hw5_label_memorization.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "id": "a7Pyt9_jH9hT"
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import time\n",
    "\n",
    "# Make sure to use only 10% of the available MNIST data.\n",
    "# Otherwise, experiment will take quite long (around 90 minutes).\n",
    "\n",
    "# (Modified version of AlexNet)\n",
    "class AlexNet(nn.Module):\n",
    "    def __init__(self, num_class=10):\n",
    "        super(AlexNet, self).__init__()\n",
    "\n",
    "        self.conv_layer1 = nn.Sequential(\n",
    "            nn.Conv2d(1, 96, kernel_size=4),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.Conv2d(96, 96, kernel_size=3),\n",
    "            nn.ReLU(inplace=True)\n",
    "        )\n",
    "        self.conv_layer2 = nn.Sequential(\n",
    "            nn.Conv2d(96, 256, kernel_size=5, padding=2),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.MaxPool2d(kernel_size=3, stride=2)\n",
    "        )\n",
    "        self.conv_layer3 = nn.Sequential(\n",
    "            nn.Conv2d(256, 384, kernel_size=3, padding=1),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.Conv2d(384, 384, kernel_size=3, padding=1),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.Conv2d(384, 256, kernel_size=3, padding=1),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.MaxPool2d(kernel_size=3, stride=2)\n",
    "        )\n",
    "\n",
    "        self.fc_layer1 = nn.Sequential(\n",
    "            nn.Dropout(),\n",
    "            nn.Linear(6400, 800),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.Linear(800, 10)\n",
    "        )\n",
    "\n",
    "    def forward(self, x):\n",
    "        output = self.conv_layer1(x)\n",
    "        output = self.conv_layer2(output)\n",
    "        output = self.conv_layer3(output)\n",
    "        output = torch.flatten(output, 1)\n",
    "        output = self.fc_layer1(output)\n",
    "        return output\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "id": "HkMMUOOXH-HY"
   },
   "outputs": [],
   "source": [
    "from torch.utils.data import DataLoader, random_split\n",
    "from torchvision import datasets\n",
    "from torchvision.transforms import transforms\n",
    "import random\n",
    "\n",
    "# MNIST dataset\n",
    "train_dataset = datasets.MNIST(root='./mnist_data/',\n",
    "                               train=True, \n",
    "                               transform=transforms.ToTensor(),\n",
    "                  \n",
    "                               download=True)\n",
    "\n",
    "test_dataset = datasets.MNIST(root='./mnist_data/',\n",
    "                              train=False, \n",
    "                              transform=transforms.ToTensor())\n",
    "\n",
    "dataset_size = len(train_dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "id": "jy91ybLTIHTU"
   },
   "outputs": [],
   "source": [
    "from torch.utils.data.sampler import SubsetRandomSampler\n",
    "# Use 10% of Train_data to reduce the computation time of this exp.\n",
    "\n",
    "train_dataset.targets = [random.randint(0,9) for idx in range(dataset_size)]\n",
    "\n",
    "train_size = int(dataset_size * 0.1)\n",
    "validation_size = int(dataset_size * 0.9)\n",
    "\n",
    "train_dataset, validation_dataset = random_split(train_dataset, [train_size, validation_size])\n",
    "\n",
    "\n",
    "#random_indices = random.sample(range(dataset_size), int(dataset_size * 0.1))\n",
    "#sample = SubsetRandomSampler(random_indices)\n",
    "\n",
    "\n",
    "# Make data_loader\n",
    "\n",
    "train_loader = torch.utils.data.DataLoader(dataset=train_dataset, batch_size=64, shuffle=True)\n",
    "\n",
