pytorch-cnn-digit-recognizer-gpu.ipynb

{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n","execution_count":1,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Introduction\n\n\n<font color = 'red'>\nContent: \n\n1. [Import Libraries](#1)\n1. [Select Device](#2)\n1. [Import Dataset](#3)\n1. [Read Data](#4)\n1. [Data Process](#5)\n    * [Transforms](#6)\n    * [Data Loader](#7)\n1. [Visualization](#8)\n1. [CNN Model](#9)\n    * [Create Model](#10)\n    * [Initial Parameters](#11)\n1. [Training](#12)\n1. [Testing](#13)\n1. [Visualization Results](#14)\n1. [Save Model](#15)\n    "},{"metadata":{},"cell_type":"markdown","source":"<a id = \"1\"></a><br>\n\n# Import Libraries\n"},{"metadata":{"trusted":true},"cell_type":"code","source":"import pandas as pd\nimport numpy as np\nimport torch\nimport torchvision\nimport torch.nn as nn\nimport torch.nn.functional as F\nimport torch.optim as optim\nfrom torchvision.transforms import transforms\nfrom torch.utils.data import DataLoader\nfrom torch.utils.data import Dataset\nfrom torchvision.utils import make_grid\nimport numpy as np\nimport matplotlib.pyplot as plt\nfrom torch.autograd import Variable\nfrom torch.utils.tensorboard import SummaryWriter\nimport time","execution_count":2,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"We initialize constant variable. if we want to change variables, we can do there easily."},{"metadata":{"trusted":true},"cell_type":"code","source":"parser = argparse.ArgumentParser(description='PyTorch MNIST Example')\n    parser.add_argument('--batch-size', type=int, default=128, metavar='N',\n                        help='input batch size for training (default: 64)')\n    parser.add_argument('--epochs', type=int, default=14, metavar='N',\n                        help='number of epochs to train (default: 14)')\n    parser.add_argument('--lr', type=float, default=1.0, metavar='LR',\n                        help='learning rate (default: 1.0)')\n    parser.add_argument('--seed', type=int, default=1, metavar='S',\n                        help='random seed (default: 1)')\n    parser.add_argument('--save-model', action='store_true', default=\"/save_model.pth\",\n                        help='For Saving the current Model')\n    args = parser.parse_args()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"2\"></a><br>\n\n# Select Device"},{"metadata":{},"cell_type":"markdown","source":"In this chapter we will use select device. There are two devices: GPU and CPU. We will choose GPU devices because GPU much faster than CPU."},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_device():\n    if torch.cuda.is_available():\n        device = 'cuda:0'\n    else:\n        device = 'cpu'\n    return device\ndevice = get_device()","execution_count":3,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Do not worry. If your devices is not available to using GPU, This function select CPU devices."},{"metadata":{},"cell_type":"markdown","source":"<a id = \"3\"></a><br>\n\n# Import Dataset"},{"metadata":{},"cell_type":"markdown","source":"we will import MNIST dataset. We will create funtion. This function is for custom dataset. Also the function is for csv file."},{"metadata":{"trusted":true},"cell_type":"code","source":"class MNISTDataset(Dataset):\n    def __init__(self, images, labels=None, transforms=None):\n        self.X = images\n        self.y = labels\n        self.transforms = transforms\n\n    def __len__(self):\n        return (len(self.X))\n\n    def __getitem__(self, i):\n        data = self.X.iloc[i, :]\n        data = np.asarray(data).astype(np.uint8).reshape(28, 28, 1)\n\n        if self.transforms:\n            data = self.transforms(data)\n\n        if self.y is not None:\n            return (data, self.y[i])\n        else:\n            return data","execution_count":4,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"4\"></a><br>\n\n# Read Data"},{"metadata":{},"cell_type":"markdown","source":"We will use two csv files. these files for train and test. But this dataset is not include validation set. If you want to use validation set, you can split training set."