pytorchmnist.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "pytorchmnist.ipynb",
      "version": "0.3.2",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "IBLsto4SYIgA",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 127
        },
        "outputId": "9e5fa7fa-96ff-48e2-f026-1a3c3f6d2cf5"
      },
      "source": [
        "from google.colab import drive\n",
        "\n",
        "drive.mount('/gdrive', force_remount=True)"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /gdrive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YjQt4HGJYQXv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import itertools\n",
        "\n",
        "import torch \n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import torch.optim as optim\n",
        "from torchvision import datasets, transforms\n",
        "\n",
        "from IPython.display import Image\n",
        "from IPython import display\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "%matplotlib inline"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RsWwWerCYgYU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "c561a5c7-54c3-4613-e21c-0f61eb5f5706"
      },
      "source": [
        "torch.cuda.get_device_name(0)"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'Tesla T4'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PWNaP3fWYi7U",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 289
        },
        "outputId": "2d0a802e-35d7-46d7-8cd6-ebefa65f1ba4"
      },
      "source": [
        "# prepare batch_loader \n",
        "batch_size=64\n",
        "trn_dataset = datasets.MNIST('../mnist_data/',\n",
        "                             download=True,\n",
        "                             train=True,\n",
        "                             transform=transforms.Compose([\n",
        "                                 transforms.ToTensor(), # image to Tensor\n",
        "                                 transforms.Normalize((0.1307,), (0.3081,)) # image, label\n",
        "                             ])) \n",
        "\n",
        "trn_loader = torch.utils.data.DataLoader(trn_dataset,\n",
        "                                         batch_size=batch_size,\n",
        "                                         shuffle=True)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "  0%|          | 0/9912422 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz to ../mnist_data/MNIST/raw/train-images-idx3-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "9920512it [00:00, 19633811.95it/s]                            \n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting ../mnist_data/MNIST/raw/train-images-idx3-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "32768it [00:00, 331506.51it/s]\n",
            "0it [00:00, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz to ../mnist_data/MNIST/raw/train-labels-idx1-ubyte.gz\n",
            "Extracting ../mnist_data/MNIST/raw/train-labels-idx1-ubyte.gz\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz to ../mnist_data/MNIST/raw/t10k-images-idx3-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "1654784it [00:00, 5174103.90it/s]                           \n",
            "8192it [00:00, 131582.48it/s]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting ../mnist_data/MNIST/raw/t10k-images-idx3-ubyte.gz\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz to ../mnist_data/MNIST/raw/t10k-labels-idx1-ubyte.gz\n",
            "Extracting ../mnist_data/MNIST/raw/t10k-labels-idx1-ubyte.gz\n",
            "Processing...\n",
            "Done!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bAN9Ql5EYyo7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "val_dataset = datasets.MNIST(\"../mnist_data/\", \n",
        "                             download=False,\n",
        "                             train=False,\n",
        "                             transform= transforms.Compose([\n",
        "                               transforms.ToTensor(),\n",
