NNDL-HW5-Q2-BeitClassifier.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fy3EdLePnOZT",
        "outputId": "f6338bb0-464f-43cb-f1b1-16945af8d466"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m5.8/5.8 MB\u001b[0m \u001b[31m79.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m452.9/452.9 KB\u001b[0m \u001b[31m41.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m7.6/7.6 MB\u001b[0m \u001b[31m96.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m182.4/182.4 KB\u001b[0m \u001b[31m10.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m213.0/213.0 KB\u001b[0m \u001b[31m9.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m132.0/132.0 KB\u001b[0m \u001b[31m14.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m140.6/140.6 KB\u001b[0m \u001b[31m17.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25h"
          ]
        }
      ],
      "source": [
        "!pip install -q transformers datasets"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Kd1bgPOypwGk"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "import torchvision\n",
        "import torchvision.transforms as transforms"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dhO4Zeh8pvyv",
        "outputId": "dd893595-36a4-453e-e393-d48f72e6f272"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Files already downloaded and verified\n",
            "Files already downloaded and verified\n"
          ]
        }
      ],
      "source": [
        "transform = transforms.Compose([transforms.ToTensor(), transforms.Resize(224), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])\n",
        "\n",
        "batch_size = 32\n",
        "\n",
        "trainset = torchvision.datasets.CIFAR10(root='./data', train=True,\n",
        "                                        download=True, transform=transform)\n",
        "trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size,\n",
        "                                          shuffle=True, num_workers=2)\n",
        "\n",
        "testset = torchvision.datasets.CIFAR10(root='./data', train=False,\n",
        "                                       download=True, transform=transform)\n",
        "testloader = torch.utils.data.DataLoader(testset, batch_size=batch_size,\n",
        "                                         shuffle=False, num_workers=2)\n",
        "\n",
        "classes = ['plane', 'car', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "u9voL61Lp-LW"
      },
      "outputs": [],
      "source": [
        "from transformers import  BeitFeatureExtractor, BeitForImageClassification\n",
        "\n",
        "model = BeitForImageClassification.from_pretrained(\"jadohu/BEiT-finetuned\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "flWnNizqqYVJ",
        "outputId": "603227d5-9d34-47bb-d8af-4fc3d1667b1e"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "BeitForImageClassification(\n",
              "  (beit): BeitModel(\n",
              "    (embeddings): BeitEmbeddings(\n",
              "      (patch_embeddings): BeitPatchEmbeddings(\n",
              "        (projection): Conv2d(3, 768, kernel_size=(16, 16), stride=(16, 16))\n",
              "      )\n",
              "      (dropout): Dropout(p=0.0, inplace=False)\n",
              "    )\n",
              "    (encoder): BeitEncoder(\n",
              "      (layer): ModuleList(\n",
              "        (0): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): Identity()\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (1): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.00909090880304575)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (2): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.0181818176060915)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (3): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.027272727340459824)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (4): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.036363635212183)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (5): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.045454543083906174)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (6): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.054545458406209946)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (7): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.06363636255264282)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (8): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.0727272778749466)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (9): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.08181818574666977)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (10): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.09090909361839294)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "        (11): BeitLayer(\n",
              "          (attention): BeitAttention(\n",
              "            (attention): BeitSelfAttention(\n",
              "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (key): Linear(in_features=768, out_features=768, bias=False)\n",
              "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "              (relative_position_bias): BeitRelativePositionBias()\n",
              "            )\n",
              "            (output): BeitSelfOutput(\n",
              "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
              "              (dropout): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BeitIntermediate(\n",
              "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BeitOutput(\n",
              "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
              "            (dropout): Dropout(p=0.0, inplace=False)\n",
              "          )\n",
              "          (layernorm_before): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "          (drop_path): BeitDropPath(p=0.10000000149011612)\n",
              "          (layernorm_after): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "        )\n",
              "      )\n",
              "    )\n",
