Assignment2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import os\n",
    "import numpy as np\n",
    "from clip import clip\n",
    "from torchvision.datasets import MNIST\n",
    "from tqdm.notebook import tqdm\n",
    "from torchvision.models import resnet50\n",
    "import torch.optim as optim\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision import datasets\n",
    "from torchvision import transforms\n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "import seaborn as sns\n",
    "import torch.nn as nn\n",
    "import matplotlib.pyplot as plt\n",
    "import random"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MNIST train dataset size： 60000\n",
      "MNIST testing dataset size： 10000\n",
      "MNIST image size： (28, 28)\n"
     ]
    }
   ],
   "source": [
    "mnist_test= MNIST(root=os.path.expanduser(\"~/.cache\"), download=True,train=False)\n",
    "mnist_train= MNIST(root=os.path.expanduser(\"~/.cache\"), download=True,train=True)\n",
    "print(\"MNIST train dataset size：\", len(mnist_train))\n",
    "print(\"MNIST testing dataset size：\", len(mnist_test))\n",
    "image, label = mnist_train[0]\n",
    "print(\"MNIST image size：\", image.size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sample images:\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1500x300 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def sample_images(dataset):\n",
    "    print(\"sample images:\")\n",
    "    fig, axes = plt.subplots(1, 5, figsize=(15, 3))\n",
    "    rand=[random.randint(0, 10000) for _ in range(5)]\n",
    "    for i in range(5):\n",
    "        image, label = dataset[rand[i]]\n",
    "        axes[i].imshow(image, cmap='gray')\n",
    "        axes[i].set_title(f'Label: {label}')\n",
    "        axes[i].axis('off')\n",
    "    plt.tight_layout()\n",
    "    plt.show()\n",
    "\n",
    "sample_images(mnist_test)\n",
    "\n",
    "def class_distribution(dataset):\n",
    "    labels = [label for _, label in dataset]\n",
    "    label_counts = np.bincount(labels)\n",
    "    num_classes = len(label_counts)\n",
    "\n",
    "    fig, ax = plt.subplots(figsize=(10, 5))\n",
    "    ax.bar(range(num_classes), label_counts, tick_label=[str(i) for i in range(num_classes)])\n",
    "    ax.set_xlabel('Class')\n",
    "    ax.set_ylabel('Count')\n",
    "    ax.set_title('MNIST Training Set Class Distribution')\n",
    "\n",
    "    ax.grid(color='gray', linestyle='--', linewidth=0.5, axis='y')\n",
    "    ax.tick_params(axis='both', which='major', labelsize=10)\n",
    "    for i, count in enumerate(label_counts):\n",
    "        ax.text(i, count + 100, str(count), ha='center', fontsize=8)\n",
    "\n",
    "    plt.tight_layout()\n",
    "    plt.show()\n",
    "\n",
    "class_distribution(mnist_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Clip Model with RN50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "CLIP(\n",
       "  (visual): ModifiedResNet(\n",
       "    (conv1): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "    (bn1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    (relu1): ReLU(inplace=True)\n",
       "    (conv2): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "    (bn2): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    (relu2): ReLU(inplace=True)\n",
       "    (conv3): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "    (bn3): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    (relu3): ReLU(inplace=True)\n",
       "    (avgpool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "    (layer1): Sequential(\n",
       "      (0): Bottleneck(\n",
       "        (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "        (downsample): Sequential(\n",
       "          (-1): AvgPool2d(kernel_size=1, stride=1, padding=0)\n",
       "          (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "          (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        )\n",
       "      )\n",
       "      (1): Bottleneck(\n",
       "        (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "      (2): Bottleneck(\n",
       "        (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "    )\n",
       "    (layer2): Sequential(\n",
       "      (0): Bottleneck(\n",
       "        (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "        (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "        (downsample): Sequential(\n",
       "          (-1): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "          (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "          (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        )\n",
       "      )\n",
       "      (1): Bottleneck(\n",
       "        (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "      (2): Bottleneck(\n",
       "        (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "      (3): Bottleneck(\n",
       "        (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "    )\n",
       "    (layer3): Sequential(\n",
       "      (0): Bottleneck(\n",
       "        (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "        (downsample): Sequential(\n",
       "          (-1): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "          (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "          (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        )\n",
       "      )\n",
       "      (1): Bottleneck(\n",
       "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "      (2): Bottleneck(\n",
       "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "      (3): Bottleneck(\n",
       "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "      (4): Bottleneck(\n",
       "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "      (5): Bottleneck(\n",
