizer, step_size = 10, gamma=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "id": "c855fdd2-d9a6-4339-b3a6-1149bead5620",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "vistrans.class_token: requires_grad=False\n",
      "vistrans.conv_proj.weight: requires_grad=False\n",
      "vistrans.conv_proj.bias: requires_grad=False\n",
      "vistrans.encoder.pos_embedding: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_0.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_1.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_2.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_3.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_4.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_5.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_6.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_7.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_8.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_9.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_10.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_11.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_12.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_13.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_14.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_15.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_16.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_17.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_18.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_19.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_20.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.ln_1.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.ln_1.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.self_attention.in_proj_weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.self_attention.in_proj_bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.self_attention.out_proj.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.self_attention.out_proj.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.ln_2.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.ln_2.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.mlp.0.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.mlp.0.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.mlp.3.weight: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_21.mlp.3.bias: requires_grad=False\n",
      "vistrans.encoder.layers.encoder_layer_22.ln_1.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.ln_1.bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.self_attention.in_proj_weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.self_attention.in_proj_bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.self_attention.out_proj.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.self_attention.out_proj.bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.ln_2.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.ln_2.bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.mlp.0.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.mlp.0.bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.mlp.3.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_22.mlp.3.bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.ln_1.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.ln_1.bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.self_attention.in_proj_weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.self_attention.in_proj_bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.self_attention.out_proj.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.self_attention.out_proj.bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.ln_2.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.ln_2.bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.mlp.0.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.mlp.0.bias: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.mlp.3.weight: requires_grad=True\n",
      "vistrans.encoder.layers.encoder_layer_23.mlp.3.bias: requires_grad=True\n",
      "vistrans.encoder.ln.weight: requires_grad=False\n",
      "vistrans.encoder.ln.bias: requires_grad=False\n",
      "cnnl.0.weight: requires_grad=True\n",
      "cnnl.0.bias: requires_grad=True\n",
      "cnnl.1.weight: requires_grad=True\n",
      "cnnl.1.bias: requires_grad=True\n",
      "cnnl.4.weight: requires_grad=True\n",
      "cnnl.4.bias: requires_grad=True\n",
      "cnnl.5.weight: requires_grad=True\n",
      "cnnl.5.bias: requires_grad=True\n",
      "head.0.weight: requires_grad=True\n",
      "head.0.bias: requires_grad=True\n"
     ]
    }
   ],
   "source": [
    "for name, param in model.named_parameters():\n",
