notebooks/CIFAR10 Classification - ViT.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CIFAR10 Image Classification with Vision in Transformers (ViT)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch \n",
    "import torch.nn as nn\n",
    "from torch.utils.data import Dataset\n",
    "\n",
    "from torchvision import datasets\n",
    "import torchvision.transforms as transforms\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda\n"
     ]
    }
   ],
   "source": [
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "print(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.manual_seed(0)\n",
    "np.random.seed(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "BATCH_SIZE = 64\n",
    "LR = 3e-4\n",
    "NUM_EPOCHES = 20"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean, std = (0.5,), (0.5,)\n",
    "\n",
    "transform = transforms.Compose([transforms.ToTensor(),\n",
    "                                transforms.Normalize(mean, std)\n",
    "                              ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n",
      "Files already downloaded and verified\n"
     ]
    }
   ],
   "source": [
    "trainset = datasets.CIFAR10('data/CIFAR10/', download=True, train=True, transform=transform)\n",
    "trainloader = torch.utils.data.DataLoader(trainset, batch_size=BATCH_SIZE, shuffle=True)\n",
    "\n",
    "testset = datasets.CIFAR10('data/CIFAR10/', download=True, train=False, transform=transform)\n",
    "testloader = torch.utils.data.DataLoader(testset, batch_size=BATCH_SIZE, shuffle=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformer_package.models import ViT"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": "ViT(\n  (dropout_layer): Dropout(p=0.2, inplace=False)\n  (embeddings): Linear(in_features=192, out_features=512, bias=True)\n  (encoders): ModuleList(\n    (0): VisionEncoder(\n      (norm1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n      (norm2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n      (attention): MultiHeadAttention(\n        (dropout_layer): Dropout(p=0.2, inplace=False)\n        (Q): Linear(in_features=512, out_features=512, bias=True)\n        (K): Linear(in_features=512, out_features=512, bias=True)\n        (V): Linear(in_features=512, out_features=512, bias=True)\n        (linear): Linear(in_features=512, out_features=512, bias=True)\n      )\n      (mlp): Sequential(\n        (0): Linear(in_features=512, out_features=2048, bias=True)\n        (1): GELU()\n        (2): Dropout(p=0.2, inplace=False)\n        (3): Linear(in_features=2048, out_features=512, bias=True)\n        (4): Dropout(p=0.2, inplace=False)\n      )\n    )\n    (1): VisionEncoder(\n      (norm1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n      (norm2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n      (attention): MultiHeadAttention(\n        (dropout_layer): Dropout(p=0.2, inplace=False)\n        (Q): Linear(in_features=512, out_features=512, bias=True)\n        (K): Linear(in_features=512, out_features=512, bias=True)\n        (V): Linear(in_features=512, out_features=512, bias=True)\n        (linear): Linear(in_features=512, out_features=512, bias=True)\n      )\n      (mlp): Sequential(\n        (0): Linear(in_features=512, out_features=2048, bias=True)\n        (1): GELU()\n        (2): Dropout(p=0.2, inplace=False)\n        (3): Linear(in_features=2048, out_features=512, bias=True)\n        (4): Dropout(p=0.2, inplace=False)\n      )\n    )\n  )\n  (norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n  (classifier): Sequential(\n    (0): Linear(in_features=512, out_features=10, bias=True)\n  )\n)"
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "image_size = 32\n",
    "channel_size = 3\n",
    "patch_size = 8\n",
    "embed_size = 512\n",
    "num_heads = 8\n",
    "classes = 10\n",
    "num_layers = 2\n",
    "hidden_size = 256\n",
    "dropout = 0.2\n",
    "\n",
    "model = ViT(image_size, channel_size, patch_size, embed_size, num_heads, classes, num_layers, hidden_size, dropout=dropout).to(device)\n",
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input Image Dimensions: torch.Size([64, 3, 32, 32])\n",
      "Label Dimensions: torch.Size([64])\n",
      "----------------------------------------------------------------------------------------------------\n",
      "Output Dimensions: torch.Size([64, 10])\n"
     ]
    }
   ],
   "source": [
    "for img, label in trainloader:\n",
    "    img = img.to(device)\n",
    "    label = label.to(device)\n",
    "    \n",
    "    print(\"Input Image Dimensions: {}\".format(img.size()))\n",
    "    print(\"Label Dimensions: {}\".format(label.size()))\n",
    "    print(\"-\"*100)\n",
    "    \n",
    "    out = model(img)\n",
    "    \n",
    "    print(\"Output Dimensions: {}\".format(out.size()))\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.Adam(params=model.parameters(), lr=LR)\n",
    "scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------------------------------------\n",
      "Epoch: 1 Train mean loss: 1399.62835824\n",
      "       Train Accuracy%:  34.782 == 17391 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 2 Train mean loss: 1213.23938024\n",
      "       Train Accuracy%:  43.97 == 21985 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 3 Train mean loss: 1147.39728940\n",