},{"metadata":{"trusted":true},"cell_type":"code","source":"df_train = pd.read_csv('../input/digit-recognizer/train.csv')\ndf_test = pd.read_csv('../input/digit-recognizer//test.csv')","execution_count":9,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"df_train.describe()","execution_count":21,"outputs":[{"output_type":"execute_result","execution_count":21,"data":{"text/plain":"              label   pixel0   pixel1   pixel2   pixel3   pixel4   pixel5  \\\ncount  42000.000000  42000.0  42000.0  42000.0  42000.0  42000.0  42000.0   \nmean       4.456643      0.0      0.0      0.0      0.0      0.0      0.0   \nstd        2.887730      0.0      0.0      0.0      0.0      0.0      0.0   \nmin        0.000000      0.0      0.0      0.0      0.0      0.0      0.0   \n25%        2.000000      0.0      0.0      0.0      0.0      0.0      0.0   \n50%        4.000000      0.0      0.0      0.0      0.0      0.0      0.0   \n75%        7.000000      0.0      0.0      0.0      0.0      0.0      0.0   \nmax        9.000000      0.0      0.0      0.0      0.0      0.0      0.0   \n\n        pixel6   pixel7   pixel8  ...      pixel774      pixel775  \\\ncount  42000.0  42000.0  42000.0  ...  42000.000000  42000.000000   \nmean       0.0      0.0      0.0  ...      0.219286      0.117095   \nstd        0.0      0.0      0.0  ...      6.312890      4.633819   \nmin        0.0      0.0      0.0  ...      0.000000      0.000000   \n25%        0.0      0.0      0.0  ...      0.000000      0.000000   \n50%        0.0      0.0      0.0  ...      0.000000      0.000000   \n75%        0.0      0.0      0.0  ...      0.000000      0.000000   \nmax        0.0      0.0      0.0  ...    254.000000    254.000000   \n\n           pixel776     pixel777      pixel778      pixel779  pixel780  \\\ncount  42000.000000  42000.00000  42000.000000  42000.000000   42000.0   \nmean       0.059024      0.02019      0.017238      0.002857       0.0   \nstd        3.274488      1.75987      1.894498      0.414264       0.0   \nmin        0.000000      0.00000      0.000000      0.000000       0.0   \n25%        0.000000      0.00000      0.000000      0.000000       0.0   \n50%        0.000000      0.00000      0.000000      0.000000       0.0   \n75%        0.000000      0.00000      0.000000      0.000000       0.0   \nmax      253.000000    253.00000    254.000000     62.000000       0.0   \n\n       pixel781  pixel782  pixel783  \ncount   42000.0   42000.0   42000.0  \nmean        0.0       0.0       0.0  \nstd         0.0       0.0       0.0  \nmin         0.0       0.0       0.0  \n25%         0.0       0.0       0.0  \n50%         0.0       0.0       0.0  \n75%         0.0       0.0       0.0  \nmax         0.0       0.0       0.0  \n\n[8 rows x 785 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>label</th>\n      <th>pixel0</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>...</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>count</th>\n      <td>42000.000000</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>...</td>\n      <td>42000.000000</td>\n      <td>42000.000000</td>\n      <td>42000.000000</td>\n      <td>42000.00000</td>\n      <td>42000.000000</td>\n      <td>42000.000000</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n      <td>42000.0</td>\n    </tr>\n    <tr>\n      <th>mean</th>\n      <td>4.456643</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.219286</td>\n      <td>0.117095</td>\n      <td>0.059024</td>\n      <td>0.02019</td>\n      <td>0.017238</td>\n      <td>0.002857</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>std</th>\n      <td>2.887730</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>6.312890</td>\n      <td>4.633819</td>\n      <td>3.274488</td>\n      <td>1.75987</td>\n      <td>1.894498</td>\n      <td>0.414264</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>min</th>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>25%</th>\n      <td>2.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>50%</th>\n      <td>4.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>75%</th>\n      <td>7.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>max</th>\n      <td>9.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>254.000000</td>\n      <td>254.000000</td>\n      <td>253.000000</td>\n      <td>253.00000</td>\n      <td>254.000000</td>\n      <td>62.