    "    print(f\"{name}: requires_grad={param.requires_grad}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "id": "1e381371-d212-48f7-803f-f861dd0d5b92",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "0eNmu2UJMg7v",
    "outputId": "26814188-ecd3-4f26-bf3f-a7578768454f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50 | Train Loss: 2.8716 | Train Acc: 0.1235 | Val Loss: 2.8403 | Val Acc: 0.1592\n",
      "Epoch 2/50 | Train Loss: 2.5237 | Train Acc: 0.1725 | Val Loss: 2.8007 | Val Acc: 0.1369\n",
      "Epoch 3/50 | Train Loss: 2.4099 | Train Acc: 0.1977 | Val Loss: 2.2771 | Val Acc: 0.1955\n",
      "Epoch 4/50 | Train Loss: 2.3032 | Train Acc: 0.2177 | Val Loss: 2.1955 | Val Acc: 0.2304\n",
      "Epoch 5/50 | Train Loss: 2.2836 | Train Acc: 0.2117 | Val Loss: 2.7105 | Val Acc: 0.0824\n",
      "Epoch 6/50 | Train Loss: 2.2280 | Train Acc: 0.2141 | Val Loss: 2.2042 | Val Acc: 0.2318\n",
      "Epoch 7/50 | Train Loss: 2.1926 | Train Acc: 0.2234 | Val Loss: 2.0605 | Val Acc: 0.2668\n",
      "Epoch 8/50 | Train Loss: 2.1570 | Train Acc: 0.2249 | Val Loss: 2.0972 | Val Acc: 0.2807\n",
      "Epoch 9/50 | Train Loss: 2.1031 | Train Acc: 0.2333 | Val Loss: 2.1421 | Val Acc: 0.2556\n",
      "Epoch 10/50 | Train Loss: 2.1107 | Train Acc: 0.2330 | Val Loss: 2.3214 | Val Acc: 0.1411\n",
      "Epoch 11/50 | Train Loss: 1.9752 | Train Acc: 0.2808 | Val Loss: 1.9702 | Val Acc: 0.3128\n",
      "Epoch 12/50 | Train Loss: 1.9201 | Train Acc: 0.2865 | Val Loss: 1.9640 | Val Acc: 0.3059\n",
      "Epoch 13/50 | Train Loss: 1.9073 | Train Acc: 0.2901 | Val Loss: 1.9492 | Val Acc: 0.3212\n",
      "Epoch 14/50 | Train Loss: 1.8980 | Train Acc: 0.2841 | Val Loss: 1.9432 | Val Acc: 0.3282\n",
      "Epoch 15/50 | Train Loss: 1.8734 | Train Acc: 0.2934 | Val Loss: 1.9613 | Val Acc: 0.3408\n",
      "Epoch 16/50 | Train Loss: 1.8520 | Train Acc: 0.2898 | Val Loss: 1.9604 | Val Acc: 0.3240\n",
      "Epoch 17/50 | Train Loss: 1.7816 | Train Acc: 0.3110 | Val Loss: 1.9272 | Val Acc: 0.3310\n",
      "Epoch 18/50 | Train Loss: 1.8159 | Train Acc: 0.2940 | Val Loss: 1.9714 | Val Acc: 0.3408\n",
      "Epoch 19/50 | Train Loss: 1.7876 | Train Acc: 0.3080 | Val Loss: 2.0035 | Val Acc: 0.3366\n",
      "Epoch 20/50 | Train Loss: 1.8344 | Train Acc: 0.2946 | Val Loss: 1.9377 | Val Acc: 0.3366\n",
      "Epoch 21/50 | Train Loss: 1.8006 | Train Acc: 0.3083 | Val Loss: 1.9603 | Val Acc: 0.3156\n",
      "Epoch 22/50 | Train Loss: 1.7771 | Train Acc: 0.3080 | Val Loss: 1.9427 | Val Acc: 0.3520\n",
      "Epoch 23/50 | Train Loss: 1.7799 | Train Acc: 0.3083 | Val Loss: 1.9573 | Val Acc: 0.3478\n",
      "Epoch 24/50 | Train Loss: 1.7916 | Train Acc: 0.2886 | Val Loss: 1.9487 | Val Acc: 0.3142\n",
      "Epoch 25/50 | Train Loss: 1.7895 | Train Acc: 0.3047 | Val Loss: 1.9376 | Val Acc: 0.3464\n",
      "Epoch 26/50 | Train Loss: 1.7330 | Train Acc: 0.3343 | Val Loss: 1.9408 | Val Acc: 0.3520\n",
      "Epoch 27/50 | Train Loss: 1.7351 | Train Acc: 0.3140 | Val Loss: 1.9490 | Val Acc: 0.3561\n",
      "Epoch 28/50 | Train Loss: 1.7435 | Train Acc: 0.3337 | Val Loss: 1.9405 | Val Acc: 0.3408\n",
      "Epoch 29/50 | Train Loss: 1.7579 | Train Acc: 0.3011 | Val Loss: 1.9641 | Val Acc: 0.3436\n",
      "Epoch 30/50 | Train Loss: 1.7144 | Train Acc: 0.3289 | Val Loss: 1.9535 | Val Acc: 0.3450\n",
      "Epoch 31/50 | Train Loss: 1.7389 | Train Acc: 0.3179 | Val Loss: 1.9775 | Val Acc: 0.3506\n",
      "Epoch 32/50 | Train Loss: 1.8144 | Train Acc: 0.2981 | Val Loss: 1.9662 | Val Acc: 0.3464\n",
      "Epoch 33/50 | Train Loss: 1.7342 | Train Acc: 0.3146 | Val Loss: 1.9496 | Val Acc: 0.3422\n",
      "Epoch 34/50 | Train Loss: 1.7077 | Train Acc: 0.3245 | Val Loss: 1.9466 | Val Acc: 0.3450\n",
      "Epoch 35/50 | Train Loss: 1.8092 | Train Acc: 0.2925 | Val Loss: 1.9469 | Val Acc: 0.3506\n",
      "Epoch 36/50 | Train Loss: 1.7220 | Train Acc: 0.3304 | Val Loss: 1.9676 | Val Acc: 0.3547\n",
      "Epoch 37/50 | Train Loss: 1.8287 | Train Acc: 0.2949 | Val Loss: 1.9806 | Val Acc: 0.3128\n",
      "Epoch 38/50 | Train Loss: 1.7765 | Train Acc: 0.3002 | Val Loss: 1.9574 | Val Acc: 0.3506\n",
      "Epoch 39/50 | Train Loss: 1.8090 | Train Acc: 0.2978 | Val Loss: 1.9546 | Val Acc: 0.3492\n",
      "Epoch 40/50 | Train Loss: 1.7346 | Train Acc: 0.3215 | Val Loss: 1.9931 | Val Acc: 0.3380\n",
      "Epoch 41/50 | Train Loss: 1.7805 | Train Acc: 0.2993 | Val Loss: 1.9849 | Val Acc: 0.3212\n",
      "Epoch 42/50 | Train Loss: 1.8200 | Train Acc: 0.2961 | Val Loss: 1.9623 | Val Acc: 0.3380\n",
      "Epoch 43/50 | Train Loss: 1.7586 | Train Acc: 0.3236 | Val Loss: 1.9690 | Val Acc: 0.3156\n",
      "Epoch 44/50 | Train Loss: 1.7689 | Train Acc: 0.2993 | Val Loss: 1.9824 | Val Acc: 0.3366\n",
      "Epoch 45/50 | Train Loss: 1.6725 | Train Acc: 0.3526 | Val Loss: 1.9636 | Val Acc: 0.3561\n",