              "    (layernorm): Identity()\n",
              "    (pooler): BeitPooler(\n",
              "      (layernorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
              "    )\n",
              "  )\n",
              "  (classifier): Linear(in_features=768, out_features=10, bias=True)\n",
              ")"
            ]
          },
          "execution_count": 4,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IkLrdDOBqppP",
        "outputId": "9260e44a-3d0c-4d3c-d5fd-c455363b7df5"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "85769674"
            ]
          },
          "execution_count": 5,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "sum([l.numel() for l in model.parameters()])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 648
        },
        "id": "gt5dEWsNrD4c",
        "outputId": "86e1ccc3-d504-43fa-e12c-00386fbfd96f"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch: 0\n",
            "\t Loss: 1.7394749669719223\n",
            "Epoch: 1\n",
            "\t Loss: 1.4057737993111004\n",
            "Epoch: 2\n",
            "\t Loss: 1.2311617908993366\n",
            "Epoch: 3\n",
            "\t Loss: 1.1053410907510146\n",
            "Epoch: 4\n",
            "\t Loss: 0.9854935388380491\n",
            "Epoch: 5\n",
            "\t Loss: 0.8862873429834118\n",
            "Epoch: 6\n"
          ]
        },
        {
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "output_type": "error",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-7-31f8d3c7cd5a>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     24\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     25\u001b[0m     \u001b[0ml\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbackward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 26\u001b[0;31m     \u001b[0moptimizer\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     27\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     28\u001b[0m     \u001b[0mlosses\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ml\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/optim/optimizer.py\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    138\u001b[0m                 \u001b[0mprofile_name\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m\"Optimizer.step#{}.step\"\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mformat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mobj\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__class__\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__name__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    139\u001b[0m                 \u001b[0;32mwith\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mautograd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprofiler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrecord_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mprofile_name\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 140\u001b[0;31m                     \u001b[0mout\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    141\u001b[0m                     \u001b[0mobj\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_optimizer_step_code\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    142\u001b[0m                     \u001b[0;32mreturn\u001b[0m \u001b[0mout\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/optim/optimizer.py\u001b[0m in \u001b[0;36m_use_grad\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m     21\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     22\u001b[0m             \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_grad_enabled\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdefaults\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'differentiable'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 23\u001b[0;31m             \u001b[0mret\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     24\u001b[0m         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     25\u001b[0m             \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_grad_enabled\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mprev_grad\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/optim/adam.py\u001b[0m in \u001b[0;36mstep\u001b[0;34m(self, closure, grad_scaler)\u001b[0m\n\u001b[1;32m    232\u001b[0m                     \u001b[0mstate_steps\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstate\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'step'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    233\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 234\u001b[0;31m             adam(params_with_grad,\n\u001b[0m\u001b[1;32m    235\u001b[0m                  \u001b[0mgrads\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    236\u001b[0m                  \u001b[0mexp_avgs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/optim/adam.py\u001b[0m in \u001b[0;36madam\u001b[0;34m(params, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, foreach, capturable, differentiable, fused, grad_scale, found_inf, amsgrad, beta1, beta2, lr, weight_decay, eps, maximize)\u001b[0m\n\u001b[1;32m    298\u001b[0m         \u001b[0mfunc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_single_tensor_adam\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    299\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 300\u001b[0;31m     func(params,\n\u001b[0m\u001b[1;32m    301\u001b[0m          \u001b[0mgrads\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    302\u001b[0m          \u001b[0mexp_avgs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/optim/adam.py\u001b[0m in \u001b[0;36m_single_tensor_adam\u001b[0;34m(params, grads, exp_avgs, exp_avg_sqs, max_exp_avg_sqs, state_steps, grad_scale, found_inf, amsgrad, beta1, beta2, lr, weight_decay, eps, maximize, capturable, differentiable)\u001b[0m\n\u001b[1;32m    410\u001b[0m                 \u001b[0mdenom\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mexp_avg_sq\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msqrt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m/\u001b[0m \u001b[0mbias_correction2_sqrt\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0madd_\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0meps\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    411\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 412\u001b[0;31m             \u001b[0mparam\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0maddcdiv_\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mexp_avg\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdenom\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalue\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0mstep_size\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    413\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    414\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ],
