model-with-mlp-1-head-aug-patch-extraction.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 132,
   "id": "d8b5acde",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:00.569974Z",
     "iopub.status.busy": "2024-10-27T05:51:00.569611Z",
     "iopub.status.idle": "2024-10-27T05:51:00.576994Z",
     "shell.execute_reply": "2024-10-27T05:51:00.576064Z",
     "shell.execute_reply.started": "2024-10-27T05:51:00.569940Z"
    },
    "id": "579e9c4f-bb81-401a-9280-d215be513669"
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import math\n",
    "import torch.utils.model_zoo as model_zoo\n",
    "import torchvision.transforms as transforms\n",
    "from torchvision.datasets import ImageFolder\n",
    "from torch.utils.data import DataLoader, Dataset\n",
    "import os\n",
    "import pandas as pd\n",
    "from PIL import Image\n",
    "import torch.optim as optim\n",
    "from torch.optim import AdamW\n",
    "from tqdm import tqdm\n",
    "import numpy as np \n",
    "from sklearn.metrics import accuracy_score, confusion_matrix, ConfusionMatrixDisplay\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import torch.nn.functional as F"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "id": "3820d51b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:01.060583Z",
     "iopub.status.busy": "2024-10-27T05:51:01.060225Z",
     "iopub.status.idle": "2024-10-27T05:51:01.065567Z",
     "shell.execute_reply": "2024-10-27T05:51:01.064250Z",
     "shell.execute_reply.started": "2024-10-27T05:51:01.060545Z"
    },
    "id": "18e6d001-c057-4580-a5ff-d1f20c1b1f1b"
   },
   "outputs": [],
   "source": [
    "model_urls = {\n",
    "    'resnet18': 'https://download.pytorch.org/models/resnet18-5c106cde.pth',\n",
    "    # 'resnet34': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth',\n",
    "    # 'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',\n",
    "    # 'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth',\n",
    "    # 'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth',\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "id": "efe78a43",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:01.422556Z",
     "iopub.status.busy": "2024-10-27T05:51:01.422235Z",
     "iopub.status.idle": "2024-10-27T05:51:01.427246Z",
     "shell.execute_reply": "2024-10-27T05:51:01.426374Z",
     "shell.execute_reply.started": "2024-10-27T05:51:01.422524Z"
    },
    "id": "489ff69f-8164-49fa-b930-ca293d6bf96a"
   },
   "outputs": [],
   "source": [
    "def conv3x3(in_planes, out_planes, stride=1):\n",
    "    \"3x3 convolution with padding\"\n",
    "    return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,\n",
    "                     padding=1, bias=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "id": "00e5fc60",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:01.736307Z",
     "iopub.status.busy": "2024-10-27T05:51:01.735987Z",
     "iopub.status.idle": "2024-10-27T05:51:01.743586Z",
     "shell.execute_reply": "2024-10-27T05:51:01.742695Z",
     "shell.execute_reply.started": "2024-10-27T05:51:01.736273Z"
    },
    "id": "ddcf8890-3b5f-4cf2-9556-2396fc691ed9"
   },
   "outputs": [],
   "source": [
    "class ChannelAttention(nn.Module):\n",
    "    def __init__(self, in_planes, ratio=16):\n",
    "        super(ChannelAttention, self).__init__()\n",
    "        self.avg_pool = nn.AdaptiveAvgPool2d(1) # 1×1×64\n",
    "        self.max_pool = nn.AdaptiveMaxPool2d(1) # 1×1×64\n",
    "\n",
    "        self.fc = nn.Sequential(nn.Conv2d(in_planes, in_planes // 16, 1, bias=False), # 1×1×4\n",
    "                               nn.ReLU(), # 1×1×4\n",
    "                               nn.Conv2d(in_planes // 16, in_planes, 1, bias=False)) # 1×1×64\n",
    "        self.sigmoid = nn.Sigmoid() # 1×1×64\n",
    "\n",
    "    def forward(self, x):\n",
    "        avg_out = self.fc(self.avg_pool(x)) # 1×1×64\n",
    "        max_out = self.fc(self.max_pool(x)) # 1×1×64\n",
    "        out = avg_out + max_out # 1×1×64\n",
    "        return self.sigmoid(out) # 1×1×64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "id": "1e920ac1",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:02.086191Z",
     "iopub.status.busy": "2024-10-27T05:51:02.085888Z",
     "iopub.status.idle": "2024-10-27T05:51:02.092974Z",
     "shell.execute_reply": "2024-10-27T05:51:02.091973Z",
     "shell.execute_reply.started": "2024-10-27T05:51:02.086160Z"
    },
    "id": "0260a2df-76c8-413f-a7bf-a1c3b6b97c1f"
   },
   "outputs": [],
   "source": [
    "class SpatialAttention(nn.Module):\n",
    "    def __init__(self, kernel_size=7):\n",
    "        super(SpatialAttention, self).__init__()\n",
    "\n",
    "        self.conv1 = nn.Conv2d(2, 1, kernel_size, padding=kernel_size//2, bias=False)\n",
    "        self.sigmoid = nn.Sigmoid()\n",
    "\n",
    "    def forward(self, x):\n",
    "        avg_out = torch.mean(x, dim=1, keepdim=True) # 56×56×1\n",
    "        max_out, _ = torch.max(x, dim=1, keepdim=True) # 56×56×1\n",
    "        x = torch.cat([avg_out, max_out], dim=1) # 56×56×2\n",
    "        x = self.conv1(x) # 56×56×1\n",
    "        return self.sigmoid(x) # 56×56×1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "id": "b0a6fa85",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:02.457762Z",
     "iopub.status.busy": "2024-10-27T05:51:02.457056Z",
     "iopub.status.idle": "2024-10-27T05:51:02.467592Z",
     "shell.execute_reply": "2024-10-27T05:51:02.466549Z",
     "shell.execute_reply.started": "2024-10-27T05:51:02.457713Z"
    },
    "id": "41fecb88-333e-4ba1-86e9-a86fd507c273"
   },
   "outputs": [],
   "source": [
    "class BasicBlock(nn.Module):\n",
    "    expansion = 1\n",
    "\n",
    "    def __init__(self, inplanes, planes, stride=1, downsample=None):\n",
    "        super(BasicBlock, self).__init__()\n",
    "        self.conv1 = conv3x3(inplanes, planes, stride) # 56*56*64 \n",
    "        self.bn1 = nn.BatchNorm2d(planes) # 56*56*64 \n",
    "        self.relu = nn.ReLU(inplace=True) # 56*56*64 \n",
    "        self.conv2 = conv3x3(planes, planes) # 56*56*64 \n",
    "        self.bn2 = nn.BatchNorm2d(planes)\n",
    "\n",
    "        self.ca = ChannelAttention(planes)\n",
    "        self.sa = SpatialAttention()\n",
    "\n",
    "        self.downsample = downsample\n",
    "        self.stride = stride\n",
    "\n",