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n  </tbody>\n</table>\n<p>8 rows × 785 columns</p>\n</div>"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df_train.head()","execution_count":10,"outputs":[{"output_type":"execute_result","execution_count":10,"data":{"text/plain":"   label  pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  \\\n0      1       0       0       0       0       0       0       0       0   \n1      0       0       0       0       0       0       0       0       0   \n2      1       0       0       0       0       0       0       0       0   \n3      4       0       0       0       0       0       0       0       0   \n4      0       0       0       0       0       0       0       0       0   \n\n   pixel8  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n0       0  ...         0         0         0         0         0         0   \n1       0  ...         0         0         0         0         0         0   \n2       0  ...         0         0         0         0         0         0   \n3       0  ...         0         0         0         0         0         0   \n4       0  ...         0         0         0         0         0         0   \n\n   pixel780  pixel781  pixel782  pixel783  \n0         0         0         0         0  \n1         0         0         0         0  \n2         0         0         0         0  \n3         0         0         0         0  \n4         0         0         0         0  \n\n[5 rows x 785 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>label</th>\n      <th>pixel0</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>...</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>4</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>5 rows × 785 columns</p>\n</div>"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df_test.describe()","execution_count":22,"outputs":[{"output_type":"execute_result","execution_count":22,"data":{"text/plain":"        pixel0   pixel1   pixel2   pixel3   pixel4   pixel5   pixel6   pixel7  \\\ncount  28000.0  28000.0  28000.0  28000.0  28000.0  28000.0  28000.0  28000.0   \nmean       0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \nstd        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \nmin        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \n25%        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \n50%        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \n75%        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \nmax        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \n\n        pixel8   pixel9  ...      pixel774      pixel775      pixel776  \\\ncount  28000.0  28000.0  ...  28000.000000  28000.000000  28000.000000   \nmean       0.0      0.0  ...      0.164607      0.073214      0.028036   \nstd        0.0      0.0  ...      5.473293      3.616811      1.813602   \nmin        0.0      0.0  ...      0.000000      0.000000      0.000000   \n25%        0.0      0.0  ...      0.000000      0.000000      0.000000   \n50%        0.0      0.0  ...      0.000000      0.000000      0.000000   \n75%        0.0      0.0  ...      0.000000      0.000000      0.000000   \nmax        0.0      0.0  ...    253.000000    254.000000    193.000000   \n\n           pixel777      pixel778  pixel779  pixel780  pixel781  pixel782  \\\ncount  28000.000000  28000.000000   28000.0   28000.0   28000.0   28000.0   \nmean       0.011250      0.006536       0.0       0.0       0.0       0.0   \nstd        1.205211      0.807475       0.0       0.0       0.0       0.0   \nmin        0.000000      0.000000       0.0       0.0       0.0       0.0   \n25%        0.000000      0.000000       0.0       0.0       0.0       0.0   \n50%        0.000000      0.000000       0.0       0.0       0.0       0.0   \n75%        0.000000      0.000000       0.0       0.0       0.0       0.0   \nmax      187.000000    119.000000       0.0       0.0       0.0       0.0   \n\n       pixel783  \ncount   28000.0  \nmean        0.0  \nstd         0.0  \nmin         0.0  \n25%         0.0  \n50%         0.0  \n75%         0.0  \nmax         0.0  \n\n[8 rows x 784 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>pixel0</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>pixel9</th>\n      <th>...