Data preparation, eyeballing and baselines.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data preparation, eyeballing and baselines"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this notebook, I will\n",
    "- First explore the Fashion MNIST dataset\n",
    "- Create a randomly shuffled training and validation set\n",
    "- Create a baseline model \n",
    "\n",
    "All of my experiments are done using just a single library `fastai`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5"
   },
   "outputs": [],
   "source": [
    "# Imports\n",
    "import numpy as np\n",
    "import pandas as pd \n",
    "from collections import Counter\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "import torch\n",
    "from fastai.vision import *\n",
    "from fastai.metrics import *\n",
    "\n",
    "# For better reproduction of the results, strongly recommended\n",
    "np.random.seed(7)\n",
    "torch.cuda.manual_seed_all(7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "_cell_guid": "79c7e3d0-c299-4dcb-8224-4455121ee9b0",
    "_uuid": "d629ff2d2480ee46fbb7e2d37f6b5fab8052498a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['train.csv', 'train', 'test.csv', 'test', '.ipynb_checkpoints', 'Identify the apparels.ipynb']\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "print(os.listdir(\"/home/jupyter/fashion_mnist/\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[PosixPath('/home/jupyter/fashion_mnist/train.csv'),\n",
       " PosixPath('/home/jupyter/fashion_mnist/train'),\n",
       " PosixPath('/home/jupyter/fashion_mnist/test.csv'),\n",
       " PosixPath('/home/jupyter/fashion_mnist/test'),\n",
       " PosixPath('/home/jupyter/fashion_mnist/.ipynb_checkpoints'),\n",
       " PosixPath('/home/jupyter/fashion_mnist/Identify the apparels.ipynb')]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_path = Path('/home/jupyter/fashion_mnist/')\n",
    "main_path.ls()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`train.csv` contains the filenames (without their extensions) of the images from the original training set and their labels. `test.csv` contains only filenames of the images from the test set. `train` and `test` folders contain the images. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = pd.read_csv(main_path/'train.csv')\n",