      "Epoch 46/50 | Train Loss: 1.7983 | Train Acc: 0.3062 | Val Loss: 1.9702 | Val Acc: 0.3408\n",
      "Epoch 47/50 | Train Loss: 1.7781 | Train Acc: 0.2892 | Val Loss: 1.9498 | Val Acc: 0.3380\n",
      "Epoch 48/50 | Train Loss: 1.8162 | Train Acc: 0.3032 | Val Loss: 2.0048 | Val Acc: 0.3324\n",
      "Epoch 49/50 | Train Loss: 1.7238 | Train Acc: 0.3307 | Val Loss: 1.9999 | Val Acc: 0.3282\n",
      "Epoch 50/50 | Train Loss: 1.8303 | Train Acc: 0.2972 | Val Loss: 1.9592 | Val Acc: 0.3254\n"
     ]
    }
   ],
   "source": [
    "train_acc_history = []\n",
    "train_loss_history = []\n",
    "val_acc_history = []\n",
    "val_loss_history = []\n",
    "\n",
    "num_epochs = 50\n",
    "for epoch in range(num_epochs):\n",
    "    model.train()\n",
    "    train_loss, train_corrects = 0.0, 0.0\n",
    "\n",
    "    for images, labels in train_loader:\n",
    "        images, labels = images.to(device), labels.to(device)\n",
    "\n",
    "        labels = torch.squeeze(labels)\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        outputs = model(images)\n",
    "        loss = criterion(outputs, labels)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        train_loss += loss.item() * images.size(0)\n",
    "        _, preds = torch.max(outputs, 1)\n",
    "        train_corrects += torch.sum(preds == labels).item()\n",
    "\n",
    "    train_loss = train_loss / len(train_dataset)\n",
    "    train_acc = train_corrects / len(train_dataset)\n",
    "    train_acc_history.append(train_acc)\n",
    "    train_loss_history.append(train_loss)\n",
    "\n",
    "    model.eval()\n",
    "    val_loss, val_corrects = 0.0, 0.0\n",
    "\n",
    "    with torch.no_grad():\n",
    "        for images, labels in validation_loader:\n",
    "            images, labels = images.to(device), labels.to(device)\n",
    "\n",
    "            outputs = model(images)\n",
    "            loss = criterion(outputs, labels)\n",
    "\n",
    "            val_loss += loss.item() * images.size(0)\n",
    "            _, preds = torch.max(outputs, 1)\n",
    "            val_corrects += torch.sum(preds == labels).item()\n",
    "\n",
    "    val_loss = val_loss / len(validation_dataset)\n",
    "    val_acc = val_corrects / len(validation_dataset)\n",
    "\n",
    "    val_acc_history.append(val_acc)\n",
    "    val_loss_history.append(val_loss)\n",
    "\n",
    "    scheduler.step()\n",
    "    \n",
    "    print(f\"Epoch {epoch+1}/{num_epochs} | Train Loss: {train_loss:.4f} | Train Acc: {train_acc:.4f} | Val Loss: {val_loss:.4f} | Val Acc: {val_acc:.4f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "id": "447f644b-52b2-4838-891c-e11ae9168ee9",
   "metadata": {
    "editable": true,
    "slideshow": {
     "slide_type": ""
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot training and validation accuracy\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.plot(range(1, num_epochs+1), train_acc_history, label='Train')\n",
    "plt.plot(range(1, num_epochs+1), val_acc_history, label='Validation')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()\n",
    "plt.show()\n",
    "\n",
    "# Plot training and validation loss\n",
    "plt.plot(range(1, num_epochs+1), train_loss_history, label='Train')\n",
    "plt.plot(range(1, num_epochs+1), val_loss_history, label='Validation')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "id": "c554cd23-8f46-4ee0-ab4d-ccc389d3b33e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Loss: 1.9354 | Test Acc: 0.3134\n"
     ]
    }
   ],
   "source": [
    "model.eval()\n",
    "test_loss, test_corrects = 0.0, 0.0\n",
    "\n",
    "with torch.no_grad():\n",
    "    for images, labels in test_loader:\n",
    "        images, labels = images.to(device), labels.to(device)\n",
    "\n",
    "        outputs = model(images)\n",
    "        loss = criterion(outputs, labels)\n",
    "\n",
    "        test_loss += loss.item() * images.size(0)\n",
    "        _, preds = torch.max(outputs, 1)\n",
    "        test_corrects += torch.sum(preds == labels).item()\n",
    "\n",
    "test_loss = test_loss / len(test_dataset)\n",
    "test_acc = test_corrects / len(test_dataset)\n",
    "\n",
    "print(f\"Test Loss: {test_loss:.4f} | Test Acc: {test_acc:.4f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c9402844-e84b-418d-88e8-7e99b5b536f9",
   "metadata": {
    "editable": true,
    "slideshow": {
     "slide_type": ""
    },
    "tags": []
   },
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "id": "17bd5e88-e4fe-4606-9d11-01b730fcf42f",
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(model.state_dict(),\"model_torch_googlenet_2.pth\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.17"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}