</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>count</th>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>...</td>\n      <td>28000.000000</td>\n      <td>28000.000000</td>\n      <td>28000.000000</td>\n      <td>28000.000000</td>\n      <td>28000.000000</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n      <td>28000.0</td>\n    </tr>\n    <tr>\n      <th>mean</th>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.164607</td>\n      <td>0.073214</td>\n      <td>0.028036</td>\n      <td>0.011250</td>\n      <td>0.006536</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>std</th>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>5.473293</td>\n      <td>3.616811</td>\n      <td>1.813602</td>\n      <td>1.205211</td>\n      <td>0.807475</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>min</th>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>25%</th>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>50%</th>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>75%</th>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>max</th>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>...</td>\n      <td>253.000000</td>\n      <td>254.000000</td>\n      <td>193.000000</td>\n      <td>187.000000</td>\n      <td>119.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n  </tbody>\n</table>\n<p>8 rows × 784 columns</p>\n</div>"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df_test.head()","execution_count":11,"outputs":[{"output_type":"execute_result","execution_count":11,"data":{"text/plain":"   pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  pixel8  \\\n0       0       0       0       0       0       0       0       0       0   \n1       0       0       0       0       0       0       0       0       0   \n2       0       0       0       0       0       0       0       0       0   \n3       0       0       0       0       0       0       0       0       0   \n4       0       0       0       0       0       0       0       0       0   \n\n   pixel9  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n0       0  ...         0         0         0         0         0         0   \n1       0  ...         0         0         0         0         0         0   \n2       0  ...         0         0         0         0         0         0   \n3       0  ...         0         0         0         0         0         0   \n4       0  ...         0         0         0         0         0         0   \n\n   pixel780  pixel781  pixel782  pixel783  \n0         0         0         0         0  \n1         0         0         0         0  \n2         0         0         0         0  \n3         0         0         0         0  \n4         0         0         0         0  \n\n[5 rows x 784 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>pixel0</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>pixel9</th>\n      <th>...</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>5 rows × 784 columns</p>\n</div>"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"These images size is 28x28  so there are 784 numbers. Each colon respresents one piksel. Training set has label and image's pixsels but test set does not have. "},{"metadata":{"trusted":true},"cell_type":"code","source":"train_labels = df_train.iloc[:, 0]\ntrain_images = df_train.iloc[:, 1:]\ntest_labels = df_test.iloc[:, 0]\ntest_images = df_test.iloc[:, 1:]\n","execution_count":13,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_labels","execution_count":14,"outputs":[{"output_type":"execute_result","execution_count":14,"data":{"text/plain":"0        1\n1        0\n2        1\n3        4\n4        0\n        ..\n41995    0\n41996    1\n41997    7\n41998    6\n41999    9\nName: label, Length: 42000, dtype: int64"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_images.size","execution_count":15,"outputs":[{"output_type":"execute_result","execution_count":15,"data":{"text/plain":"32928000"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_images.shape","execution_count":16,"outputs":[{"output_type":"execute_result","execution_count":16,"data":{"text/plain":"(42000, 784)"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_labels","execution_count":17,"outputs":[{"output_type":"execute_result","execution_count":17,"data":{"text/plain":"0        0\n1        0\n2        0\n3        0\n4        0\n        ..