    "sub_file = pd.read_csv(main_path/'test.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "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>id</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   id  label\n",
       "0   1      9\n",
       "1   2      0\n",
       "2   3      0\n",
       "3   4      3\n",
       "4   5      0"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A reminder about the labels: \n",
    "\n",
    "![](https://i.ibb.co/p3JDHxd/Screenshot-from-2019-05-09-13-27-12.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9    6000\n",
       "8    6000\n",
       "7    6000\n",
       "6    6000\n",
       "5    6000\n",
       "4    6000\n",
       "3    6000\n",
       "2    6000\n",
       "1    6000\n",
       "0    6000\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Distribution of the labels in the training set\n",
    "train.label.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "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>id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>60001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>60002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>60003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>60004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>60005</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      id\n",
       "0  60001\n",
       "1  60002\n",
       "2  60003\n",
       "3  60004\n",
       "4  60005"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sub_file.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### DataBunch preparation\n",
    "\n",
    "In order to use fastai, I needed to prepare the dataset in a format that accepted by library. There are several advantages of using a DataBunch. The ability to play around several helper functions that support this format is just one of them.\n",
    "\n",
    "The following code block creates separate bunches for training, validation (from the original training set) and test set. Along with it, it does th following - \n",
    "\n",
    "- Applies the default transformations for data augmentation\n",
    "- Adds labels to the images present in the different bunches\n",
    "- Resizes the images to 28*28 (which is the original dimension of the images as mentioned by Zalando Research)\n",
    "- Prepares mini-btaches of 128 images in the three splits\n",