    "    def forward(self, x):\n",
    "        residual = x # (56×56×64, 56×56×64), (56×56×64, 28x28x128), (28x28x128, 14x14x256)\n",
    "\n",
    "        out = self.conv1(x) # (56*56*64, 56×56×64), (28*28*128, 28*28*128), (14x14x256, 14x14x256)\n",
    "        out = self.bn1(out) # (56*56*64, 56×56×64), (28*28*128, 28*28*128), (14x14x256, 14x14x256)\n",
    "        out = self.relu(out) # (56*56*64, 56×56×64), (28*28*128, 28*28*128), (14x14x256, 14x14x256)\n",
    "\n",
    "        out = self.conv2(out) # (56*56*64, 56×56×64), (28*28*128, 28*28*128), (14x14x256, 14x14x256)\n",
    "        out = self.bn2(out) # (56*56*64, 56×56×64), (28*28*128, 28*28*128), (14x14x256, 14x14x256)\n",
    "\n",
    "        out = self.ca(out) * out # (1*1*64) * (56*56*64) = (56*56*64; 56×56×64); (28*28*128; 28*28*128); (14x14x256, 14x14x256);\n",
    "        out = self.sa(out) * out # (56×56×1) * (56×56×64) = (56×56×64; 56×56×64); (28x28x128; 28*28*128); (14x14x256, 14x14x256);\n",
    "\n",
    "        if self.downsample is not None:\n",
    "            residual = self.downsample(x) # 28*28*128, 14x14x256 \n",
    "\n",
    "        out += residual # (56×56×64, 56×56×64), (28*28*128, 28*28*128), (14x14x256, 14x14x256)\n",
    "        out = self.relu(out) # (56×56×64, 56×56×64), (28*28*128, 28*28*128), (14x14x256, 14x14x256)\n",
    "\n",
    "        return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "id": "4d586a30",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:02.900969Z",
     "iopub.status.busy": "2024-10-27T05:51:02.900215Z",
     "iopub.status.idle": "2024-10-27T05:51:02.909622Z",
     "shell.execute_reply": "2024-10-27T05:51:02.908770Z",
     "shell.execute_reply.started": "2024-10-27T05:51:02.900936Z"
    }
   },
   "outputs": [],
   "source": [
    "class PatchExtraction(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(PatchExtraction, self).__init__()\n",
    "\n",
    "        self.depthwise_conv1 = nn.Conv2d(256, 256, kernel_size=4, stride=4, padding=1, groups=256)  # Input: 14x14x256, Output: 4×4×256     \n",
    "        self.pointwise_conv1 = nn.Conv2d(256, 256, kernel_size=1, stride=1, padding=0)  # Output: 4×4×256\n",
    "        \n",
    "        self.depthwise_conv2 = nn.Conv2d(256, 256, kernel_size=2, stride=2, padding=0, groups=256)  # Output: 2×2×256\n",
    "        self.pointwise_conv2 = nn.Conv2d(256, 256, kernel_size=1, stride=1, padding=0)  # Output: 2×2×256\n",
    "\n",
    "        self.conv3 = nn.Conv2d(256, 256, kernel_size=1, stride=1, padding=0)  # Output: 2×2×256\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = F.relu(self.depthwise_conv1(x))\n",
    "        x = F.relu(self.pointwise_conv1(x))\n",
    "\n",
    "        x = F.relu(self.depthwise_conv2(x))\n",
    "        x = F.relu(self.pointwise_conv2(x))\n",
    "\n",
    "        x = F.relu(self.conv3(x))\n",
    "        \n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "id": "27eefd53",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:03.505986Z",
     "iopub.status.busy": "2024-10-27T05:51:03.505131Z",
     "iopub.status.idle": "2024-10-27T05:51:03.525409Z",
     "shell.execute_reply": "2024-10-27T05:51:03.524407Z",
     "shell.execute_reply.started": "2024-10-27T05:51:03.505946Z"
    },
    "id": "42b16778-f805-4570-8455-179026a3c972"
   },
   "outputs": [],
   "source": [
    "class ResNet(nn.Module):\n",
    "\n",
    "    def __init__(self, block, layers, num_heads, num_classes=7):\n",
    "        self.inplanes = 64\n",
    "        self.num_heads = num_heads\n",
    "        super(ResNet, self).__init__()\n",
    "        self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3,\n",
    "                               bias=False) # 112*112*64\n",
    "        self.bn1 = nn.BatchNorm2d(64) # 112*112*64\n",
    "        self.relu = nn.ReLU(inplace=True) # 112*112*64\n",
    "        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) # 56*56*64\n",
    "        self.layer1 = self._make_layer(block, 64, layers[0]) # 56×56×64\n",
    "        self.layer2 = self._make_layer(block, 128, layers[1], stride=2) # 28*28*128\n",
    "        self.layer3 = self._make_layer(block, 256, layers[2], stride=2) # 14*14*256 \n",
    "        self.patch_extraction = PatchExtraction() #  2×2×256\n",
    "\n",
    "        self.attention = nn.MultiheadAttention(256, num_heads)\n",
    "        # Input will be changed to the size 4x32x256 (see below forward code) which will be used as query key value \n",
    "        # weight matrix will have the size 256*(256) as default value (256) is default value we can change this by using vdim and kdim attributes of the class\n",
    "        # else weight matrix will have embed_dim * kdim for key matrix , after multiplying 4x32x256 will be the ouptut like this we will be having key query \n",
    "        # value matrix (Q)x(K.T)  query and key transpose multiplication will produce of size 4x32x4 and dividing it with sqrt(256) , this dividing saves the \n",
    "        # variance and makes it around 1 and also it gets trained without gradient saturation and finally it applies soft max on to this making each row value\n",
    "        #sum up to 1 of 4x4 matrix this for 32 batch size and multiplying it with 4x32x256 i.e 4x32x4 with 4x32x256 value matrix outputs the matrix 4x32x256\n",
    "        self.layer_norm = nn.LayerNorm(256)\n",
    "        self.global_avg_pool = nn.AdaptiveAvgPool2d(1) \n",
    "\n",
    "        self.classifier = nn.Linear(256, num_classes)\n",
    "\n",
    "        for m in self.classifier.modules():\n",
    "            if isinstance(m, nn.Linear):\n",
    "                fan_out = m.weight.size(0)  \n",
    "                fan_in = m.weight.size(1) if m.weight.dim() > 1 else 0  \n",
    "                init_range = 1.0 / math.sqrt(fan_in + fan_out)\n",
    "                m.weight.data.uniform_(-init_range, init_range)\n",
    "                if m.bias is not None:\n",
    "                    m.bias.data.zero_()\n",
    "\n",
    "    def _make_layer(self, block, planes, blocks, stride=1):\n",
    "        downsample = None\n",
    "        if stride != 1 or self.inplanes != planes * block.expansion:\n",
    "            downsample = nn.Sequential(\n",
    "                nn.Conv2d(self.inplanes, planes * block.expansion,\n",
    "                          kernel_size=1, stride=stride, bias=False), # 28*28*128, 14×14×256, \n",