\n27995    0\n27996    0\n27997    0\n27998    0\n27999    0\nName: pixel0, Length: 28000, dtype: int64"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_images.size","execution_count":18,"outputs":[{"output_type":"execute_result","execution_count":18,"data":{"text/plain":"21924000"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_images.shape","execution_count":19,"outputs":[{"output_type":"execute_result","execution_count":19,"data":{"text/plain":"(28000, 783)"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"5\"></a><br>\n\n# Data Process"},{"metadata":{},"cell_type":"markdown","source":"<a id = \"6\"></a><br>\n\n# Transforms"},{"metadata":{},"cell_type":"markdown","source":"We must convert images to tensor."},{"metadata":{"trusted":true},"cell_type":"code","source":"transform = transforms.Compose(\n    [transforms.ToPILImage(),\n     transforms.ToTensor(),\n     transforms.Normalize((0.5, ), (0.5, ))\n])","execution_count":23,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"7\"></a><br>\n\n# Data Loader"},{"metadata":{},"cell_type":"markdown","source":"we applied transorms. In this chapter we will load data."},{"metadata":{"trusted":true},"cell_type":"code","source":"train_data = MNISTDataset(df_train.iloc[:, 1:], train_labels, transform)\ntest_data = MNISTDataset(df_test.iloc[:, 0:], test_labels, transform)","execution_count":25,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"trainloader = DataLoader(train_data, batch_size=args.batch_size, shuffle=True)\ntestloader = DataLoader(test_data, batch_size=args.batch_size, shuffle=True)","execution_count":26,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"We selected batch size is 128. There are  42000 images and our batch size is 128 so Each bath size has got 330 images."},{"metadata":{},"cell_type":"markdown","source":"<a id = \"8\"></a><br>\n\n# Visualization"},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_random_images(num):\n\n    indices = list(range(len(test_data)))\n    np.random.shuffle(indices)\n    idx = indices[:num]\n    from torch.utils.data.sampler import SubsetRandomSampler\n    sampler = SubsetRandomSampler(idx)\n    loader = torch.utils.data.DataLoader(test_data,\n                       sampler=sampler, batch_size=num)\n    dataiter = iter(loader)\n    images, labels = dataiter.next()\n    return images, labels\n\n\nto_pil = transforms.ToPILImage()\nimages, labels = get_random_images(5)\nfig=plt.figure(figsize=(10,10))\nfor ii in range(len(images)):\n    image = to_pil(images[ii])\n    sub = fig.add_subplot(1, len(images), ii+1)\n    plt.axis('off')\n    plt.imshow(image)\nplt.show()","execution_count":27,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 720x720 with 5 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"for images, labels in trainloader:\n    break\nim = make_grid(images,nrow=5)\ninv_normalize = transforms.Normalize(\n    mean= [-0.485/0.225,\n    -0.456/0.224,\n    -0.406/0.225],\n    std= [1/0.225, 1/0.224, 1/0.225])\nim_inv = inv_normalize(im)\nplt.figure(figsize=(8,16))\nplt.imshow(np.transpose(im_inv.numpy(),(1,2,0)))\nplt.show()","execution_count":34,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 576x1152 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"9\"></a><br>\n\n# CNN Model"},{"metadata":{},"cell_type":"markdown","source":"<a id = \"10\"></a><br>\n\n# Create Model"},{"metadata":{},"cell_type":"markdown","source":"We make six layers. we will use dropout function to drop overfitting. "},{"metadata":{"trusted":true},"cell_type":"code","source":"class CNN(nn.Module):\n    def __init__(self):\n        super(CNN, self).