       "        (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "    )\n",
       "    (layer4): Sequential(\n",
       "      (0): Bottleneck(\n",
       "        (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "        (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "        (downsample): Sequential(\n",
       "          (-1): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "          (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "          (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        )\n",
       "      )\n",
       "      (1): Bottleneck(\n",
       "        (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "      (2): Bottleneck(\n",
       "        (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace=True)\n",
       "        (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "        (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace=True)\n",
       "        (avgpool): Identity()\n",
       "        (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu3): ReLU(inplace=True)\n",
       "      )\n",
       "    )\n",
       "    (attnpool): AttentionPool2d(\n",
       "      (k_proj): Linear(in_features=2048, out_features=2048, bias=True)\n",
       "      (q_proj): Linear(in_features=2048, out_features=2048, bias=True)\n",
       "      (v_proj): Linear(in_features=2048, out_features=2048, bias=True)\n",
       "      (c_proj): Linear(in_features=2048, out_features=1024, bias=True)\n",
       "    )\n",
       "  )\n",
       "  (transformer): Transformer(\n",
       "    (resblocks): Sequential(\n",
       "      (0): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (1): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (2): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (3): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (4): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (5): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (6): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (7): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (8): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (9): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (10): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "      (11): ResidualAttentionBlock(\n",
       "        (attn): MultiheadAttention(\n",
       "          (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): Sequential(\n",
       "          (c_fc): Linear(in_features=512, out_features=2048, bias=True)\n",
       "          (gelu): QuickGELU()\n",
       "          (c_proj): Linear(in_features=2048, out_features=512, bias=True)\n",
       "        )\n",
       "        (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      )\n",
       "    )\n",
       "  )\n",
       "  (token_embedding): Embedding(49408, 512)\n",
       "  (ln_final): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       ")"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model, preprocess = clip.load(name=\"RN50\", device=device, download_root='/shareddata/clip/')\n",
    "model.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "class_names = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "templates = ['a photo of the number: \"{}\".']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e6bda28ba42e486d982b047f1d129360",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/10 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def zeroshot_classifier(classnames, template):\n",
    "    with torch.no_grad():\n",
    "        zeroshot_weights = []\n",
    "        for classname in tqdm(classnames):\n",
    "            texts = [template.format(classname) for template in templates] \n",
    "            texts = clip.tokenize(texts).to(device) \n",
    "            class_embeddings = model.encode_text(texts) \n",
    "            class_embeddings /= class_embeddings.norm(dim=-1, keepdim=True)\n",
    "            class_embedding = class_embeddings.mean(dim=0)\n",
    "            class_embedding /= class_embedding.norm() \n",
    "            zeroshot_weights.append(class_embedding)\n",
    "        zeroshot_weights = torch.stack(zeroshot_weights, dim=1).to(device)\n",
    "    return zeroshot_weights\n",
    "zeroshot_weights = zeroshot_classifier(class_names, templates)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def model_inference(model, image, zeroshot_weights):\n",
    "    image_input = preprocess(image).unsqueeze(0).to(device)\n",
    "    with torch.no_grad():\n",
    "        image_features = model.encode_image(image_input)\n",
    "    image_features /= image_features.norm(dim=-1, keepdim=True) \n",
    "    logits = 100. * image_features @ zeroshot_weights\n",
    "    return logits\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def accuarcy(model,data,zeroshot_weights):\n",
    "    predictions=[]\n",
    "    acc=0\n",
    "    for image, class_id in data:\n",
    "        logits=model_inference(model, image, zeroshot_weights)\n",
    "        prediction = logits.topk(1)[1]\n",
    "        predictions.append(prediction.cpu().item())\n",
    "        if(prediction==class_id):\n",
    "            acc+=1\n",
    "    \n",
    "    return acc/len(mnist_test),predictions\n",
    "\n",
    "result=accuarcy(model,mnist_test,zeroshot_weights)\n",
    "accuracy=result[0]\n",
    "predictions=result[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy for clip model with RN50 in Mnist: 58.10 %\n"
     ]
    }
   ],
   "source": [
    "print(\"Accuracy for clip model with RN50 in Mnist:\",format(accuracy*100, '.2f'),\"%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "classification_report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.73      0.73      0.73       980\n",
      "           1       0.83      0.68      0.75      1135\n",
      "           2       0.41      0.54      0.46      1032\n",