    "                nn.BatchNorm2d(planes * block.expansion), # 28*28*128, 14×14×256\n",
    "            )\n",
    "\n",
    "        layers = []\n",
    "        layers.append(block(self.inplanes, planes, stride, downsample))\n",
    "        self.inplanes = planes * block.expansion\n",
    "        for i in range(1, blocks):\n",
    "            layers.append(block(self.inplanes, planes))\n",
    "\n",
    "        return nn.Sequential(*layers)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.conv1(x)\n",
    "        x = self.bn1(x)\n",
    "        x = self.relu(x)\n",
    "        x = self.maxpool(x)\n",
    "\n",
    "        x = self.layer1(x)\n",
    "        x = self.layer2(x)\n",
    "        x = self.layer3(x)\n",
    "\n",
    "        x = self.patch_extraction(x) \n",
    "\n",
    "        batch_size, channels, height, width = x.shape\n",
    "        x = x.view(batch_size, channels, height*width).permute(2,0,1)\n",
    "\n",
    "        query = x\n",
    "        key = x\n",
    "        value = x\n",
    "\n",
    "        attn_output, _ = self.attention(query, key, value)\n",
    "        x = self.layer_norm(attn_output + x)\n",
    "        x = x.permute(1,2,0).view(batch_size, channels, height, width) # 32*256*2*2\n",
    "        x = self.global_avg_pool(x).view(batch_size, -1) # 32*256\n",
    "\n",
    "        logits = self.classifier(x)\n",
    "\n",
    "        return logits"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "id": "d09ec97b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:34:56.687318Z",
     "iopub.status.busy": "2024-10-27T08:34:56.686517Z",
     "iopub.status.idle": "2024-10-27T08:34:56.693444Z",
     "shell.execute_reply": "2024-10-27T08:34:56.692553Z",
     "shell.execute_reply.started": "2024-10-27T08:34:56.687276Z"
    }
   },
   "outputs": [],
   "source": [
    "def resnet18_cbam(pretrained=True, **kwargs):\n",
    "    \"\"\"Constructs a ResNet-18 model.\n",
    "\n",
    "    Args:\n",
    "        pretrained (bool): If True, returns a model pre-trained on ImageNet\n",
    "    \"\"\"\n",
    "    num_heads = 1\n",
    "    model = ResNet(BasicBlock, [2, 2, 2, 2], num_heads)\n",
    "#     if pretrained:\n",
    "#         pretrained_state_dict = model_zoo.load_url(model_urls['resnet18'])\n",
    "#         now_state_dict        = model.state_dict()\n",
    "#         now_state_dict.update(pretrained_state_dict)\n",
    "#         model.load_state_dict(now_state_dict, strict=False)\n",
    "    \n",
    "    if pretrained:\n",
    "        checkpoint = torch.load(r\"/kaggle/input/qdawsfegdrhfjghk/aug_epoch_12_acc_82.20_newmodel.pth\")\n",
    "        model.load_state_dict(checkpoint['model_state_dict'],strict=False)\n",
    "\n",
    "    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "id": "6ecdcac2",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:34:58.017229Z",
     "iopub.status.busy": "2024-10-27T08:34:58.016577Z",
     "iopub.status.idle": "2024-10-27T08:34:58.719346Z",
     "shell.execute_reply": "2024-10-27T08:34:58.718015Z",
     "shell.execute_reply.started": "2024-10-27T08:34:58.017190Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_30/3634161233.py:17: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.\n",
      "  checkpoint = torch.load(r\"/kaggle/input/qdawsfegdrhfjghk/aug_epoch_12_acc_82.20_newmodel.pth\")\n"
     ]
    }
   ],
   "source": [
    "model = resnet18_cbam()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "id": "e5efb0f7",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:35:02.860040Z",
     "iopub.status.busy": "2024-10-27T08:35:02.859301Z",
     "iopub.status.idle": "2024-10-27T08:35:02.866377Z",
     "shell.execute_reply": "2024-10-27T08:35:02.865273Z",
     "shell.execute_reply.started": "2024-10-27T08:35:02.859998Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device(type='cuda')"
      ]
     },
     "execution_count": 292,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "id": "f8a0158b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:35:03.079002Z",
     "iopub.status.busy": "2024-10-27T08:35:03.078669Z",
     "iopub.status.idle": "2024-10-27T08:35:03.093381Z",
     "shell.execute_reply": "2024-10-27T08:35:03.091715Z",
     "shell.execute_reply.started": "2024-10-27T08:35:03.078969Z"
    },
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "model = model.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 294,
   "id": "64b143e8",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:35:06.297608Z",
     "iopub.status.busy": "2024-10-27T08:35:06.297226Z",
     "iopub.status.idle": "2024-10-27T08:35:06.302558Z",
     "shell.execute_reply": "2024-10-27T08:35:06.301587Z",
     "shell.execute_reply.started": "2024-10-27T08:35:06.297570Z"
    },
    "id": "03cd3076-fd58-4de5-a993-fac853812bce"
   },
   "outputs": [],
   "source": [
    "transform = transforms.Compose([\n",
    "    transforms.Resize((224, 224)),\n",
    "    transforms.ToTensor(),\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 295,
   "id": "6ccab0f6",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:35:06.687934Z",
     "iopub.status.busy": "2024-10-27T08:35:06.687232Z",
     "iopub.status.idle": "2024-10-27T08:35:06.693388Z",
     "shell.execute_reply": "2024-10-27T08:35:06.692358Z",
     "shell.execute_reply.started": "2024-10-27T08:35:06.687894Z"
    }
   },
   "outputs": [],
   "source": [
    "aug_transform = transforms.Compose([\n",
    "    transforms.Resize((224, 224)),  # Resize the image to 224x224\n",
    "    transforms.RandomHorizontalFlip(),  # Randomly flip the image horizontally\n",
    "    transforms.RandomRotation(10),  # Randomly rotate the image by up to 5 degrees\n",
    "    transforms.ColorJitter(brightness=0.2, contrast=0.2, saturation=0.2, hue=0.1),  # Slightly change brightness and contrast\n",
    "    transforms.ToTensor(),  # Convert the image to a PyTorch tensor\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c86a09c",
   "metadata": {
    "id": "a2eb0ecd-292c-41ab-b5ee-d250558f39d9"
   },
   "source": [
    "# Class Labels are 0-Indexed CAREFULL"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "id": "c08cc324",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:13.052087Z",
     "iopub.status.busy": "2024-10-27T05:51:13.051799Z",
     "iopub.status.idle": "2024-10-27T05:51:13.060228Z",