    "- Normalizes the dataset with the mean and std of the original MNIST dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_img = ImageList.from_df(sub_file, path=main_path, folder='test', suffix='.png')\n",
    "\n",
    "databunch = ImageList.from_df(train, path=main_path, folder='train', suffix='.png')\\\n",
    "            .split_by_rand_pct(0.1)\\\n",
    "            .label_from_df()\\\n",
    "            .add_test(test_img)\\\n",
    "            .transform(get_transforms(do_flip=True), size=28)\\\n",
    "            .databunch(path='.', bs=128, device= torch.device('cuda:0'))\\\n",
    "            .normalize(mnist_stats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A sneak peek into the new training and validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x432 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "databunch.show_batch(rows=3, figsize=(7,6))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAcQAAAGoCAYAAADYaZU+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzt3Xmw1fV9//EXgiKiLMpyQZBFEBXBLdYF+akNKqQal9TUWI2JppWJmaTa0Y5aY1obp51EMxMnGRg7tlErTWvGDR2sW4IF94psEkW5sl0W2Rdxg99ffvL6vMP55ELuveece5+PGWc+536+53u+l/O534/f9/uzdNq1a9cuAQDQwe1T7QsAAKAW0CECACA6RAAAJNEhAgAgiQ4RAABJdIgAAEiiQwQAQBIdYtHQoUPVqVOn3/tv9OjR1b40dFDPPfecOnfurBEjRlT7UtCB/OAHP9jtvXDx4sXVvrQW1aXaF1DLXn31VX322Wfp9bZt2zRmzBhdeumlVbwqdFSrV6/WlVdeqbPPPrvd3YhQ+4YOHaoXX3wx+1nfvn2rdDWtgw6xIH7Z99xzjz755BNdffXVVboidFQ7d+7UX/7lX+raa6/Vjh076BDR5jp37qyGhoZqX0arImS6B6ZOnarzzz9fAwcOrPaloIO5/fbb1alTJ914443VvhR0UMuXL9egQYM0aNAgTZo0SbNnz672JbU4OsRmeu211/T666/rmmuuqfaloIN5/vnnNWXKFN1///3aZx/+ZNH2Tj75ZN1333168sknNW3aNPXu3Vvjx4/X008/Xe1La1GETJtp6tSpGjZsmM4555xqXwo6kA8++ECXX3657r333nYfrkLtmjRpUvZ6/PjxWrFihX70ox/p7LPPrtJVtTz+d7MZNm/erGnTpumv//qv1alTp2pfDjqQ+fPna+XKlTr//PPVpUsXdenSRf/4j/+od999V126dNGDDz5Y7UtEB3XqqaeqsbGx2pfRonhCbIYHHnhAH3/8sb75zW9W+1LQwZx00kmaN29e9rOf//znmj59up588kkNHjy4SleGju6NN95od+2PDrEZpk6dqgsvvFD9+/ev9qWgg+nevbuOOeaY7Gf9+vXTfvvt93s/B1rL9ddfr/POO09Dhw7V5s2bdc899+jpp5/Wo48+Wu1La1F0iH/ASy+9pLlz5+quu+6q9qUAQFU0NTXp61//utauXauePXtq7NixeuaZZ/Snf/qn1b60FtVp165du6p9EQAAVBuDagAAEB0iAACS6BABAJBEhwgAgKQ2HmXKpPbqYvxU67TBzp07p7LvjjJgwIDsuJ/+9KepfPLJJ2d1+++/fyrv2LEjq/Nr7tq1a8XjunT53Z/zfvvtl9UtXbo0le+7776s7pe//GUqNzU17fZzpZZpP7RB7oPVVmqDPCECACA6RAAAJDExH2g1gwYNyl737t07lT/++OOsrhTG8TDsJ598kso7d+7MjvOdMGJdnz59UjmucPP666+nsodMCW+io+EJEQAA0SECACCJDhEAAEnkEIE