    "test_loader = torch.utils.data.DataLoader(dataset=test_dataset, batch_size=64, shuffle=False)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "kmwUHokeIhk9",
    "outputId": "cda27283-ff5d-4937-f0ce-cb692dea3a98"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True\n"
     ]
    }
   ],
   "source": [
    "print(torch.cuda.is_available())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "KsgkbbGuIOxZ",
    "outputId": "d1040034-6264-4395-8f38-b64fb54a5275"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 2 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 3 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 4 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 5 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 6 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 7 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 8 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 9 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 10 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 11 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 12 : Train Loss 2.30, Accuracy 0.09%\n",
      "Epoch 13 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 14 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 15 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 16 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 17 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 18 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 19 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 20 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 21 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 22 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 23 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 24 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 25 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 26 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 27 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 28 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 29 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 30 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 31 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 32 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 33 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 34 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 35 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 36 : Train Loss 2.30, Accuracy 0.11%\n",
      "Epoch 37 : Train Loss 2.30, Accuracy 0.11%\n",
      "Epoch 38 : Train Loss 2.30, Accuracy 0.10%\n",
      "Epoch 39 : Train Loss 2.30, Accuracy 0.11%\n",
      "Epoch 40 : Train Loss 2.30, Accuracy 0.11%\n",
      "Epoch 41 : Train Loss 2.30, Accuracy 0.11%\n",
      "Epoch 42 : Train Loss 2.30, Accuracy 0.11%\n",
      "Epoch 43 : Train Loss 2.30, Accuracy 0.11%\n",
      "Epoch 44 : Train Loss 2.30, Accuracy 0.12%\n",
      "Epoch 45 : Train Loss 2.30, Accuracy 0.11%\n",
      "Epoch 46 : Train Loss 2.30, Accuracy 0.11%\n",
      "Epoch 47 : Train Loss 2.30, Accuracy 0.12%\n",
      "Epoch 48 : Train Loss 2.30, Accuracy 0.12%\n",
      "Epoch 49 : Train Loss 2.30, Accuracy 0.12%\n",
      "Epoch 50 : Train Loss 2.30, Accuracy 0.12%\n",
      "Epoch 51 : Train Loss 2.30, Accuracy 0.12%\n",
      "Epoch 52 : Train Loss 2.30, Accuracy 0.12%\n",
      "Epoch 53 : Train Loss 2.30, Accuracy 0.12%\n",
      "Epoch 54 : Train Loss 2.29, Accuracy 0.12%\n",
      "Epoch 55 : Train Loss 2.29, Accuracy 0.12%\n",
      "Epoch 56 : Train Loss 2.29, Accuracy 0.12%\n",
      "Epoch 57 : Train Loss 2.29, Accuracy 0.13%\n",
      "Epoch 58 : Train Loss 2.29, Accuracy 0.13%\n",
      "Epoch 59 : Train Loss 2.29, Accuracy 0.13%\n",