     "shell.execute_reply": "2024-10-27T05:51:13.059406Z",
     "shell.execute_reply.started": "2024-10-27T05:51:13.052055Z"
    },
    "id": "b7badb95-b47f-42ab-85dc-d1e02aa63d0b"
   },
   "outputs": [],
   "source": [
    "class CustomImageDataset(Dataset):\n",
    "    def __init__(self, image_dir, csv_file, transform=None):\n",
    "        self.image_dir = image_dir\n",
    "        self.transform = transform\n",
    "\n",
    "        self.data_frame = pd.read_csv(csv_file)\n",
    "\n",
    "        assert 'image' in self.data_frame.columns\n",
    "        assert 'label' in self.data_frame.columns\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.data_frame)\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        img_name = self.data_frame.iloc[idx, 0]  \n",
    "        class_label = self.data_frame.iloc[idx, 1] - 1  \n",
    "\n",
    "        class_folder = os.path.join(self.image_dir, str(class_label+1))  \n",
    "        img_path = os.path.join(class_folder, img_name)\n",
    "\n",
    "        image = Image.open(img_path).convert('RGB')\n",
    "\n",
    "        if self.transform:\n",
    "            image = self.transform(image)\n",
    "\n",
    "        return image, class_label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "id": "96b29e69",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:13.839968Z",
     "iopub.status.busy": "2024-10-27T05:51:13.839215Z",
     "iopub.status.idle": "2024-10-27T05:51:13.844305Z",
     "shell.execute_reply": "2024-10-27T05:51:13.843304Z",
     "shell.execute_reply.started": "2024-10-27T05:51:13.839928Z"
    },
    "id": "aa27ec16-927e-4e09-b749-8a6d18011972"
   },
   "outputs": [],
   "source": [
    "image_directory = r\"/kaggle/input/raf-db-dataset/DATASET/train\"  \n",
    "csv_file_path = r\"/kaggle/input/raf-db-dataset/train_labels.csv\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "id": "913926b7",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:14.083946Z",
     "iopub.status.busy": "2024-10-27T05:51:14.083624Z",
     "iopub.status.idle": "2024-10-27T05:51:14.107421Z",
     "shell.execute_reply": "2024-10-27T05:51:14.106413Z",
     "shell.execute_reply.started": "2024-10-27T05:51:14.083913Z"
    },
    "id": "a4ebf758-cb70-4921-aac6-3fc093533fc4"
   },
   "outputs": [],
   "source": [
    "train_dataset = CustomImageDataset(image_dir=image_directory, csv_file=csv_file_path, transform=transform)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "id": "a968594f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:14.311729Z",
     "iopub.status.busy": "2024-10-27T05:51:14.311402Z",
     "iopub.status.idle": "2024-10-27T05:51:14.327427Z",
     "shell.execute_reply": "2024-10-27T05:51:14.326669Z",
     "shell.execute_reply.started": "2024-10-27T05:51:14.311694Z"
    }
   },
   "outputs": [],
   "source": [
    "aug_train_dataset = CustomImageDataset(image_directory, csv_file_path, aug_transform)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "id": "824f4f5d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:14.537851Z",
     "iopub.status.busy": "2024-10-27T05:51:14.537532Z",
     "iopub.status.idle": "2024-10-27T05:51:14.543088Z",
     "shell.execute_reply": "2024-10-27T05:51:14.542215Z",
     "shell.execute_reply.started": "2024-10-27T05:51:14.537818Z"
    }
   },
   "outputs": [],
   "source": [
    "train_loader = DataLoader(train_dataset, batch_size=32, num_workers=4, pin_memory=True, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "id": "035eb1d2",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:15.360306Z",
     "iopub.status.busy": "2024-10-27T05:51:15.359963Z",
     "iopub.status.idle": "2024-10-27T05:51:15.364743Z",
     "shell.execute_reply": "2024-10-27T05:51:15.363880Z",
     "shell.execute_reply.started": "2024-10-27T05:51:15.360270Z"
    }
   },
   "outputs": [],
   "source": [
    "aug_train_loader = DataLoader(aug_train_dataset, batch_size=32, num_workers=4, pin_memory=True, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "id": "7fbcf0c4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:15.596907Z",
     "iopub.status.busy": "2024-10-27T05:51:15.596586Z",
     "iopub.status.idle": "2024-10-27T05:51:15.617450Z",
     "shell.execute_reply": "2024-10-27T05:51:15.616523Z",
     "shell.execute_reply.started": "2024-10-27T05:51:15.596873Z"
    },
    "id": "f74qGWRn4mYD"
   },
   "outputs": [],
   "source": [
    "torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "id": "aafce9d5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:16.023477Z",
     "iopub.status.busy": "2024-10-27T05:51:16.023182Z",
     "iopub.status.idle": "2024-10-27T05:51:16.028979Z",
     "shell.execute_reply": "2024-10-27T05:51:16.028012Z",
     "shell.execute_reply.started": "2024-10-27T05:51:16.023441Z"
    },
    "id": "e6c77aa4-7e13-483f-938a-64df57fea96f"
   },
   "outputs": [],
   "source": [
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = optim.AdamW(model.parameters(), lr=0.001, weight_decay=0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "id": "17ac4e6c",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:16.240805Z",
     "iopub.status.busy": "2024-10-27T05:51:16.240241Z",
     "iopub.status.idle": "2024-10-27T05:51:16.244986Z",
     "shell.execute_reply": "2024-10-27T05:51:16.244184Z",
     "shell.execute_reply.started": "2024-10-27T05:51:16.240758Z"
    },
    "id": "6BiC9Ry0nqd-"
   },
   "outputs": [],
   "source": [
    "os.makedirs('models', exist_ok=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "id": "bdbab063",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 55
    },
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:16.469988Z",
     "iopub.status.busy": "2024-10-27T05:51:16.469275Z",
     "iopub.status.idle": "2024-10-27T05:51:16.479674Z",
     "shell.execute_reply": "2024-10-27T05:51:16.478510Z",
     "shell.execute_reply.started": "2024-10-27T05:51:16.469935Z"
    },
    "id": "TOE5GMazqsMl",
    "outputId": "c30bc8a6-b2f4-42ae-9b4c-73bda762e30a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/kaggle/working'"
      ]
     },
     "execution_count": 160,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "os.getcwd()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "id": "2d59d8b4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:16.681009Z",
     "iopub.status.busy": "2024-10-27T05:51:16.680632Z",
     "iopub.status.idle": "2024-10-27T05:51:16.686669Z",