__init__()\n        self.layer1 = nn.Sequential(\n            nn.Conv2d(1, 32, kernel_size=3, padding=1),\n            nn.ReLU(),\n            nn.Dropout2d(0.4))\n        self.layer2 = nn.Sequential(\n            nn.Conv2d(32, 32, kernel_size=3, padding=1),\n            nn.Dropout2d(0.4),\n            nn.MaxPool2d(2),\n            nn.ReLU())\n        self.layer3 = nn.Sequential(\n            nn.Conv2d(32, 64, kernel_size=3, padding=1),\n            nn.Dropout2d(0.4),\n            nn.ReLU())\n        self.layer4 = nn.Sequential(\n            nn.Conv2d(64, 64, kernel_size=3, padding=1),\n            nn.Dropout2d(0.4),\n            nn.ReLU(),\n            nn.MaxPool2d(2))\n        self.layer5 = nn.Sequential(\n            nn.Conv2d(64, 128, kernel_size=3, padding=1),\n            nn.Dropout2d(0.4),\n            nn.ReLU())\n        self.layer6 = nn.Sequential(\n            nn.Conv2d(128, 128, kernel_size=3, padding=1),\n            nn.Dropout2d(0.4),\n            nn.MaxPool2d(2),\n            nn.ReLU())\n\n        self.fc1 = nn.Linear(1152, 10)\n\n\n\n    def forward(self, x):\n        out = self.layer1(x)\n        out = self.layer2(out)\n        out = self.layer3(out)\n        out = self.layer4(out)\n        out = self.layer5(out)\n        out = self.layer6(out)\n        out = out.view(out.size(0), -1)\n        out = F.softmax(self.fc1(out))\n        return out\n\n\ncnn = CNN().to(device)","execution_count":35,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"11\"></a><br>\n\n# Initial Parameters"},{"metadata":{"trusted":true},"cell_type":"code","source":"random_seed = 1\ntorch.backends.cudnn.enabled = False\ntorch.manual_seed(args.seed)\n\nexamples = enumerate(testloader)\nbatch_idx, (example_data, example_targets) = next(examples)\n\ncriterion = nn.CrossEntropyLoss()\noptimizer = torch.optim.Adam(cnn.parameters(), lr=args.lr)\n\nepochs = args.epochs","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"12\"></a><br>\n\n# Training"},{"metadata":{"trusted":true},"cell_type":"code","source":"trainloss = []\n\n\ndef train(cnn):\n    start = time.time()\n\n    for epoch in range(epochs):\n        running_loss = 0\n\n        for i, data in enumerate(trainloader, 0):\n            inputs, labels = data[0].to(device), data[1].to(device)\n\n            optimizer.zero_grad()\n\n            outputs = cnn(inputs)\n            loss = criterion(outputs, labels)\n            loss.backward()\n            trainloss.append(loss)\n            optimizer.step()\n\n            running_loss += loss.item()\n            print('[Epoch %d] loss: %.3f' %\n                  (epoch + 1, running_loss / len(trainloader)))\n\n            predicted = torch.max(outputs.data, 1)[1]\n\n\n            if i % 100 == 99:\n                grid = torchvision.utils.make_grid(inputs)\n\n\n            \n    end = time.time()\n\n    print('Done Training')\n    print('%0.2f minutes' % ((end - start) / 60))\n\n    return cnn\n\n\ntrain(cnn)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"13\"></a><br>\n\n# Testing"},{"metadata":{"trusted":true},"cell_type":"code","source":"testloss = []\n\ndef test():\n    correct = 0\n    total  = len(test_data)\n    with torch.no_grad():\n        for data in testloader:\n            inputs, labels = data[0].to(device), data[1].to(device)\n            outputs = cnn(inputs)\n\n            _, predicted = torch.max(outputs.data, 1)\n            correct += (predicted == labels).sum().item()\n\n    loss = criterion(outputs, labels)\n    testloss.append(loss)\n    testcorrect.append(correct)\n    print('Accuracy of the network on test images: %0.3f %%' % (\n            100 * correct / total))\n\ntest()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"14\"></a><br>\n\n# Visualization Results"},{"metadata":{"trusted":true},"cell_type":"code","source":"def imshow(img):\n    img = img / 2 + 0.5 \n    npimg = img.numpy()\n    plt.imshow(np.transpose(npimg, (1, 2, 0)))\n    plt.show()\n\n\n\ndataiter = iter(testloader)\nimages, labels = dataiter.next()\n\nimshow(torchvision.utils.make_grid(images))\nprint('GroundTruth: ', ' '.join('%5s' % classes[labels[j]] for j in range(4))\n      \n      \nplt.plot(trainloss, label='Training loss')\nplt.plot(testloss, label='Validation loss')\nplt.legend()\nplt.show()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<a id = \"15\"></a><br>\n\n# Save Model"},{"metadata":{"trusted":true},"cell_type":"code","source":"def save_model():\n    model_path = input(\"{}\".format(args.save_model))\n    torch.save(cnn.state_dict(),'{}.pth'.format(model_path))\n    print(\"saved model to {}\".format(model_path))","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}