      "source": [
        "import torch\n",
        "from torch import nn\n",
        "\n",
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
        "epochs = 20\n",
        "\n",
        "optimizer = torch.optim.Adam(model.parameters())\n",
        "loss = nn.CrossEntropyLoss().to(device)\n",
        "model.to(device)\n",
        "\n",
        "model.train()\n",
        "for epoch in range(epochs): \n",
        "  print(\"Epoch:\", epoch)\n",
        "  losses = []\n",
        "  \n",
        "  for idx, data in enumerate(trainloader):\n",
        "    inputs, labels = data\n",
        "    pixel_values = inputs.to(device)\n",
        "    labels = labels.to(device)\n",
        "\n",
        "    optimizer.zero_grad()\n",
        "    outputs = model(pixel_values=pixel_values.squeeze(1))\n",
        "    l = loss(outputs.logits, labels)\n",
        "\n",
        "    l.backward()\n",
        "    optimizer.step()\n",
        "\n",
        "    losses.append(l.item())\n",
        "\n",
        "  print(\"\\t Loss:\", sum(losses)/len(losses))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zpTWY268s2GV",
        "outputId": "771a26f5-e5f2-4c95-9190-f6bef5a888d8"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting torchmetrics\n",
            "  Downloading torchmetrics-0.11.0-py3-none-any.whl (512 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m512.4/512.4 KB\u001b[0m \u001b[31m34.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: torch>=1.8.1 in /usr/local/lib/python3.8/dist-packages (from torchmetrics) (1.13.0+cu116)\n",
            "Requirement already satisfied: numpy>=1.17.2 in /usr/local/lib/python3.8/dist-packages (from torchmetrics) (1.21.6)\n",
            "Requirement already satisfied: packaging in /usr/local/lib/python3.8/dist-packages (from torchmetrics) (21.3)\n",
            "Requirement already satisfied: typing-extensions in /usr/local/lib/python3.8/dist-packages (from torchmetrics) (4.4.0)\n",
            "Requirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /usr/local/lib/python3.8/dist-packages (from packaging->torchmetrics) (3.0.9)\n",
            "Installing collected packages: torchmetrics\n",
            "Successfully installed torchmetrics-0.11.0\n"
          ]
        }
      ],
      "source": [
        "!pip install torchmetrics"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RaZu1dvUaedi",
        "outputId": "203eb3d0-d687-42a7-eaa6-cd54326cc5c0"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "test loss : 0.742321598724303\n",
            "test accuracy : 0.9503218531608582\n",
            "test precision : 0.8155014514923096\n"
          ]
        }
      ],
      "source": [
        "from torchmetrics.classification import BinaryF1Score, BinaryAccuracy, BinaryPrecision , BinaryRecall\n",
        "\n",
        "acc = BinaryAccuracy()\n",
        "prec = BinaryPrecision()\n",
        "\n",
        "softmax = nn.Softmax(dim=1)\n",
        "\n",
        "targets = []\n",
        "model_predicts = []\n",
        "losses = []\n",
        "model.eval() \n",
        "acc_list = []\n",
        "pre_list = []\n",
        "with torch.no_grad():\n",
        "  for idx, data in enumerate(trainloader):\n",
        "    inputs, labels = data\n",
        "    \n",
        "    X = inputs.to(device)\n",
        "    l = labels.to(device)\n",
        "\n",
        "    out = model(X.squeeze(1))\n",
        "    out = out.logits\n",
        "    loss_t = loss(out, l)\n",
        "    losses.append(loss_t.item())\n",
        "    \n",
        "    out = softmax(out)\n",
        "\n",
        "    ls = torch.zeros(out.shape[0], 10)\n",
        "    for i in range(out.shape[0]):\n",
        "      ls[i][l[i]] = 1.\n",
        "\n",
        "    acc_list.append(acc(out.cpu(), ls.cpu()))\n",
        "    pre_list.append(prec(out.cpu(), ls.cpu()))\n",
        "\n",
        "    targets.extend((torch.max(ls, 1)[1]).data.cpu().numpy())\n",
        "    model_predicts.extend((torch.max(out, 1)[1]).data.cpu().numpy())\n",
        "\n",
        "  print(f'test loss : {sum(losses)/len(losses)}')\n",
        "  print(f'test accuracy : {sum(acc_list)/len(acc_list)}')\n",
        "  print(f'test precision : {sum(pre_list)/len(pre_list)}')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 428
        },
        "id": "bT8CyVlDa1Bi",
        "outputId": "a4eef4c1-e2d8-4159-fc4f-622cb1102cd4"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 864x504 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "import seaborn as sns\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "from matplotlib import pyplot as plt\n",
        "\n",
        "cf_matrix = confusion_matrix(targets, model_predicts)\n",
        "\n",
        "df_cm = pd.DataFrame(cf_matrix/np.sum(cf_matrix)*10, \n",
        "                     index = [i for i in classes],\n",
        "                     columns = [i for i in classes]\n",
        "                     )\n",
        "plt.figure(figsize = (12, 7))\n",
        "sns.heatmap(df_cm, annot=True)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "6XyjOIE7oMNH"
      },
      "outputs": [],
      "source": []
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}