     "shell.execute_reply": "2024-10-27T05:51:16.685939Z",
     "shell.execute_reply.started": "2024-10-27T05:51:16.680973Z"
    }
   },
   "outputs": [],
   "source": [
    "test_image_directory = r\"/kaggle/input/raf-db-dataset/DATASET/test\"\n",
    "test_csv_file_path = r\"/kaggle/input/raf-db-dataset/test_labels.csv\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "id": "59163cc5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:17.547383Z",
     "iopub.status.busy": "2024-10-27T05:51:17.546676Z",
     "iopub.status.idle": "2024-10-27T05:51:17.557433Z",
     "shell.execute_reply": "2024-10-27T05:51:17.556322Z",
     "shell.execute_reply.started": "2024-10-27T05:51:17.547343Z"
    }
   },
   "outputs": [],
   "source": [
    "test_dataset = CustomImageDataset(test_image_directory, test_csv_file_path, transform)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "id": "c865f0d6",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:17.808044Z",
     "iopub.status.busy": "2024-10-27T05:51:17.807698Z",
     "iopub.status.idle": "2024-10-27T05:51:17.812527Z",
     "shell.execute_reply": "2024-10-27T05:51:17.811666Z",
     "shell.execute_reply.started": "2024-10-27T05:51:17.808011Z"
    }
   },
   "outputs": [],
   "source": [
    "test_loader = DataLoader(test_dataset, batch_size=32, num_workers=4, pin_memory=True, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "id": "815efa56",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:18.541843Z",
     "iopub.status.busy": "2024-10-27T05:51:18.541430Z",
     "iopub.status.idle": "2024-10-27T05:51:18.553907Z",
     "shell.execute_reply": "2024-10-27T05:51:18.552969Z",
     "shell.execute_reply.started": "2024-10-27T05:51:18.541803Z"
    },
    "id": "29a08383-0aa7-4122-86ae-3bcf8da03420"
   },
   "outputs": [],
   "source": [
    "def train(model, train_loader, test_loader, criterion, optimizer, device, num_epochs=20):\n",
    "    best_accuracy = 75\n",
    "    model.train()\n",
    "    for epoch in range(1, num_epochs+1):\n",
    "        running_loss = 0.0\n",
    "\n",
    "        for images, labels in tqdm(train_loader):\n",
    "            images, labels = images.to(device), labels.to(device)\n",
    "\n",
    "            optimizer.zero_grad()\n",
    "\n",
    "            outputs=model(images)\n",
    "            loss = criterion(outputs,labels)\n",
    "\n",
    "            loss.backward()\n",
    "            optimizer.step()\n",
    "\n",
    "            running_loss += loss.item()\n",
    "        epoch_loss = running_loss / len(train_loader)\n",
    "        print(f'Epoch [{epoch}/{num_epochs}], Loss: {epoch_loss:.4f}')\n",
    "        if epoch > 1 :\n",
    "            model.eval()\n",
    "            correct = 0\n",
    "            total = 0\n",
    "            with torch.no_grad():\n",
    "                for images, labels in tqdm(test_loader):\n",
    "                    images = images.to(device)\n",
    "                    labels = labels.to(device)\n",
    "                    \n",
    "                    logits = model(images)\n",
    "                    \n",
    "                    _, predicted = torch.max(logits, 1)\n",
    "                    total += labels.size(0)  \n",
    "                    correct += (predicted == labels).sum().item()\n",
    "            accuracy = 100*(correct/total)\n",
    "            print(f\"test_Acc : {accuracy}\")\n",
    "                \n",
    "            if accuracy > best_accuracy or epoch%20==0:\n",
    "                best_accuracy = accuracy\n",
    "                torch.save({'iter': epoch,\n",
    "                        'model_state_dict': model.state_dict(),\n",
    "                        'optimizer_state_dict': optimizer.state_dict(),},\n",
    "                        os.path.join('models', f\"aug_epoch_{epoch}_acc_{accuracy:.2f}_newmodel.pth\"))\n",
    "                print('Model saved.')\n",
    "            model.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "31ed7bab",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "execution": {
     "iopub.execute_input": "2024-10-27T05:02:29.106272Z",
     "iopub.status.busy": "2024-10-27T05:02:29.105879Z",
     "iopub.status.idle": "2024-10-27T05:16:14.963501Z",
     "shell.execute_reply": "2024-10-27T05:16:14.962452Z",
     "shell.execute_reply.started": "2024-10-27T05:02:29.106232Z"
    },
    "id": "58970af0-2447-4401-aeee-de55d874ca94",
    "outputId": "b138da59-58d3-410c-a550-7dfb98216bd1"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.14it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/20], Loss: 1.3731\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.31it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [2/20], Loss: 0.9905\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:05<00:00, 18.99it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 68.48109517601043\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [3/20], Loss: 0.7745\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.95it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 70.73011734028684\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.28it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [4/20], Loss: 0.6318\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.03it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 73.82659713168188\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [5/20], Loss: 0.5022\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.45it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 75.91264667535854\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [6/20], Loss: 0.3788\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.32it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 76.17340286831812\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [7/20], Loss: 0.2866\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.42it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 76.49934810951761\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.28it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [8/20], Loss: 0.2066\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.18it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 76.76010430247719\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [9/20], Loss: 0.1458\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.49it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 