/WszXfe7444/PXvfq1SuVPRcYzxFzd55v9OP23Xff7DifdrFly5aszqd1DBs2LKvr27fvbq8f6Gh4QgQAQHSIAABIImQK/NE8dDlx4sRUPuOMM7LjevToUfEcPmXi008/zep8JRxfqcbDoFK+AsqmTZsqnr9fv34V64COjL8EAABEhwgAgCQ6RAAAJEmddrXh+kys8l5dLMXVOm3wqKOOSuXp06encszNeW6wtJNE3KnC84b+vtWrV2fHvf/++6ncs2fPrG748OGpHKd8LFiwIJVfeeWVVH722Wez42bNmqU/Fm2Q+2C1sdsFAAB/AB0iAABi2gWwW3uyOe6NN9642/fFHS1814o4ZcJXmYkhTefnjNd02GGHpfKBBx5Y8X3xuoYOHbrb961fvz47buHCham8YcOGiteI9q0U8vU2GVMGpb8hr/Pzx/d4XdxNxleGeuyxxyp+VglPiAAAiA4RAABJdIgAAEgihwjsVimHGHecP+KII1J5x44dFc/50UcfpbLnE+P5o27duu32/L6km5Tn/+L5161bl8oxR+nnHzx4cCqPHDkyO87rYg7R80WVdv9A/fLv15cqjMsMervzvLiUt/FSnrw0Benoo49O5auuuiqr27ZtWyo//fTTWd2HH35Y8fMcT4gAAIgOEQAASYRMgd0qhf3OPvvs7LWvCuPviyFNDyHFIekeQorv8/Ctl2NY1KdJxN0u4obBzlfC8bDuiBEjsuM8hDp37tysjjBp/SmlBWJd7969U3ns2LGpvH379uy4zZs3p3JDQ0NW58fGVZa8ztMCJ5xwQnbcJZdcUvGz/e/BpyBJ0m9/+1s1B0+IAACIDhEAAEmETIE9Nnr06Ox1pQ124899NGcMi7oYCvXwlZ/Tw1OStHHjxlT2EXdSHq7t3r17VrdmzZrdXkc8rjSCFvWvUjuTpEMPPTSVv/KVr6Syh9ilPHQ+ZMiQrM5XOjr44IOzurfffnu3xx1++OHZcS+//HIqDxgwIKt78MEHU3nZsmVZXRzxWglPiAAAiA4RAABJdIgAAEgih9jm4satno/yTVzfe++97LjXX389lWPc3vkqElI5V4Xm8xUz+vTpk9X5ah3+7x1Xn6mUa5Ty79RXlZHyaRG+4kY8v69UU9ol4KCDDsrqDjjggFT2nTDicaXdOsgv1j9vM7H9eJv0jaK3bt2aHderV69U9nYr5W0w5vR8mkTpb8inJ02ZMiWrizl1F/PylfCECACA6BABAJBEyLRVxLDopEmTUvmrX/1qVjdw4MBUbmpqSmUPPUjSnDlzUvm2227L6nyjzDgUOYZQsXf69euXynE6godxPCwaFz72Ielx0WKf+uDtQMqHrzc2NqZy/G4POeSQVI5t0D8vbh7s1+xTN2JYy4fKe5hVImRaj0oLyseVh1auXJnK//3f/13xHN7OfIqElC8OH+9v3/3ud1PZF/CO04fOPffcVI6rMXk7jtfV3NQRT4gAAIgOEQAASXSIAABIIocoqfJuAlLzV/H//ve/n8pf+tKXsjofph83xvTh9mPGjEllzwdJ+arvcajzT3/601Q+/vjjszpfLgx7zzcBjvk//059iSvPx0VxGLi/jnWVlp2KeTsfGh9X++/Ro0cqx+HwPtXCcy1xU1WfIvTss8/u9prQPnlOrnRP9DazePHirM6XZ4v3pT//8z9PZc8vLl++PDuutGuF37t9Cpsk9e3bt+L7HE+IAACIDhEAAEmETCWVhx/7EHvfGNZDpFL+iB6H+HqIIa5W4kPn/ZE/hlb9nLfccktW9yd/8iep/J//+Z9Z3aJFi4Q/nodMY8jRQ5relkqrY8Swk3/3cbqGt5FSW/JwVen8sX36dXr4NPIpPXHIO+pDKT3kSmHR0kbClT4rvvZpTFI+tcjDos8//3x2nP+tnXnmmVmdpwni9cf7aSU8IQIAIDpEAAAk0SECACCJHKIkaeTIkal85ZVXZnXnnXdeKvuQ+rh8l+deYg6oUo5JyofK+zJFvXv3zo7zof4xPu65zVNOOSWr8yXfsPd82ak4ZNxf+3cdc3zeLuKya76cWvzuPefn331sg36OuLSat8+4U4Uf6+eP0y48B8RSbbXL211sg/79lsZOlJR2xaj0WVLefnwZQKny/dOnm0nSzTffnMpxNxYfL7FkyZKsjt0uAADYA3SIAACoDkOmcSivhwRKj8VXXXVVKl922WVZnQ+pj7sEeLjKh+7GYbwehoob+C5btiyVY4jBh/B7+HTBggXZcT5M2VcMkfIwXZwS8IUvfEH443l4JrYzD/eUQkj+vhh29RBqXJnGz+9h0rhrRWlHju3bt6dyKZTrodD4t1aakoHW19ypM6WpP36fiiFHbwfr16//