      "       Train Accuracy%:  47.228 == 23614 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 4 Train mean loss: 1107.55177343\n",
      "       Train Accuracy%:  49.094 == 24547 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 5 Train mean loss: 1072.51819640\n",
      "       Train Accuracy%:  50.772 == 25386 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 6 Train mean loss: 1049.94498992\n",
      "       Train Accuracy%:  51.904 == 25952 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 7 Train mean loss: 1023.63963330\n",
      "       Train Accuracy%:  52.93 == 26465 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 8 Train mean loss: 1001.64437157\n",
      "       Train Accuracy%:  53.81 == 26905 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 9 Train mean loss: 975.06941009\n",
      "       Train Accuracy%:  55.324 == 27662 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 10 Train mean loss: 955.83791620\n",
      "       Train Accuracy%:  56.082 == 28041 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 11 Train mean loss: 934.50161016\n",
      "       Train Accuracy%:  56.914 == 28457 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 12 Train mean loss: 915.14476407\n",
      "       Train Accuracy%:  57.992 == 28996 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 13 Train mean loss: 893.43885159\n",
      "       Train Accuracy%:  58.948 == 29474 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 14 Train mean loss: 872.27832127\n",
      "       Train Accuracy%:  60.204 == 30102 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 15 Train mean loss: 853.43125552\n",
      "       Train Accuracy%:  60.706 == 30353 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 16 Train mean loss: 837.26983732\n",
      "       Train Accuracy%:  61.29 == 30645 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 17 Train mean loss: 819.46424955\n",
      "       Train Accuracy%:  62.18 == 31090 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 18 Train mean loss: 795.84717381\n",
      "       Train Accuracy%:  63.626 == 31813 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 19 Train mean loss: 778.38073927\n",
      "       Train Accuracy%:  64.438 == 32219 / 50000\n",
      "-------------------------------------------------\n",
      "-------------------------------------------------\n",
      "Epoch: 20 Train mean loss: 761.25517702\n",
      "       Train Accuracy%:  65.018 == 32509 / 50000\n",
      "-------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "loss_hist = {}\n",
    "loss_hist[\"train accuracy\"] = []\n",
    "loss_hist[\"train loss\"] = []\n",
    "\n",
    "for epoch in range(1, NUM_EPOCHES+1):\n",
    "    model.train()\n",
    "    \n",
    "    epoch_train_loss = 0\n",
    "        \n",
    "    y_true_train = []\n",
    "    y_pred_train = []\n",
    "        \n",
    "    for batch_idx, (img, labels) in enumerate(trainloader):\n",
    "        img = img.to(device)\n",
    "        labels = labels.to(device)\n",
    "        \n",
    "        preds = model(img)\n",
    "        \n",
    "        loss = criterion(preds, labels)\n",
    "        optimizer.zero_grad()\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        \n",
    "        y_pred_train.extend(preds.detach().argmax(dim=-1).tolist())\n",
    "        y_true_train.extend(labels.detach().tolist())\n",
    "            \n",
    "        epoch_train_loss += loss.item()\n",
    "    \n",
    "    #scheduler.step(epoch_train_loss)\n",
    "    \n",
    "    loss_hist[\"train loss\"].append(epoch_train_loss)\n",
    "    \n",
    "    total_correct = len([True for x, y in zip(y_pred_train, y_true_train) if x==y])\n",
    "    total = len(y_pred_train)\n",
    "    accuracy = total_correct * 100 / total\n",
    "    \n",
    "    loss_hist[\"train accuracy\"].append(accuracy)\n",
    "    \n",
    "    print(\"-------------------------------------------------\")\n",
    "    print(\"Epoch: {} Train mean loss: {:.8f}\".format(epoch, epoch_train_loss))\n",
    "    print(\"       Train Accuracy%: \", accuracy, \"==\", total_correct, \"/\", total)\n",
    "    print(\"-------------------------------------------------\")\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(loss_hist[\"train accuracy\"])\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(loss_hist[\"train loss\"])\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy%:  58.69 == 5869 / 10000\n"
     ]
    }
   ],
   "source": [
    "with torch.no_grad():\n",
    "    model.eval()\n",
    "    \n",
    "    y_true_test = []\n",
    "    y_pred_test = []\n",
    "    \n",
    "    for batch_idx, (img, labels) in enumerate(testloader):\n",
    "        img = img.to(device)\n",
    "        label = label.to(device)\n",
    "    \n",
    "        preds = model(img)\n",
    "        \n",
    "        y_pred_test.extend(preds.detach().argmax(dim=-1).tolist())\n",
    "        y_true_test.extend(labels.detach().tolist())\n",
    "        \n",
    "    total_correct = len([True for x, y in zip(y_pred_test, y_true_test) if x==y])\n",
    "    total = len(y_pred_test)\n",
    "    accuracy = total_correct * 100 / total\n",
    "    \n",
    "    print(\"Test Accuracy%: \", accuracy, \"==\", total_correct, \"/\", total)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  }
 ],
 "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}