75.19556714471969\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.27it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [10/20], Loss: 0.1244\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.39it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 75.13037809647979\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.27it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [11/20], Loss: 0.0999\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.35it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 76.53194263363756\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [12/20], Loss: 0.0575\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 22.59it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 76.36897001303781\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.28it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [13/20], Loss: 0.0656\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.31it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 77.15123859191657\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [14/20], Loss: 0.0921\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.05it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 77.73794002607562\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.27it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [15/20], Loss: 0.0495\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.33it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 78.16166883963494\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [16/20], Loss: 0.0695\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.93it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 75.65189048239895\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.30it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [17/20], Loss: 0.0451\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.85it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 77.11864406779661\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.30it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [18/20], Loss: 0.0422\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.79it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 75.78226857887876\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [19/20], Loss: 0.0545\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 22.00it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 76.95567144719688\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:37<00:00, 10.30it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [20/20], Loss: 0.0529\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.82it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 77.54237288135593\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "train(model, train_loader, test_loader, criterion, optimizer, device, num_epochs=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "id": "abc15593",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T05:51:24.845573Z",
     "iopub.status.busy": "2024-10-27T05:51:24.845197Z",
     "iopub.status.idle": "2024-10-27T06:24:28.463593Z",
     "shell.execute_reply": "2024-10-27T06:24:28.462456Z",
     "shell.execute_reply.started": "2024-10-27T05:51:24.845535Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:47<00:00,  8.11it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/40], Loss: 0.6193\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.42it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [2/40], Loss: 0.5053\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 19.97it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.5632333767927\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.44it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [3/40], Loss: 0.4596\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 22.85it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.24771838331161\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.45it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [4/40], Loss: 0.4096\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.39it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.85658409387223\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.46it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [5/40], Loss: 0.3955\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 22.03it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.40026075619296\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.44it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [6/40], Loss: 0.3616\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.13it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 78.0638852672751\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.36it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [7/40], Loss: 0.3418\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.87it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.23728813559322\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.45it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [8/40], Loss: 0.3234\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.13950456323337\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:46<00:00,  8.34it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [9/40], Loss: 0.2921\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.45it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.50847457627118\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.46it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [10/40], Loss: 0.2692\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.08it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.79139504563233\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:46<00:00,  8.32it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [11/40], Loss: 0.2696\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.29it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.82398956975229\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.45it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [12/40], Loss: 0.2412\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.98it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 82.20338983050848\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.37it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [13/40], Loss: 