      "Epoch 60 : Train Loss 2.29, Accuracy 0.14%\n",
      "Epoch 61 : Train Loss 2.29, Accuracy 0.12%\n",
      "Epoch 62 : Train Loss 2.28, Accuracy 0.14%\n",
      "Epoch 63 : Train Loss 2.28, Accuracy 0.13%\n",
      "Epoch 64 : Train Loss 2.28, Accuracy 0.14%\n",
      "Epoch 65 : Train Loss 2.28, Accuracy 0.13%\n",
      "Epoch 66 : Train Loss 2.27, Accuracy 0.15%\n",
      "Epoch 67 : Train Loss 2.27, Accuracy 0.14%\n",
      "Epoch 68 : Train Loss 2.27, Accuracy 0.14%\n",
      "Epoch 69 : Train Loss 2.26, Accuracy 0.15%\n",
      "Epoch 70 : Train Loss 2.26, Accuracy 0.15%\n",
      "Epoch 71 : Train Loss 2.25, Accuracy 0.15%\n",
      "Epoch 72 : Train Loss 2.25, Accuracy 0.15%\n",
      "Epoch 73 : Train Loss 2.24, Accuracy 0.17%\n",
      "Epoch 74 : Train Loss 2.24, Accuracy 0.16%\n",
      "Epoch 75 : Train Loss 2.22, Accuracy 0.18%\n",
      "Epoch 76 : Train Loss 2.21, Accuracy 0.18%\n",
      "Epoch 77 : Train Loss 2.19, Accuracy 0.19%\n",
      "Epoch 78 : Train Loss 2.18, Accuracy 0.19%\n",
      "Epoch 79 : Train Loss 2.17, Accuracy 0.19%\n",
      "Epoch 80 : Train Loss 2.15, Accuracy 0.21%\n",
      "Epoch 81 : Train Loss 2.12, Accuracy 0.23%\n",
      "Epoch 82 : Train Loss 2.09, Accuracy 0.24%\n",
      "Epoch 83 : Train Loss 2.06, Accuracy 0.26%\n",
      "Epoch 84 : Train Loss 2.01, Accuracy 0.28%\n",
      "Epoch 85 : Train Loss 1.97, Accuracy 0.29%\n",
      "Epoch 86 : Train Loss 1.91, Accuracy 0.32%\n",
      "Epoch 87 : Train Loss 1.85, Accuracy 0.35%\n",
      "Epoch 88 : Train Loss 1.79, Accuracy 0.37%\n",
      "Epoch 89 : Train Loss 1.71, Accuracy 0.40%\n",
      "Epoch 90 : Train Loss 1.62, Accuracy 0.43%\n",
      "Epoch 91 : Train Loss 1.54, Accuracy 0.46%\n",
      "Epoch 92 : Train Loss 1.44, Accuracy 0.50%\n",
      "Epoch 93 : Train Loss 1.33, Accuracy 0.54%\n",
      "Epoch 94 : Train Loss 1.26, Accuracy 0.57%\n",
      "Epoch 95 : Train Loss 1.12, Accuracy 0.62%\n",
      "Epoch 96 : Train Loss 1.03, Accuracy 0.65%\n",
      "Epoch 97 : Train Loss 0.95, Accuracy 0.68%\n",
      "Epoch 98 : Train Loss 0.88, Accuracy 0.70%\n",
      "Epoch 99 : Train Loss 0.77, Accuracy 0.74%\n",
      "Epoch 100 : Train Loss 0.71, Accuracy 0.76%\n",
      "Epoch 101 : Train Loss 0.65, Accuracy 0.78%\n",
      "Epoch 102 : Train Loss 0.55, Accuracy 0.81%\n",
      "Epoch 103 : Train Loss 0.54, Accuracy 0.82%\n",
      "Epoch 104 : Train Loss 0.49, Accuracy 0.84%\n",
      "Epoch 105 : Train Loss 0.45, Accuracy 0.85%\n",
      "Epoch 106 : Train Loss 0.41, Accuracy 0.86%\n",
      "Epoch 107 : Train Loss 0.39, Accuracy 0.87%\n",
      "Epoch 108 : Train Loss 0.35, Accuracy 0.89%\n",
      "Epoch 109 : Train Loss 0.32, Accuracy 0.89%\n",
      "Epoch 110 : Train Loss 0.31, Accuracy 0.90%\n",
      "Epoch 111 : Train Loss 0.26, Accuracy 0.91%\n",
      "Epoch 112 : Train Loss 0.25, Accuracy 0.92%\n",
      "Epoch 113 : Train Loss 0.24, Accuracy 0.92%\n",
      "Epoch 114 : Train Loss 0.23, Accuracy 0.92%\n",
      "Epoch 115 : Train Loss 0.21, Accuracy 0.93%\n",
      "Epoch 116 : Train Loss 0.21, Accuracy 0.93%\n",
      "Epoch 117 : Train Loss 0.19, Accuracy 0.94%\n",
      "Epoch 118 : Train Loss 0.17, Accuracy 0.94%\n",
      "Epoch 119 : Train Loss 0.17, Accuracy 0.95%\n",
      "Epoch 120 : Train Loss 0.15, Accuracy 0.95%\n",
      "Epoch 121 : Train Loss 0.16, Accuracy 0.95%\n",
      "Epoch 122 : Train Loss 0.15, Accuracy 0.95%\n",
      "Epoch 123 : Train Loss 0.15, Accuracy 0.95%\n",
      "Epoch 124 : Train Loss 0.15, Accuracy 0.95%\n",
      "Epoch 125 : Train Loss 0.12, Accuracy 0.96%\n",
      "Epoch 126 : Train Loss 0.12, Accuracy 0.96%\n",
      "Epoch 127 : Train Loss 0.10, Accuracy 0.97%\n",