        "                               transforms.Normalize((0.1307, ),(0.3081, ))\n",
        "                           ]))\n",
        "\n",
        "val_loader = torch.utils.data.DataLoader(val_dataset,\n",
        "                                         batch_size=batch_size,\n",
        "                                         shuffle=True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n0-4P6jhY0sj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "num_batches = len(trn_loader)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "va2uE7zcY5em",
        "colab_type": "text"
      },
      "source": [
        "MNIST 그림을 확인해보자."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gRbcf3DkY2tT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "data = next(iter(trn_loader))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZFKBOWE2ZCFK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "img, label = data"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CPqYrWq1ZEwx",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "c4df8daa-c74a-4e0b-8079-35b785330a1f"
      },
      "source": [
        "img.size() # batch_size, num_channel, x, y"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([64, 1, 28, 28])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JZCMnoHwZG8B",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "e000cb62-1073-4442-8e28-e5ee969eff02"
      },
      "source": [
        "label.size() # 0~9까지의 숫자가 batch_size 만큼"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([64])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1cBD6YoaZKLJ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 357
        },
        "outputId": "708d9ebc-a2bb-4c8a-9d90-066802313352"
      },
      "source": [
        "# create figure for plotting\n",
        "row_num = 5\n",
        "col_num = 4\n",
        "fig, ax = plt.subplots(row_num, col_num, figsize=(6,6))\n",
        "for i, j in itertools.product(range(row_num), range(col_num)):\n",
        "    ax[i,j].get_xaxis().set_visible(False)\n",
        "    ax[i,j].get_yaxis().set_visible(False)\n",
        "    \n",
        "    \n",
        "for k in range(row_num*col_num):\n",
        "    i = k//col_num\n",
        "    j = k%col_num\n",
        "    ax[i,j].cla() # clear the current axis\n",
        "    ax[i,j].imshow(img[k,:].data.cpu().numpy().reshape(28,28), cmap='Greys')"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUwAAAFUCAYAAABP6wWpAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzt3XmczWX/x/HXkDWlNEiFURR3G1JS\ntCkVLTctiJafNty3RI9U2kR3lruF3JVukjvu3PxCq1/RQntoU9nTSMoyZcua5feH+3Ouc86cGdfM\nnO85Z855P/85M2e9+pq+5/29ls+VtXfvXkREZP/KJLsBIiKlhU6YIiKedMIUEfGkE6aIiCedMEVE\nPOmEKSLi6YDCHszOzt6bk5OToKYkXm5uLnl5eVnJ+nwd3+Do2AYnk49toSfMnJwc5s2bF0yrUkCz\nZs2S+vk6vsHRsQ1OJh9bXZKLiHgqNGGKSKSNGzcC0Lp1awAWL14MwIgRI0LP6datW+IbJgmhhCki\n4imlEuaWLVtCP+/evTvmc8qVKwdApUqVEtImyWxbt24FYNasWQA899xzAMyfPx+AUaNGAUqVmUIJ\nU0TEU1IS5sqVKwEYPHhwxP3Tpk0L/bxu3ToArJpSVta+UX6bztCmTRsA7rnnHgBq164dXIMlY91+\n++0AjBs3LubjSpbxV6FCBQB27twJwOTJk0OPXXXVVUlpk1HCFBHxlJCEuXz5cgCGDBkCwL///W8A\ntm/fDrj06CM3NxeAf/7znwBMnToVcH1MDRs2LHF7RczYsWMBKFMmMlu8++67yWhOWrNjvWvXLsAd\n8/ArTyVMEZFSItCEaf07L730EuBGHOPJ+jpPOeUUAKZMmQLARRddFPfPkvR33333Aa6P3GZj2DxL\n9VkGJ4jzQ7wpYYqIeAokYdoo+PPPPw/k7/8xdevWBaBr164A9OvXL/RYlSpVYr4mLy8PgLfffhuA\na665BnD9oXfffTcA2dnZQPLXiwfp008/BWDbtm2Am6M6adKkiOf98MMPALzxxhuh+y688EIArrvu\nOgCuvvpqAA44IKWm5iac9aeff/75ALRq1QpQskymmTNnhn7+8ccfAahTp05S2qKEKSLiKZA4YSnQ\nkmX0KLjNnRw0aFCR39uSY6dOnQCXKH/66ScAvv32W8CNYqZjwtywYQMAbdu2Bdz65oLYHFVL4wCf\nfPIJANdeey0A9957LwAfffQRAEcccUQcW5z6li5dCsCMGTMi7reZHZI8jRo1Cv182GGHJbElSpgi\nIt4S2mFlfWjWPxQP3bt3B9zoZiY4+OCDAdeHuWjRoojHbaXE2WefDbikX758+dBzbBXFpk2bABg4\ncCAAxx13HOCSuyXPdGcJ88svvwTcFcqpp56atDZlmksvvRRw/y9v3rwZgFWrVoWeY/UmDjzwwAS3\nbh8lTBERT4EkzCZNmgCwevVqAH799VcAGjRoAEDZsmXj9llDhw6N23uVFpYYjz322IjborC0aX3C\nTZs2BeCpp54CYPbs2UD6J0yb0bFs2TIAHn/8cQCaN2+etDZlKps1M3r0aMCNU9jqPnDjIzVq1Ehs\n4/5LCVNExFOgfZjVq1ePuI2nHTt2AG6EOHokfsGCBXH/zHT03XffAXDrrbcCrpL4K6+8krQ2JVK9\nevWA/HOF//znPwNw1FFHeb3P2rVrQz8Xd78buwKzW8mvV69eALzzzjtJ+XwlTBERT6V2WYfNH7Rk\nabfWt2F9cRLbH3/8Abg+SlslNGzYMEAV7X3ZFU6PHj1C97322msRz9mzZw9Q8Io3e9zmDE+YMAGA\n+vXrx7exacAqGSWLEqaIiKdSlzDt2zt8XXQ463tL1jyt0sKOkx1P67Nr3Lhx0tpUmlgl9g8//BBw\ne/yUxBdffAG4+Yjvv/8+EMwYQCpr164dAIcffjjgZtuAO0Z2vE866aSEtk0JU0TEU6lJmDZfzuZm\n2Sh5NJtXKPmF78RpK3ms6ovNfcs0Z511FuBWk3z//feA2zsqus/M+iwtWVriKah/MvwzCtpZYP36\n9YBLTbbqqFatWjHbkO4qV64MRK5MM1Yzc3/1E4KihCki4kknTBERT3G9JLepKlbQ1or62uZGBbHy\nYx07dgzdZ5ePVmCid+/eEe9pDjnkEMANYthEZMnPCkuAKyp8xRVXAFCtWrWktCnZrMiGXQbbdDW7\n1P7ss88A93dl04fs8tkuxcPLCFpxZvPQQw8V2gbrbjrmmGOK+V8hiaKEKSLiKS4J0wZgbNmSJcq9\ne/cCBXd2Rz/+6quvhh775ZdfAFfQNnqCuhk8eDAALVu2LOF/RfoLL2Jg+vfvn/iGpCBbjnjBBRcA\nLmF26NABgNNPPx3IPyndhKfK/SVK8WOLKWKxbVjOPPNMoPBBt3hSwhQR8VSihGnFPK0g8Ny5c0vU\nGNsi18fIkSMBuOWWW0r0mZlk/PjxoZ9tepYmqhduzZo1QMHJUoIzZswYAM4999x8jz3zzDOAm2a0\nv3GSeFHCFBHxVKKEaSPZBSVL20rBNj1buHAhEJl0istGL8OLHkhs1ncZnpI6d+6cpNakNtuYz25t\nW9ejjz4acCPZ1sdu2yiEL9+zUW9b2ldYXxzA8uXLAVeEw8Tqc84k1j952mmnhe6bM2dOxHOsL1MJ\nU0QkxZQoYdoItY12m/bt2wOuTJX1A9kSMXt+9OtiKeg5llJt9NI2Q5P8bAvicDavVQpXtWpVwJXB\ns3maVnDZtvJ4/vnnQ6+xn227C0unBbnsssuAxI30lha2lU2LFi1C90UnzETTv5CIiKcSJUwrNhs9\nN9IKGFj6+/bbbyMej36+/R6+8dShhx4a8dpYKQngzjvvBNzGa9q8yrF0PmrUKCCy31Lbx/qxhBk9\nt3LIkCGA2+Bv5syZocdsBkffvn0BlzTt99tuuw1ws0usX3TEiBERn5Gpq6+i3XjjjaGfo49Roilh\nioh4KlHCtG9S+8Y033zzDbD/lT7m5ptvBmD48OGh+ypUqAC40UcbtVyxYgUAb775JuDmYV188cUR\n94ePrGUqO1YTJ04E4IYbbgg9tr9/EylcdEK3/nmAPn36RDxWs2ZNwBVptuRoZcysVFymFQoujZQw\nRUQ8lShh2sh0165dS9QI23Ar1nw1m8tmm5pFV0SKtr85b5lk2bJlEb9ff/31SWpJ+rO0GP1zuIK2\n7FWyLD2UMEVEPJUoYdq8MVvRkwiWIJUk9y96Y67wfjaRdFDSq9uiUsIUEfFUajZBE382O8Gq1V91\n1VXJbI5IiRx//PGhn8M38ksGJUwREU9KmGnI9j2y+qI2Omv71oCrMC4i/pQwRUQ8KWGmIavyYsmy\nYsWKgFKlSEkpYYqIeFLCTEPly5cH3FpyEYkPJUwREU9ZhVU9z8rKWgekc0ypu3fv3qQt5NXxDY6O\nbXAy+dgWesIUERFHl+QiIp50whQR8aQTpoiIJ50wRUQ86YQpIuJJJ0wREU86YYqIeNIJU0TEk06Y\nIiKeCi2+kZ2dvTcnJydBTUm83Nxc8vLyspL1+Tq+wdGxDU4mH9tCT5g5OTnMmzcvmFalgGbNmiX1\n83V8g6NjG5xMPra6JBcR8ZQS9TCtAMhzzz0Xuu/mm28G4OqrrwZg0qRJiW+YiEgYJUwREU8pkTBz\nc3MBuOWWW0L3ZWXt63OtUKFCMpokIpKPEqaIiKeUSJgPP/xwvvtOO+00AJ566qlENycjbNu2DYC7\n774bgNtvvx2AevXqJa1NIqlOCVNExFNSE2ZeXh4AH3/8cb7Hxo4dC8BBBx2U0DZliqlTpwIwcuRI\nwCVMKbmnn34agL/85S+Au0rq2bNn0tok8aGEKSLiKakJ88EHHwRg8eLFALRq1Sr0WO3atZPSpkzx\nn//8J+L33bt3J6kl6cFSJbhkaZQs04cSpoiIp6QkzNWrVwMwc+ZMACpWrAjA6NGjQ89R32Wwjj32\nWADeeOMNAF544QUABg4cmLQ2Sem2ceNGADZt2gS4OdQ1atSI+2ctXboUgI4dOwLw9ddfA9CtWzcA\nHn30UQCqVq0a189VwhQR8ZTQhGn9ZDb3b9myZYDru7TUI8GzPjdL9zfeeGMym1PqRfdbQubNIbZk\nedtttwHwzjvvAO5v64knnijxZ2zduhWAW2+9FXDJ0lYG2uya6dOnA7Bq1aoSf2Y4JUwREU8JTZib\nN28GXH+ZsX4HCZb1HQPs2LEDcMe+bt26SWlTaWczPGLJtNFxm9nSrl07AF555RUAnnzySQBGjBgB\nuDR4zz33AHDiiScCcO655wKuz3Pnzp0ArFmzJvQZl19+OeCS5cknnwxAmTL7st/nn38OuApo8aaE\nKSLiKanzMNu2bQvAFVdckcxmZIwFCxbku69FixZJaEn66N+/f7KbkHJuuukmAJo3bw7AAw88AMCr\nr74KuIQ5ePDgiN8rV64MuIrn1if61Vdfhd7bkqO9JroP05LmjBkz4vsf9V9KmCIinhKaMF966aWI\n3+3buUqVKolsRsaaNWtWspuQdmxNvuRnfZPTpk2LuP+zzz4DYOLEiQDMnz8fgIULFwIwe/ZswCVO\nuwX4/fffAahZsybg+j27dOkCuKvWoChhioh4SkjC3LVrF+ASjm3R2bhx40R8vPzXokWL8t3XpEmT\nJLQkvWXa/Muisr5NuzXRK4VsjrCt5gGXPm1u56BBg4JtbBQlTBERTwlJmNu3bwfgxRdfBOCEE04A\nIvsmwp8HLg3ZaoHWrVsDbk6X9Vk0bNgQgCOPPDKQtqcDm8+2fPnyfI/Vr18/0c0RicnWfRe2/ttG\nycP3/0okJUwREU8JSZgTJkyI+P2uu+6K+H3u3LmA24sc3MhZQcaNGwdA9erVAfj+++8BjbjHYqOS\ntgoC4LDDDgPggANSYluntGJXQ1Iya9euBWDevHmh+5o2bQrA4YcfnpQ2KWGKiHjSCVNExFNCrsd+\n/vnniN/vu+8+AIYPHw64Mm82rSDcKaecAsAxxxwDQHZ2NuAGMN58803AFR/u06dPXNueDqzIgS0f\nA+jXrx+Qf+BN/EQX3ejQoUPo5+OOOy7RzUlLVixmy5YtoftatmwJwPr16wFXRKYg8d7qRglTRMRT\nUnr8V6xYEXFrrrzyytDPtkHa0UcfDUC5cuUAN0hh3yyTJ08GoHfv3oArLKrk5NhE4HCnn356ElqS\nPmy6m9FAT/xZ0Y7wKyMrEGy3lj7tOdHFOc466yzATUMsaaFsJUwREU9JnVNiU1ssTVo6BJcoC2Lf\nLMOGDQNgw4YNAOzZsyfu7UxH9s0rkmi2zYSNXdhV4iOPPAIUnBbBFd+waUVWCi6aFRa+4IILAGjQ\noEFc2q6EKSLiKakJ05ZK2rdAUYwZMwaA7777DoBevXoB6rsMZ33EVjYr1kZdIoliyfKSSy4BXCEN\nS5DRtyb89/feew9wCTLe2+jujxKmiIinpCZM245zwIABBT7H5ruNHDky4n7rs7zjjjsAN6JmJerF\nFTCx7Y27du2azOaklehRctm/vLw8wCVLuyps06YNkL/4b6z/l5Pd966zi4iIp4QkTFtpYt/Kn3zy\nCeDSY+fOnb3fq2/fvoDrBzn77LOB/P0eAuPHj4/4PdH9PelMW1MUXa1atQDXD2kj3AWNOxTUp5lM\nSpgiIp4SkjArVaoEwNtvvw3A0KFDARg4cOB+X2vrcm2E94YbbgBUxs3Hyy+/nOwmiITY3Or99UMu\nXboUcPMwU4kSpoiIp4SOklvStFHxwkbHpeSssktQm9pnMtvoTHNb488SpvVdWtHgVKCEKSLiSfsT\npDGr0PLxxx8DUK1atWQ2J6307NkTcDM/wpOmVS5SXczisRoT1od50UUXJbM5EZQwRUQ8KWGmsWuv\nvTbiVuLPKuyEU7IsmebNmwOuD9NqX4JbHVSjRo3ENwwlTBERb0qYIiVgaXLKlClJbkn6sT5M29sH\nXB1Mq8QV7z179kcJU0TEkxKmiKSkiRMnAq5+BEC7du0AlzQTTQlTRMSTEqaIpKSOHTtG3KYCJUwR\nEU9ZhVUEycrKWgesKPAJpV/dvXv3Vk/Wh+v4BkfHNjiZfGwLPWGKiIijS3IREU86YYqIeNIJU0TE\nk06YIiKedMIUEfGkE6aIiCedMEVEPOmEKSLiSSdMERFPOmGKiHgqtFpRdnb23pycnAQ1JfFyc3PJ\ny8vLStbn6/gGR8c2OJl8bAs9Yebk5DBv3rxgWpUCmjVrltTP1/ENjo5tcDL52OqSXETEk06YIiKe\ndMIUEfFU6reouP/++wEYNmwYAAsXLgTg6KOPTlqbRCQ9KWGKiHgqtQlzx44dAMydOxeASpUqAftG\n8CS2DRs2hH4eM2YMAEOHDgXgxBNPBODNN98EoHz58gluXWratWsXAHv27AFg9+7dAHz11VcAfPzx\nxwCMHTs29JoFCxYAMHnyZACuvPJKALKykjaDrVTbuHFj6Of27dsDcOqppwLu7zdRlDBFRDyV2oT5\n5ZdfAjBjxgwAqlatCkCZMvoOiLZlyxYAnnnmmdB99957b8RzZs2aBUCHDh0AeP311xPTuBT1xx9/\nADBlyhQAnnjiCcAlS0uesdjf4K233gpAu3btAKhcuXIwjU0zluLt38ASOri/09NOOy3h7QIlTBER\nbwlJmHl5eQBUqVIFgIoVKxb7vaw/47zzzit5w9KcfVP36NEDgAkTJuz3Ne+//36gbUp1O3fuBODO\nO+8E4B//+IfX6w4++ODQz9a/9thjjwFKlkW1ZMkSAI4//vgktyQ/JUwREU+BJswPPvgAgJtuugmA\nzZs3A64fqEaNGkV+T9tHffv27RH3K3E6liwvv/xyAKZPn+792lT8Vk8kS4X7S5ZNmjQBoH///oBL\nlQC1a9cu0mdu27YNgKVLlwLQqFEjAMqVK1ek90kX/fr1S3YTCqSEKSLiKdCE2aVLFwB++umniPst\nAcVTp06d4v6epZUlyqIkSxOelDKJ9Y1bwoxmMwwuueQSAA455BCgZP2TdpVkc2Jvv/12ANatWwdA\ntWrViv3epZnNgElFSpgiIp4CSZg20vrLL79E3H/NNdcAUL169WK/d3Tfkn0LX3rppcV+z3Tx888/\nA3DzzTcX+z1Gjx4NwD333ANArVq1St6wUsD6C2vWrAnA+vXrIx63FT1t2rQBSpYsbZXatddeC8DU\nqVOL/V7pyGZzFDYuYX+n3bt3BxK3wk8JU0TEUyAJs2/fvkD+vsozzjhj34ceUPSPtXXQtuLC2HuV\nZG5nuvjoo48AWLt2bczHGzRoEPp51KhRgJsje8sttwCuL69p06aA6086/PDDA2hx6rDE+NprrwFu\npNpW9IwfPx6ASZMmAdC5c2cAHnroIQCOOuqo0HtFrxn//vvvAbjvvvsAt4pq69atEc+3/mOri5Cp\nLOUXxq4AomfLBE0JU0TEU1wTpvVdfv311xH316lTB4CuXbsW+72XL18O5O9bsm/tTGbz9wrqu2zY\nsCHgEijAoYceGvEcS1I2s2HNmjUR753uCdNYHVX77z/nnHMA+OabbwC3Euhf//pXxO0NN9wQeo8B\nAwYArvLTHXfcAbg1/caujrp16wZErvXPZJbeU5ESpoiIp7gmzNmzZwP5+y6thl34etuislGxaHXr\n1i32e5Z2lnauuuoqADZt2hTzeTazIDpVhiuo38jmBGYam2dp1XGeeuopAJ5++mkAVq9eHfH8cePG\nxfw5Fqs1+uyzzwJw3XXXlbS5acWulHr16pXkluSnhCki4ikuCdOSTUHVj60uYBAWLVoEuDXQmbSX\nz7vvvgvA/PnzYz5uNQN99rC2OW82km4VYx588EEA2rZtC2TebARLmlY/1NJP8+bNAfjxxx+938tG\nv+fMmQPAn/70p7i1M5341IWwv1P790kUJUwREU9xSZgjRowA3LwyY/1iRx55ZJHf0+YH2je4zY+L\ndtdddwHwyCOPAPlH0dOZrUGOZqOvM2fOBOCggw7yfs+//vWvAPTu3RuA7777DnCj5bb3T6azqllF\nYevQlSwLZ5XWC2PjIomevaGEKSLiSSdMERFPcbkkf/jhh2Pe/9577wH5LwntctumxdiUodzc3NBz\n3nrrLSD/9A1jl0S2rCx8aVqmO/PMM4GiXYpL4WxJ7t///nfATWwvCvublsLZAGNhFi5cCLii5In6\nW1fCFBHxFGgB4bJlywKuVJgV0LACBvZ7cdjgg03+PfbYY4v9XqXN77//Drhv2Wjh25LGy6effgpk\n3qDP8OHDAbdtwp49eyIejy7CAW4gokWLFoAbOBM/Vv7RyhXGYoPAq1atAtzy36ApYYqIeAo0YR53\n3HFFen74pGjbIC16YrBNmbFkad/imeTXX38FCk6YtulcUVgJMisUEc1nqkc6sb7zRx99FMifLG36\nlSXQ6JJuAE8++SQA559/PuAWeFg6UtHr2F588UWg8A35bOFAcTZSLAklTBERT3FJmFYAw1KK76Re\nS5QXXHABENkPZIUi6tWrF/Ea+z0Tk6WJLpocXUTVyuvZ0sjC2EwFK0VmZd6iZyG0atWqpM0uFaxf\n3f6+bJsVKxxz/fXXAy5ZFsb+v7AlkbawIzqtSmzRf4PhTjrpJACqVKmS0DYpYYqIeIpLwrRCDbZM\n0YqqFsQKOti3eOPGjfM9Z8WKFTFfa6Xjtm3bBmRmOX9bamrz1aI30Zo7dy5QeMK0ftA+ffoAbuMp\nY9/qrVu3BvIn/XRlW3RE951bYRnbysPHDz/8ALhkWabMvnxSkk0AM0msZGls7rb9/WqUXEQkxcR1\nlNy+hQsq8xYPtlWFFfawTbqOOeaYwD4zVVnh2eiEacVuv/rqK8CNfE+cODHfc3777beY7239oyNH\njgQS31eUajp16uT9XEuUdoyNHUPbDFCKr2fPngDUr18/oZ+rhCki4inQeZhBstUuixcvBjIzYbZp\n0wZw89VsRYkVVbbb5557zvs9bXWWzRVMVN9QqrA1ybZax+Zj2gZ/2dnZMV+3cuXK0M+DBg0C8q/w\nadKkSXwbm8EsrRdny+6SUMIUEfGUsgmzWrVqgBtRtzmadr9tSepT2SRdWT+jbZ9r/Tq2UqIobER9\n+vTpgDvOmcb+u20lj23jfPnllxf7Pe3YvvHGGyVsnVx44YWAS/GJpoQpIuIpZROm9SVZwrRbyc9W\nodj2rjYqbvVIrTqUzc8EuO222wCoU6cOAN27dwegcuXKwTe4FLDqRDbPz+ZUWuWtgrZMifUe999/\nP5CZc4aLo3bt2oDbhiJ8+2j7Wy5XrlziG4YSpoiIt5RNmFJ0NmJoo7F227dv36S1qbSy2QJ226hR\nIwBefvnlpLUpU9gI+JQpU5LckvyUMEVEPOmEKSLiSSdMERFPOmGKiHjSCVNExJNOmCIinnTCFBHx\nlFXY/jtZWVnrgNilz9ND3b179yat/LWOb3B0bIOTyce20BOmiIg4uiQXEfGkE6aIiCedMEVEPOmE\nKSLiSSdMERFPOmGKiHjSCVNExJNOmCIinnTCFBHxVOgWFdnZ2XtzcnIS1JTEy83NJS8vLytZn6/j\nGxwd2+Bk8rEt9ISZk5PDvHnzgmlVCmjWrFlSP1/HNzg6tsHJ5GOrS3IREU86YYqIeNIJU0TEk06Y\nIiKedMIUEfFU6Ch5Kvrjjz8A+OSTTwDo1q0bAMuXLwfACiK//PLLAFx++eWJbqKIJMm2bdsAKFu2\nLACrVq0C9o3sA2RllWwmlhKmiIinlE+Ylhjnzp0LwKOPPgrAlClTIp5n3xzlypUDoGLFiolqopRi\nv/32GwCvv/56xP0vvvgiADNmzPB6n/CtXv7nf/4HgHPOOQeALl26AC71SOH27NkDwK5duwB3NQmw\ndu1aAF544QUAOnXqBMDSpUsBmDBhAgDly5cHYNGiRQDcdNNNgDtPPPvss8VqmxKmiIinlE+YX3zx\nBQAtWrSI+Xj16vs2d+vZsycADRo0AODCCy9MQOuktBoyZAgAw4YNA2Djxo0xn2eJxFZ/LFy4EIDm\nzZsD8O6770Y8D+Bf//pXxO0777wDwD//+U8AKlSoEKf/ivRiaf/OO+8E4Pnnnwegdu3aoeesXLkS\ncGn9888/B+Cggw6KuK1fvz4ABx54IADr1q0D4JprrilRG5UwRUQ8pWzCbN++PRDZfwHum8W+QcaP\nHw/AKaecksDWZY7ff/8dgLy8PMD1Ib311lsRz9uxYwcAr776KgDz589PVBOLxFKMJcsqVaoA0Lt3\nb8BdqUSrXLky4P47K1WqBLjjE27w4MEAjBgxAnD9asOHDweUMAty++23A+549erVC3DHE2D79u0A\nlCmzL+vZmIX9brd2jO3fy/o0NUouIpIgKZUwrb8SYPr06YAbKTP9+/cHYMCAAQlrV2m1ZMkSALZs\n2QJAnTp1Qo/NmjULgDVr1gCubyg3NxeAb7/9FoBff/0VgNWrV8f8DPsmf+ihh4D8VwSpxkbDrc/S\n5uk++OCDXq+3PjFjVzyLFy8O3Tdt2rSI59SqVQuAAw5Iqf/dUo6NdNucSfs3sXQf/bOPeKd5JUwR\nEU8p8ZVn3/IfffRR6D5LltYn8be//Q2APn36JLh1pc/PP/8MQMOGDUv8Xoceeijg5hSee+65gJud\ncNlllwFwxBFHlPizEqFt27YAPPbYYwC0bNmySK+3PjFLlPZ3+dJLLxX4mvfffx9wI7gS6emnnwbg\nq6++Alz/d7Vq1ZLWpoIoYYqIeEpqwrQ5VO+99x7g+trCWR9Ev379EtewUsqSZXTF6O7duwMuFQJc\nccUVEc+xlSrRo4jHHnsskD4rp7KzswE3IuvL0o+N2BaWKHv06AHARRddBEDdunWL3M5MYP3i//73\nvwGoWbMm4OZM2m3Tpk1Dr0n236ESpoiIp6QkzA8//BBw/UmxkqV55ZVXEtKm0szmAnbt2hVw39w2\nV/KCCy5ITsPSwIIFCwA47bTTANi9e/d+X7N+/XrAzTgIX2currKYHVObG2ui+5Vtriy4EXT7d7AZ\nCXYlFDQlTBERT0lJmNZ/VFiyfO2114Dk7+yYynbu3Am4PjObW2nrZZUsS85mGgwaNAhwszcsPY4d\nOzbfa/7zn/9E3NpacptX2Lhx4+AaXArYzILoZGkzMDp37gy4mRmx3HvvvQCcccYZAHz66aeAWwEY\nFCVMERFPCU2Y9m1sI47Rwr9PK0TtAAAO70lEQVRRrNqQaggWbNmyZYAbZbR5a7FSjxSPzQO+6667\nIu63pPn444/ne43N+rD16ra+fubMmYDbDeD8888PoMWpb/To0YCrE2qj4FWrVgX81ntbPcwjjzwS\ncDUlbMVZUJQwRUQ8JSRhbtq0CYBnnnkGyD9qePbZZwMwderU0H1KlgXbsGED4FbZ2LGyGo82gljY\nXEHrLzrssMMCa2c6s3XhsVbv2L9L69atAbj44osBt5LtyiuvBNwuAlbDNVNYaj/vvPOK/R42P/uq\nq64C3PxYq1NqM3DiTQlTRMSTTpgiIp4SckluGxWFl28LZ53i4RNUpWDWaW4TgI866igAvvzySwBu\nueUWoODljuAuaayIRHjpN4kPKwVnJeX+/Oc/AzB79mzAbcz11FNPAXDCCSckuomlnk1RtO482wYk\nqEFjJUwREU+BJkzr1P7ss89iPn7//fcDrty/+LHpROann34C3KCapUUrtnzccccB8MADD4Re88EH\nHwBuGZ8SZnAOPvhgwE1BssnWtkTYJnJPnDgxCa0r3exv265O7WrVyvAVteDw/ihhioh4CiRh2qRe\n60+waTDR7NuhJP0MNknYSptZ2bJkl4EK0tChQwFo06ZNxP12PK0QgW38ZGwTLnCbxln6P/nkk4Np\nrITYksiRI0cCrq/ZCgynG5tOaAk7EWw7kKCmJSphioh4CjRh7m+JnvWx2ehhrL5MG+m17V1tgqqN\nCFv5rYEDBwJu2810dsghhwD5iwDvT6x+ykceeQRwaUeC16pVq2Q3IVAdO3YEXD+5zQqwv794bkz2\n9ddfA/Djjz8CcP3118f9M8IpYYqIeErqFhX2rWBb6sZKTFayrKBCBf/4xz8AV+JMCvbHH3/ku+/E\nE09MQksknVn/oRWytr71U089FXB9uFZ0oyTbD48aNQpwy4Hr1atX7PfyoYQpIuIpJbbZtX4H69MM\nZ32XtpjeSpiNGDEC0NalRRErhduKiEy3ZMkSAMaMGQPA1VdfDcS3gPXmzZsBuO222+L2nqlo3Lhx\nAAwYMABwGxjadjOnn3464Aq/2NYqVji4sCLANm/40ksvBeCHH34A3EZzQV9pKmGKiHgKJGFan4Sd\n7W0FSkG2bdsGRK5gsY2QnnjiCSC4ck2ZwArZhm8oZ7MJrBRZprMCwbZSxPrG5s2bB7jtee0Kpygs\nWdpKHiskbNKtH9nm/9p8YNuqw/7+Pv74Y8ClebtatNuisDoK9p7F+fcpCiVMERFPgSZMW1li8wZt\nDqUP2yI2nVfslJStoNq6dSvgCrNu374dcHMsbT5s+DxXO75aQ77PY489BsARRxwBuOLLjRo1Alxa\nsq1hi8L6R+fMmRNxvxXQnTx5cjFaXHrY/8M2P9Nu7Xxgs2SsapOl+libJNo2u3fffTcA7du3B6B6\n9epBND0fJUwREU+BjpJb0rQqOTYia31qNnoWS/Q6aMnPNpOzfkibi2Z9wtF69uwZ+tn6iGWfo48+\nGnApp1evXoDbZsLmFE6YMKHYn2FJy1b6WLJM5FrrVGKVhGzLDrtds2YNEPvvuEaNGhGvTTQlTBER\nTwmZh2lp0b5Z030tbaLYtsSvvfYaAG+//TbgKrLb5ls2z1Wp0l/Dhg0BWLVqFQArV64E8lcWmjZt\nGgDr1q0L3Wd1Lo1dWVm9g3jO7UxHNWvWTHYTCqSEKSLiKSVW+kjJ2DbFdjto0KBkNict1a5dG4Au\nXbpE3B/9u6Q3JUwREU86YYqIeNIJU0TEk06YIiKedMIUEfGUZXvmxHwwK2sdsCJxzUm4unv37k3M\nItQYdHyDo2MbnEw+toWeMEVExNEluYiIJ50wRUQ86YQpIuJJJ0wREU86YYqIeNIJU0TEk06YIiKe\ndMIUEfGkE6aIiCedMEVEPBVacT07O3uv7QOcjnJzc8nLy8tK1ufr+AZHxzY4mXxsCz1h5uTkhDZV\nT0fJ3oxKxzc4OrbByeRjq0tyERFPOmGKiHjSCVNExJO22c0Av/32GwBDhgwJ3ffoo48CYPVQrU/q\nlFNOSXDrREoPJUwREU9KmGnsiy++AKBFixYA7Nq1K/RYVlZWxK2I7J8SpoiIJyXMNJSXlwdA586d\nAZcse/ToEXpOy5YtAejQoQMABxygP4Vwv//+OwBLliwBoH///gDMmDEDcH2/ltAbN24ceu3NN98M\nRB5vKdiGDRsAeO655wD46aefABg9ejQAV155Zei5J554IgB33HFHIpsYooQpIuJJsSINWWpctmwZ\nAPXq1QNg0KBBoecceuihiW9YKdK6dWuAfCtaCur7/frrr0M/9+vXD4CTTjoJgDPPPDOwdpZmzzzz\nDACPPfYYAD/88AOQP72PHz8+9JpKlSoBsHv3bgB69+4NQIUKFRLQYiVMERFvgSbMVatWAdCwYUMA\nDjzwQADmzp0LQO3atYP8+Iy1efPmiN/tW1ipcv+uu+46IH+yLFNmX7Zo1KgRADfddBMADz30EOD6\n4QC2bt0KwKRJkwAlzGhTp04FXL/wpk2bIh63/uALL7wQiDxPPPDAAwDcfffdANx6662AEqaISMoJ\nNGHOnj0bgC1btkTc2rzAzz77DIAjjzyy2J+xceNGAKZPnw64keFMNGfOHAAWLFgAQMeOHQGN1hbF\njh07Yt4/YsQIAHr27Blxv11F2cqpcA0aNIhz60q39evXA9CnTx/AJUtLmuauu+4CoEqVKvnew64A\nVqxYEVg7C6OEKSLiKZCEaUnyb3/7W8T95cqVA+CXX34B3Chi9+7dAffNAnDQQQcBBa9E2bNnD+BS\nlI2wZXLC/PDDDwE379JGFDXH0p+N2J5//vkA1K9fH4Czzz7b+z0OP/xwwKUh2cdmEvz4448AHHLI\nIQD07dsX8Otjt9RZp04dAFauXAlA1apV49vYAihhioh4CiR6DB48GICFCxcCrs/Cvr1tBNJm69vz\n7Rbg1VdfBeCcc84B3MoLWx9t77l06dKI52Ui68e142tOOOGEZDSnVDvqqKMAt1qnINYfN2XKlHyP\nHXHEEUDiUk9pY1eNtmqnOLM37Aq0oD7noChhioh4imvCtNn3NjPf+hkefvjhiOfZnhnvvvsuAE8+\n+SQAjzzySOg5l112GQB169YF9j8qNmzYsBK1vTT79NNPAVi9enXE/Z06dUpGczKC/a1a33m4Jk2a\nAK6f3eZwSiRLh3beKFu2bJHf409/+lOJ36Mo9C8pIuIprglz+fLlgBu5uvHGGwE3WhvNvg2sP9Jm\n7QOMGzcOgO3bt0e8xlaxDBgwAIBWrVoB0LRp05I2P+3Mnz8fcH2c4VujVqxYMRlNKrUsLT799NMA\njB07tsDnWtUdO+41atQA3Oqgk08+ObB2liY2lmGzaWwVT1H06tULcH3PxXmPolDCFBHxFNeE2a5d\nu4jfL7nkkiK9vnLlyqGfo1dUGKuVZ2t4re9S/UT5tW3bNuL3s846K/SzJXRbjWIjuxLb999/D7h1\n+T5eeumliN8nTJgAwEcffQS4/rdMYTNZ/v73vwNw5513Au5v0W7t8fbt2wNuLARg0aJFALz//vsA\nvPjii4Bbv3/FFVcAcPzxxwfxn6CEKSLiK5B5mDY6a9VG4smSZa1atYDM+5YuCftWBjjvvPMAt9rC\n1uI3b9488Q0rBSwVFqRmzZqhny3dWB+drZm2W1s1ZH39mdafbGMb06ZNA+CTTz6JeNySp60xz87O\nDj1mM0Gia2Ymam8qJUwREU86YYqIeIrrJbldLp9xxhlAfC817HLy+eefB2D48OEAHHzwwXH7jHQz\nefJkwE0nGjlyZOixzz//HHCl4GyA6K9//Svg/i1lHxuAsEEHc/HFFwORU4Wsm2PdunWAW/JrJeJ+\n++03wA0C2XSjTGHH57333gPc39z//d//RTzPivREL8hIJiVMERFPcU2YQZRWs+VTNlHYytXbt5K4\nb+zy5csDsHPnTsANPtgWIbYYANw0jG7dugHwv//7vxG31uGeqNL/qc4KaQwZMsT7NdWrVwdcWre/\nYVt88fbbbwOZlzCNlXu0ojHPPvtsxON21fPWW2+F7rv66qsBVypu8eLFgbcznBKmiIinlK8sO3To\nUABeeOEFIHJTd9nHpgINHDgQcBtE2QRgW6oXzhYJWJ/w2rVrAbetyF/+8hcAxowZE1SzM4aVIrOF\nHBMnTkxmc1JOrK0owE11i8VSqU1BMva3/vjjj8epdZGUMEVEPKVswrSJqbm5uYCb7Gsji5KfTQi2\n0fAPPvgAgJ9//hmIvfzRCqPYSLolzP1N1BZJJlsU069fv4j7bUuRoChhioh4StmEaaNgNrJrc9hs\nJFjyq1atGuCWi1qBZiu6YYkT3NJSkdLINpo78MADAbfxohVJCYoSpoiIp5RNmK1btwZcv5u2LPX3\n5ptvAu4Yzpo1C4CWLVuGnmPzL41tc2ESVcwgVdlcVuvLtUIZxfk7/PXXXwFXbMIUVFhb9m/btm2A\n21I6UZQwRUQ8pVTCDF+/bOX9bS6WtiwtOlu1Y8lyyZIlocf2V8o/0+e73nfffYCb72d958cccwwA\nZ555pvd7DRo0CHDbrdh72WoqKTo7P0RvYRP+Nx4EJUwREU8pkTDz8vKAyGo6tjGaCtoWn42aW0Wi\nd955J/SYjaDbWn2b72oK2iIkU0yaNCnid+vTtPl/Nt8veh5gLNGrpexKyrYHkaKzTc/sytOKM9uK\ntaAoYYqIeEqJhGm1Ga1OIKh/Jwg2ah79s+Q3atQoAO6//34AvvzyS8CNzn7zzTcAXHvttd7v2aVL\nF8BdPUnxWYUuW6dvCdO2BbF/ryZNmsT1c5UwRUQ8JTVh2kiX1b0L78PUqLgkk1VSt83i5syZA7j5\nmLbdbvhVUUFef/31iPeyOpBSch06dAAizx3gzi3xpoQpIuIpqQnT6gJaZaJevXqFHrPKO5m2Bamk\nFqs436pVq4j7r7nmmmQ0R6LY3l52GzQlTBERT0lNmM2aNYv4vUePHqGfy5Ytm+jmiIgUSglTRMRT\nSiTMPXv2JLMZIiJelDBFRDxl2Qh1zAezstYBKxLXnISru3fv3urJ+nAd3+Do2AYnk49toSdMERFx\ndEkuIuJJJ0wREU86YYqIeNIJU0TEk06YIiKe/h9qyK7LULyJHQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x432 with 20 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8rTHW2ZPZMSi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# construct model on cuda if available\n",
        "\n",
        "use_cuda = torch.cuda.is_available()\n",
        "\n",
        "class CNNClassifier(nn.Module):\n",
        "    \n",
        "    def __init__(self):\n",
        "        # 항상 torch.nn.Module을 상속받고 시작\n",
        "        super(CNNClassifier, self).__init__()\n",
        "        conv1 = nn.Conv2d(1, 6, 5, 1) # 6@24*24\n",
        "        # activation ReLU\n",
        "        pool1 = nn.MaxPool2d(2) # 6@12*12\n",
        "        conv2 = nn.Conv2d(6, 16, 5, 1) # 16@8*8\n",
        "        # activation ReLU\n",
        "        pool2 = nn.MaxPool2d(2) # 16@4*4\n",
        "        \n",
        "        self.conv_module = nn.Sequential(\n",
        "            conv1,\n",
        "            nn.ReLU(),\n",