      "           3       0.33      0.79      0.47      1010\n",
      "           4       0.71      0.60      0.65       982\n",
      "           5       0.98      0.32      0.48       892\n",
      "           6       0.49      0.39      0.43       958\n",
      "           7       0.59      0.70      0.64      1028\n",
      "           8       0.86      0.50      0.63       974\n",
      "           9       0.79      0.52      0.62      1009\n",
      "\n",
      "    accuracy                           0.58     10000\n",
      "   macro avg       0.67      0.58      0.59     10000\n",
      "weighted avg       0.67      0.58      0.59     10000\n",
      "\n",
      "confusion matrix\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Axes: >"
      ]
     },
     "execution_count": 109,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"classification_report\")\n",
    "print(classification_report(mnist.targets.cpu(), predictions))\n",
    "print(\"confusion matrix\")\n",
    "matrix=confusion_matrix(mnist.targets.cpu(), predictions)\n",
    "fmt = 'd' \n",
    "annot_kws = {'fontsize': 12}  \n",
    "sns.heatmap(matrix, annot=True, fmt=fmt, cmap='Blues', cbar=False, annot_kws=annot_kws)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "class SimpleCNN(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "         \n",
    "        self.conv1 = nn.Conv2d(1, 32, kernel_size=3)  \n",
    "        self.pool1 = nn.MaxPool2d(2)                  \n",
    "        self.drop1 = nn.Dropout(p=0.15)\n",
    "        self.conv2 = nn.Conv2d(32, 64, kernel_size=3)   \n",
    "        self.pool2 = nn.MaxPool2d(2)\n",
    "        self.drop2 = nn.Dropout(p=0.15)\n",
    "        self.fc1 = nn.Linear(1600, 64)          \n",
    "        self.fc2 = nn.Linear(64, 10)\n",
    "    \n",
    "    def forward(self, x):\n",
    "        x = self.drop1(self.pool1(torch.relu(self.conv1(x))))     \n",
    "        x = self.drop2(self.pool2(torch.relu(self.conv2(x))))\n",
    "        x = torch.flatten(x, start_dim=1)\n",
    "        x = torch.relu(self.fc1(x))\n",
    "        x = self.fc2(x)\n",
    "       \n",
    "        return x\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SimpleCNN(\n",
       "  (conv1): Conv2d(1, 32, kernel_size=(3, 3), stride=(1, 1))\n",
       "  (pool1): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "  (drop1): Dropout(p=0.15, inplace=False)\n",
       "  (conv2): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1))\n",
       "  (pool2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "  (drop2): Dropout(p=0.15, inplace=False)\n",
       "  (fc1): Linear(in_features=1600, out_features=64, bias=True)\n",
       "  (fc2): Linear(in_features=64, out_features=10, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model2 = SimpleCNN()\n",
    "model2.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "LEARNING_RATE = 1e-1\n",
    "MOMENTUM = 0.9\n",
    "STEP=5\n",
    "GAMMA=0.5\n",
    "optimizer = optim.SGD(model2.parameters(), lr=LEARNING_RATE, momentum=MOMENTUM)\n",
    "scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=STEP, gamma=GAMMA)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def Output(model2, image, target): \n",
    "    criterion = nn.CrossEntropyLoss()\n",
    "    output = model2(image)\n",
    "    loss=criterion(output,target)\n",
    "    return output, loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "EPOCHS=30\n",
    "BATCH_SIZE=128\n",
    "Trans = transforms.Compose([transforms.ToTensor(), \n",
    "                               transforms.Normalize(mean = (0.1307,),std = (0.3081,))\n",
    "                               ])\n",
    "\n",
    "train_set = datasets.MNIST(\"data\",train=True,download=True,transform=Trans)\n",
    "test_set = datasets.MNIST(\"data\",train=False,download=True,transform=Trans)\n",
    "train_loader = DataLoader(train_set,batch_size=BATCH_SIZE,shuffle=True)\n",
    "test_loader = DataLoader(test_set,batch_size=BATCH_SIZE,shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def test(model,optimizer,scheduler):\n",
    "    for epoch in range(EPOCHS):\n",
    "        running_cls_loss = 0.0\n",
    "        for batch_idx, (image, target) in enumerate(train_loader):\n",
    "            image = image.to(device)\n",
    "            target = target.to(device)\n",
    "            outputs, loss = Output(model2, image, target)\n",
    "            _, preds = torch.max(outputs, 1) \n",
    "            loss_data = loss.data.item()\n",
    "            if np.isnan(loss_data):\n",
    "                raise ValueError('loss is nan while training')\n",
    "            running_cls_loss += loss.item()\n",
    "            loss.backward()\n",
    "            optimizer.step()\n",
    "            optimizer.zero_grad()\n",
    "        scheduler.step()\n",
    "        if  (epoch +1) == EPOCHS:\n",
    "            model.eval()\n",
    "            val_loss = 0.0\n",
    "            val_corrects = 0\n",
    "            for batch_idx, (image, target) in enumerate(test_loader):\n",
    "                image = image.to(device)\n",
    "                target = target.to(device)\n",
    "                outputs, loss = Output(model, image, target)\n",
    "                _, preds = torch.max(outputs, 1)  \n",
    "                val_loss += loss.item()\n",
    "                val_corrects += torch.sum(preds == target.data)    \n",
    "            \n",
    "            val_acc = val_corrects.double() / len(test_set)\n",
    "    return val_acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy for CNN in Mnist: 99.25 %\n"
     ]
    }
   ],
   "source": [
    "testA=test(model2,optimizer,scheduler)\n",
    "print(\"Accuracy for CNN in Mnist:\",format(testA*100, '.2f'),\"%\")"
   ]
  }
 ],
 "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}