0.2327\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.04it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.4328552803129\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.47it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [14/40], Loss: 0.2184\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.98it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.16036505867015\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:46<00:00,  8.31it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [15/40], Loss: 0.2104\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.28it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.25814863103\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.46it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [16/40], Loss: 0.1949\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 21.90it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.87744458930901\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.47it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [17/40], Loss: 0.1816\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.77it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.33507170795306\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.39it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [18/40], Loss: 0.1769\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.87it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.87744458930901\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.46it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [19/40], Loss: 0.1810\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.91it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.25814863103\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:46<00:00,  8.35it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [20/40], Loss: 0.1646\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.19it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.82398956975229\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.41it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [21/40], Loss: 0.1539\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.33it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.14993481095176\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:46<00:00,  8.34it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [22/40], Loss: 0.1509\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 24.00it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.80182529335072\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.46it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [23/40], Loss: 0.1530\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.04it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.46544980443285\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.36it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [24/40], Loss: 0.1493\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.75it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 78.78096479791395\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.52it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [25/40], Loss: 0.1278\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.21it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.01955671447197\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.42it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [26/40], Loss: 0.1265\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.42it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.76923076923077\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.50it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [27/40], Loss: 0.1238\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.14it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.34550195567145\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.42it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [28/40], Loss: 0.1267\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.86it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.51890482398957\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.49it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [29/40], Loss: 0.1182\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 22.45it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.14993481095176\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.50it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [30/40], Loss: 0.1071\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.39it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.80182529335072\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.48it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [31/40], Loss: 0.1120\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.31it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.97522816166884\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.46it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [32/40], Loss: 0.1087\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.49it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.58409387222947\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.38it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [33/40], Loss: 0.1094\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.53it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.77966101694916\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.49it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [34/40], Loss: 0.1145\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.37it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.58409387222947\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:46<00:00,  8.32it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [35/40], Loss: 0.1000\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.71it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 82.13820078226858\n",