        "            pool1,\n",
        "            conv2,\n",
        "            nn.ReLU(),\n",
        "            pool2\n",
        "        )\n",
        "        \n",
        "        fc1 = nn.Linear(16*4*4, 120)\n",
        "        # activation ReLU\n",
        "        fc2 = nn.Linear(120, 84)\n",
        "        # activation ReLU\n",
        "        fc3 = nn.Linear(84, 10)\n",
        "\n",
        "        self.fc_module = nn.Sequential(\n",
        "            fc1,\n",
        "            nn.ReLU(),\n",
        "            fc2,\n",
        "            nn.ReLU(),\n",
        "            fc3\n",
        "        )\n",
        "        \n",
        "        # gpu로 할당\n",
        "        if use_cuda:\n",
        "            self.conv_module = self.conv_module.cuda()\n",
        "            self.fc_module = self.fc_module.cuda()\n",
        "        \n",
        "    def forward(self, x):\n",
        "        out = self.conv_module(x) # @16*4*4\n",
        "        # make linear\n",
        "        dim = 1\n",
        "        for d in out.size()[1:]: #16, 4, 4\n",
        "            dim = dim * d\n",
        "        out = out.view(-1, dim)\n",
        "        out = self.fc_module(out)\n",
        "        return F.softmax(out, dim=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9LbJf2GpZT2Z",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "cnn = CNNClassifier()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "19HI4JuYZdBj",
        "colab_type": "text"
      },
      "source": [
        "Training model\n",
        "What do we need? \n",
        "\n",
        "  1.loss ex) MSELoss\n",
        "  2.backpropagation method ex) SGD ... \n",
        "  2.1 update할 parameter들 \n",
        "  2.2 learning rate\n",
        "\n",
        "  .3hyperparamters ex) num_epochs ..."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Vgj0cRXSZWD5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# loss\n",
        "criterion = nn.CrossEntropyLoss()\n",
        "# backpropagation method\n",
        "learning_rate = 1e-3\n",
        "optimizer = optim.Adam(cnn.parameters(), lr=learning_rate)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I6lgUynVZoBo",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# hyper-parameters\n",
        "num_epochs = 2"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1D4qzuvkZpdI",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "5a3bc80a-bc65-4312-818e-223cf2d323b3"
      },
      "source": [
        "num_batches"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "938"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3q9k_fTjZqpI",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "cda59ff2-3cdd-4c6b-cc7e-6b7a7a5a8f56"
      },
      "source": [
        "batch_size\n"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ONG9kCsnZslQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 601
        },
        "outputId": "e513b606-10bc-48bf-be4b-4d67f28173ed"
      },
      "source": [
        "# create figure for plotting\n",
        "row_num = 2\n",
        "col_num = 4\n",
        "fig, ax = plt.subplots(row_num, col_num, figsize=(6,6))\n",
        "for i, j in itertools.product(range(row_num), range(col_num)):\n",
        "    ax[i,j].get_xaxis().set_visible(False)\n",
        "    ax[i,j].get_yaxis().set_visible(False) \n",
        "    \n",
        "trn_loss_list = []\n",
        "val_loss_list = []\n",
        "\n",
        "for epoch in range(num_epochs):\n",
        "    trn_loss = 0.0\n",
        "    for i, data in enumerate(trn_loader):\n",
        "        x, label = data\n",
        "        if use_cuda:\n",
        "            x = x.cuda()\n",
        "            label = label.cuda()\n",
        "        # grad init\n",
        "        optimizer.zero_grad()\n",
        "        # forward propagation\n",
        "        model_output = cnn(x)\n",
        "        # calculate loss\n",
        "        loss = criterion(model_output, label)\n",