      "Epoch 128 : Train Loss 0.12, Accuracy 0.96%\n",
      "Epoch 129 : Train Loss 0.10, Accuracy 0.97%\n",
      "Epoch 130 : Train Loss 0.11, Accuracy 0.97%\n",
      "Epoch 131 : Train Loss 0.10, Accuracy 0.97%\n",
      "Epoch 132 : Train Loss 0.12, Accuracy 0.96%\n",
      "Epoch 133 : Train Loss 0.10, Accuracy 0.97%\n",
      "Epoch 134 : Train Loss 0.07, Accuracy 0.98%\n",
      "Epoch 135 : Train Loss 0.09, Accuracy 0.97%\n",
      "Epoch 136 : Train Loss 0.08, Accuracy 0.97%\n",
      "Epoch 137 : Train Loss 0.09, Accuracy 0.97%\n",
      "Epoch 138 : Train Loss 0.08, Accuracy 0.98%\n",
      "Epoch 139 : Train Loss 0.09, Accuracy 0.97%\n",
      "Epoch 140 : Train Loss 0.06, Accuracy 0.98%\n",
      "Epoch 141 : Train Loss 0.07, Accuracy 0.98%\n",
      "Epoch 142 : Train Loss 0.05, Accuracy 0.98%\n",
      "Epoch 143 : Train Loss 0.07, Accuracy 0.98%\n",
      "Epoch 144 : Train Loss 0.07, Accuracy 0.98%\n",
      "Epoch 145 : Train Loss 0.06, Accuracy 0.98%\n",
      "Epoch 146 : Train Loss 0.06, Accuracy 0.98%\n",
      "Epoch 147 : Train Loss 0.05, Accuracy 0.98%\n",
      "Epoch 148 : Train Loss 0.06, Accuracy 0.98%\n",
      "Epoch 149 : Train Loss 0.05, Accuracy 0.98%\n",
      "Epoch 150 : Train Loss 0.06, Accuracy 0.98%\n",
      "Total training time: 520.6623151302338\n"
     ]
    }
   ],
   "source": [
    "learning_rate = 0.1\n",
    "batch_size = 64\n",
    "epochs = 150\n",
    "\n",
    "loss_list = []\n",
    "acc_list = []\n",
    "\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "model = AlexNet().to(device)\n",
    "loss_function = torch.nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)\n",
    "\n",
    "tick = time.time()\n",
    "\n",
    "\n",
    "for epoch in range(epochs):\n",
    "    train_loss, correct, total, cnt = 0, 0, 0, 0\n",
    "    \n",
    "    for images, labels in train_loader:\n",
    "        \n",
    "      model.train()\n",
    "      images, labels = images.to(device), labels.to(device)\n",
    "\n",
    "      optimizer.zero_grad()\n",
    "      output = model(images)\n",
    "\n",
    "      loss = loss_function(output, labels)\n",
    "      loss.backward()\n",
    "\n",
    "      pred = output.max(1, keepdim=True)[1]\n",
    "      correct += pred.eq(labels.view_as(pred)).sum().item()\n",
    "      total += labels.size(0)\n",
    "      train_loss += loss.item()\n",
    "      cnt += 1\n",
    "\n",
    "      optimizer.step()\n",
    "\n",
    "    accuracy = correct/total\n",
    "    avg_loss = train_loss/cnt\n",
    "\n",
    "    loss_list.append(avg_loss)\n",
    "    acc_list.append(accuracy)\n",
    "    print(f\"Epoch {epoch + 1} : Train Loss {avg_loss:.2f}, Accuracy {accuracy:.2f}%\")\n",
    "    \n",
    "tock = time.time()\n",
    "print(f\"Total training time: {tock - tick}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 313
    },
    "id": "Cw_RiN6QxQcc",
    "outputId": "0dc32cc3-e91e-44be-cda5-b2a5fa53e3f1"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[]"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "fig, ax1 = plt.subplots()\n",
    "plt.title('Training with Randomized Label')\n",
    "plt.xlabel('Epochs')\n",
    "ax1.plot(np.arange(epochs), acc_list, color='red')\n",
    "ax1.tick_params(axis='y', labelcolor = 'red')\n",
    "plt.ylabel('Acuuracy', color='red')\n",
    "\n",
    "ax2 = ax1.twinx()\n",
    "ax2.plot(np.arange(epochs), loss_list, color='blue')\n",
    "ax2.tick_params(axis='y', labelcolor = 'blue')\n",
    "plt.ylabel('Loss', color='blue')\n",
    "plt.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "FPT_cDFQx_zj"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "provenance": []
  },
  "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}