      "Model saved.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.44it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [36/40], Loss: 0.0966\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.62it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.47588005215124\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.40it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [37/40], Loss: 0.0955\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.42it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 80.60625814863103\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.48it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [38/40], Loss: 0.1049\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.27it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.49804432855281\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.41it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [39/40], Loss: 0.0929\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:03<00:00, 24.33it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 81.45371577574967\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 384/384 [00:45<00:00,  8.46it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [40/40], Loss: 0.0984\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 23.74it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_Acc : 79.79139504563233\n",
      "Model saved.\n"
     ]
    }
   ],
   "source": [
    "train(model, aug_train_loader, test_loader, criterion, optimizer, device, num_epochs=40)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 296,
   "id": "379231eb",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:35:25.743105Z",
     "iopub.status.busy": "2024-10-27T08:35:25.742673Z",
     "iopub.status.idle": "2024-10-27T08:35:31.244657Z",
     "shell.execute_reply": "2024-10-27T08:35:31.243351Z",
     "shell.execute_reply.started": "2024-10-27T08:35:25.743064Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:05<00:00, 17.50it/s]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "82.20338983050848"
      ]
     },
     "execution_count": 296,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.eval()\n",
    "correct = 0\n",
    "total = 0\n",
    "with torch.no_grad():\n",
    "    for images, labels in tqdm(test_loader):\n",
    "        images = images.to(device)\n",
    "        labels = labels.to(device)\n",
    "        \n",
    "        logits = model(images)\n",
    "        \n",
    "        _, predicted = torch.max(logits, 1)\n",
    "        total += labels.size(0) \n",
    "        correct += (predicted == labels).sum().item()\n",
    "accuracy = 100*(correct/total)\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "accuracy\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "id": "2ed8d849",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:36:02.001317Z",
     "iopub.status.busy": "2024-10-27T08:36:02.000397Z",
     "iopub.status.idle": "2024-10-27T08:36:06.303899Z",
     "shell.execute_reply": "2024-10-27T08:36:06.302695Z",
     "shell.execute_reply.started": "2024-10-27T08:36:02.001253Z"
    },
    "id": "3f08e78f-164c-44f4-ab33-76d3355754af"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 96/96 [00:04<00:00, 22.38it/s]\n"
     ]
    }
   ],
   "source": [
    "all_preds=[]\n",
    "all_labels=[]\n",
    "with torch.no_grad():\n",
    "    for images, labels in tqdm(test_loader):\n",
    "        images = images.to(device)\n",
    "        labels = labels.to(device)\n",
    "        outputs = model(images)\n",
    "        _, predicted = torch.max(outputs,1)\n",
    "#         predicted+=1\n",
    "        \n",
    "        all_preds.extend(predicted.cpu().numpy())\n",
    "        all_labels.extend(labels.cpu().numpy())\n",
    "\n",
    "all_preds = np.array(all_preds)\n",
    "all_labels = np.array(all_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "id": "4446c79b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:36:07.415844Z",
     "iopub.status.busy": "2024-10-27T08:36:07.415433Z",
     "iopub.status.idle": "2024-10-27T08:36:07.433482Z",
     "shell.execute_reply": "2024-10-27T08:36:07.432595Z",
     "shell.execute_reply.started": "2024-10-27T08:36:07.415802Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8220338983050848"
      ]
     },
     "execution_count": 299,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy = accuracy_score(all_labels, all_preds)\n",
    "accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "id": "2ae8a128",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:36:10.563002Z",
     "iopub.status.busy": "2024-10-27T08:36:10.562342Z",
     "iopub.status.idle": "2024-10-27T08:36:10.572364Z",
     "shell.execute_reply": "2024-10-27T08:36:10.571158Z",
     "shell.execute_reply.started": "2024-10-27T08:36:10.562959Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 258,    8,    4,   12,   14,    5,   28],\n",
       "       [  13,   39,    2,    3,   10,    3,    4],\n",
       "       [   1,    2,   76,   12,   24,   12,   33],\n",
       "       [   8,    0,   14, 1093,   15,    3,   52],\n",
       "       [   2,    2,   18,   22,  364,    3,   67],\n",
       "       [   5,   10,   13,   14,   10,  102,    8],\n",
       "       [  10,    1,   18,   29,   30,    2,  590]])"
      ]
     },
     "execution_count": 300,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cm = confusion_matrix(all_labels, all_preds)\n",
    "cm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 301,
   "id": "086747fd",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-10-27T08:36:12.395566Z",
     "iopub.status.busy": "2024-10-27T08:36:12.395156Z",
     "iopub.status.idle": "2024-10-27T08:36:12.829087Z",
     "shell.execute_reply": "2024-10-27T08:36:12.828191Z",
     "shell.execute_reply.started": "2024-10-27T08:36:12.395522Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 7))\n",
    "sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', xticklabels=[f'Class {i}' for i in range(6)], yticklabels=[f'Class {i}' for i in range(6)])\n",
    "plt.ylabel('True Label')\n",
    "plt.xlabel('Predicted Label')\n",
    "plt.title('Confusion Matrix')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "93edcc25",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "gpuType": "T4",
   "provenance": []
  },
  "kaggle": {
   "accelerator": "nvidiaTeslaT4",
   "dataSources": [
    {
     "datasetId": 3758654,
     "sourceId": 6504606,
     "sourceType": "datasetVersion"
    },
    {
     "datasetId": 5956378,
     "sourceId": 9732886,
     "sourceType": "datasetVersion"
    },
    {
     "datasetId": 5956988,
     "sourceId": 9733710,
     "sourceType": "datasetVersion"
    }
   ],
   "dockerImageVersionId": 30787,
   "isGpuEnabled": true,
   "isInternetEnabled": true,
   "language": "python",
   "sourceType": "notebook"
  },
  "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": "3.11.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}