        "        # back propagation \n",
        "        loss.backward()\n",
        "        # weight update\n",
        "        optimizer.step()\n",
        "        \n",
        "        # trn_loss summary\n",
        "        trn_loss += loss.item()\n",
        "        # del (memory issue)\n",
        "        del loss\n",
        "        del model_output\n",
        "        \n",
        "        # 학습과정 출력\n",
        "        if (i+1) % 100 == 0: # every 100 mini-batches\n",
        "            with torch.no_grad(): # very very very very important!!!\n",
        "                val_loss = 0.0\n",
        "                for j, val in enumerate(val_loader):\n",
        "                    val_x, val_label = val\n",
        "                    if use_cuda:\n",
        "                        val_x = val_x.cuda()\n",
        "                        val_label =val_label.cuda()\n",
        "                    val_output = cnn(val_x)\n",
        "                    v_loss = criterion(val_output, val_label)\n",
        "                    val_loss += v_loss\n",
        "\n",
        "            # draw last val dataset\n",
        "            for k in range(row_num*col_num):\n",
        "                ii = k//col_num\n",
        "                jj = k%col_num\n",
        "                ax[ii,jj].cla() # clear the current axis\n",
        "                ax[ii,jj].imshow(val_x[k,:].data.cpu().numpy().reshape(28,28), cmap='Greys')\n",
        "            \n",
        "            display.clear_output(wait=True)\n",
        "            display.display(plt.gcf()) # get a reference to a current figure\n",
        "                \n",
        "            print(\"label: {}\".format(val_label[:row_num*col_num]))\n",
        "            print(\"prediction: {}\".format(val_output.argmax(dim=1)[:row_num*col_num]))\n",
        "            del val_output\n",
        "            del v_loss\n",
        "            \n",
        "            print(\"epoch: {}/{} | step: {}/{} | trn loss: {:.4f} | val loss: {:.4f}\".format(\n",
        "                epoch+1, num_epochs, i+1, num_batches, trn_loss / 100, val_loss / len(val_loader)\n",
        "            ))            \n",
        "            \n",
        "            trn_loss_list.append(trn_loss/100)\n",
        "            val_loss_list.append(val_loss/len(val_loader))\n",
        "            trn_loss = 0.0"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAV0AAAEJCAYAAAA3l/RWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIAYfksLuwAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x432 with 8 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "label: tensor([0, 3, 6, 9, 9, 3, 7, 7], device='cuda:0')\n",
            "prediction: tensor([0, 3, 6, 9, 9, 3, 2, 7], device='cuda:0')\n",
            "epoch: 2/2 | step: 900/938 | trn loss: 1.4890 | val loss: 1.4974\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAV0AAAEJCAYAAAA3l/RWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIAYfksLuwAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x432 with 8 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B6brOBdkZwJg",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "outputId": "3e45f1cc-fbf0-491e-a07b-fb4a3e1b86f5"
      },
      "source": [
        "plt.figure(figsize=(16,9))\n",
        "x_range = range(len(trn_loss_list))\n",
        "plt.plot(x_range, trn_loss_list, label=\"trn\")\n",
        "plt.plot(x_range, val_loss_list, label=\"val\")\n",
        "plt.legend()\n",
        "plt.xlabel(\"training steps\")\n",
        "plt.ylabel(\"loss\")"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0, 0.5, 'loss')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA7MAAAIaCAYAAADyehr9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n5zjT9M7yAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 1152x648 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6xwFgmLmaBAX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}