4y5W5b+FuNpWpelnq1evzo7z1W82bNiQ1fnvGv992O0CAIA9QIcIAIDqMGQaH8Mrrajws5/9LHt90UUXpXIMNfk5Yoih0shAD0FJ0jvvvJPKcfNLHyG6YsWKrM5/Hx/FGhf39lFTc+fOzep8JFYplIK95yMx4yhN/zf3cGcMi3o4MoYtPdQdRyl7CN7rSmGnGN7088eQbKW/oTgStiXCaNh7pVCo1/mG4b6xgJS3XV/hSpJOPPHEVP6f//mfrO4//uM/9vh6Y8jdQ/qTJ0/O6t58881UfuaZZ1LZ00hSfi+Nf0Mu/o3GjYYr4QkRAADRIQIAIIkOEQAASXWSQ/RYdFzF47rrrkvlr33ta6k8cODA7DjPk8QpE6UV2z1O7bmimF/xc8ZNLD12Hjdu9dVAPvjgg1SO0z+OPvroVPZpHFK+8aavqCLluc44zBrN5999zPF5fs5zFXGll0ob/Up5ziN+994+PT/tm/5KeW485gVLOcRKQ/bjFJ61a9emMrnqtuftYNKkSVmd56v9nhVXevH74muvvZbVPfzww6kcp2t961vfSuV//dd/rXiN3i58OpskXXLJJan8wgsvZHU+hcKnDMX7vf8bxL8vvydv3ry54nWV8IQIAIDoEAEAkFRDIVNf7PXSSy/N6r7zne+kckNDQ1bnj9ceJoohIw8jxFBQabFaf0T3urh4socYNm7cmNX5VIuVK1dmdT69wn+3VatWZcc1NTWlcgzX+gLeMQzi57n11luF5olhRW8Hcbi3DwWfN29eKsc26GHv0qovcUqPtzsPLcVQv7fJOOXDQ/pxCLpfv0/xiAsw+/vi+eMUEDRPaeWt6Oqrr05lDz9K+bSFmTNnpvLSpUuz4zzsHaeA+bSgV155Jas7/vjjU3nChAmp/Otf/zo77sgjj0zlL3/5y1mdX1ds/34f9PZeCovGe7WHReNKTTGEWglPiAAAiA4RAABJdIgAAEhq4xyiD8O9/vrrszrfgSIureZDumPs2XMjHlMuxeIjj+OXpmB4/Drmh/yz41Bhz+34FAlJmj9/fir77xnzkJ4L2LRpU1bnuYCYf/K8DznE5ovL7/m0iPjd+/QHX8JvwIAB2XGed4vt2PMf/fv3z+p8moe3Lc/9SXmbj+3Yl11rbGzM6k4//fRU9lzgniyNRQ5x75SWooxTBXwZtl/96ldZ3XvvvZfKPi0r3m+83fkSb5L0pS99KZV9Gcl4nXEch/N7vOcMpXzsRpwy5Nfl99n4b+DtOrZBv3/uSW42O65ZRwEA0M7RIQIAoDYOmY4bNy6Vx4wZk9U9/vjjqRx3evDh33E4rT+G+2NxfESOUxWchyniKiQekvVhyaXQZAx3+lD5Up0PifZVa+LruGmm123ZsiWrY1PXvVMK6cRVWTw8420k7jTg7ay0ak0ME/n7PJ0Q/xZiqsGtWbMmlZcvX57VnXXWWansfzdxRScPm8W/EzSfhyPjd+3TA+Lfsu94E+9vvgLNkCFDUvnll1/OjiuF9P2+G9u/f/eeTvjGN76RHfdP//RPqRxDsn6OmHbwf4fS7kP+dxPbpx9b2gS7hCdEAABEhwgAgCQ6RAAAJLVxDtGXtbrllluyOo8bx2WhPEcTl13zvIlPb4g5w1J+0cW4tOcKfVmrmEP0HFMcSu1x7xgT96WJfGh8zPN47jGeHy2v1M5KQ8G9bfXt2zc7zvOLkbcDb2dS/n17Xim2cW+DMXfsOcQ4JN3zK74jii8XKOW/d8xvofk8txaXgPTvJt6LPCftUySkvG15bnDUqFHZcQ888EAqx3EKvkvP4YcfntX5uAVffvKxxx7LjvOdWuLfSWlaUCXxXuf/JqXzR80dS8ETIgAAokMEAEBSG4dM45QDF1c1by5/pPZwZAxNekg2PoY3dziwhzdKO2bEx3OfkuFhUSlf4SN+Nqonhl88ZBq/ew+f+yozcRqEh0JjuKc0LchDmv6+GFr142Ib9zYYV1Ly93m6ojQ1hA2B957/u8aQqX9vMWTqO9ksWLAgq/N24ZuQxxWXxo8fn8pxKoK3+RgK9c/2kGlsSz51o7QaU2k3Gf9biP8G/vvEFbtKIdPmtleeEAEAEB0iAACS2jhkunjx4t99cHhkLi3o6iGqGGLwBY4rrVojlUMRHnoqrWjgq8rEEYN+DkaB1r9SWD2G/r0NesgotkFv17HOQ+dxM9NK4dp4jR6iigsfeyg0LtzsbddHN7///vsVj8Pe83tfDDl6+L00Wj2mV/x+5PfZ0mYI8R7mbTCG8H1B75EjR1a8Dg99+ohTqXLoP35eaTSqt+N4H/fzx/YfZwVUwhMiAACiQwQAQBIdIgAAkto4h+hirq6Uu/PY9rp161rtmoDPxdyI5zVi7sVf+7DwUhuPORTPD8VdMrzOcyMxP+TXGHPtns/p16+fKvFVVOJQ9f/93/+t+D40n68Q4zv5SPm/ecyDeZ4t5tYqrdIV85DNnX4Q26e/9nxcbIPexuNUOj9HacUu/93i30JpxRn/m42f3dx+gydEAABEhwgAgKQqhkyBWhZDOv46hlN92kVcmN55XQxllRbL9pCpT4uIm7h6SCoOm/cVdEqbHy9btqziNcWVQbB33nvvvVSOC6iXVquKG6e7SuHy2B6bO3Ujhhz92NI0tdL0s9JqXpU2z44r7fhxpXBt/Ozmtl2eEAEAEB0iAACS6BABAJBEDhHYrZif8DxhHJLuPA8Th9R7XczfeM6je/fuWZ1/np8zHlfKUfpybTEv4zth+LQL/50lpjy1FM/P7ckuP6U8s3/flcr1qK3z1jwhAgAgOkQAACQRMgV2y8OIUj6NIW786+bOnZvKo0aNyupKK9V4aCvuBOOhVg/l+hQMKZ9aMWzYsKzOw21xpR03Z86cVI6/Z/w8tC02ZW59PCECACA6RAAAJNEhAgAgiRwisFu+w4okvfjii6k8YcKErG769OmpfP/996fyZZddlh3nO3zHHKXnDeNyW74sl++A4LuYS9LmzZt3e1ysi+f36RtDhgxJ5ZhrXLFihYD2jCdEAABEhwgAgCSp0656X8oAAIAWwBMiAACiQwQAQBIdIgAAkugQAQCQRIcIAIAkOkQAACTRIQIAIIkOEQAASXSIAABIokMEAEASHSIAAJLoEAEAkESHCACAJDpEAAAk0SECACCJDhEAAEl0iAAASKJDBABAEh0iAACS6BABAJBEhwgAgCQ6RAAAJNEhAgAgiQ4RAABJdIgAAEiiQwQAQBIdIgAAkugQAQCQRIcIAIAkOsQ/6Mknn9Rxxx2nrl27aujQobrrrruqfUnoQH70ox/p1FNPVe/evdWrVy+dfvrpmjFjRrUvCx3M/fffrxNPPFG9e/dWt27ddNRRR+nOO+/Url27qn1pLapLtS+glr322mu64IIL9Ld/+7eaNm2aXn75ZU2ePFkHHHCAJk+eXO3LQwfw3HPP6aqrrtJJJ52kbt266Z577tF5552n3/zmNxo3bly1Lw8dRL9+/XTrrbdq1KhR6tq1q1544QV9+9vfVpcuXfS9732v2pfXYjrtam9dfAu67LLL1NjYqNmzZ6ef3XDDDXrooYe0ZMmSKl4ZOrIxY8bonHPO0Z133lntS0EHdtFFF0mSHn744SpfScshZFowa9YsTZw4MfvZxIkT1djYqOXLl1fpqtCR7dy5U1u2bFGfPn2qfSnooHbt2qVXXnlFs2bN0llnnVXty2lRdIgFTU1NamhoyH72+eumpqZqXBI6uDvuuEMbN27UFVdcUe1LQQezadMmHXjggeratatOPfVUfec739F3v/vdal9WiyKHuJc6depU7UtAB/Pzn/9cd9xxhx577DENGjSo2peDDuaggw7SnDlztH37ds2ePVs33XSTBg4cqG9961vVvrQWQ4dYMGDAAK1atSr72erVqyXp954cgdb04x//WLfddpsee+wxTZgwodqXgw5on3320YgRIyRJY8eO1YYNG/T3f//37apDJGRaMG7cOD311FPZz2bMmKEhQ4bwf+hoM9///vf1D//wD3ryySfpDFEzdu7cqY8++qjal9GieEIsuO6663Taaafplltu0RVXXKFXXnlFd999t37yk59U+9LQQfzN3/yNpk6dqmnTpmnUqFEpYtGtWzf17NmzyleHjuK2227T+PHjNXz4cH3yySeaOXOm/uVf/kXf/OY3q31pLYppF3/AE088oZtvvlmLFi1SQ0ODvve97+n666+v9mWhg6iUq77yyiv17//+7217MeiwrrvuOj3++ONasWKF9t9/fw0fPlxXXXWVJk+erM6dO1f78loMHSIAACKHCACAJDpEAAAk0SECACCJDhEAAEltPO2iNVZ38XP26tUrlbt27Zodd8ABB6Ty//t//y+rmzRpUioPGzYsq+vevXsqP/PMM6kc15I8/PDDU3nu3LlZ3S9+8YtUXrZsWVa3Zs2aVN6xY0cqx3+rlhj7xPiplmmDp5xySvba29o++/zu/zEPOeSQiucYOHBg9nrTpk2pHNfJ9blePtUitjM/x/bt27O6Tz/9tOK1tCXaYHVXufL71E033ZTVDR8+PJXffvvtrM7blrfHI444Ijvu0UcfTeW46Lff36qp1AZ5QgQAQHU4MT/Oeenfv38qn3HGGan85S9/OTvu0EMPTeWtW7dmddu2bUvluGh3jx49UvnVV19N5bgXnR83ZsyYrO7aa69N5UWLFmV1zz77bCrPmzcvlTdv3izUppdeeil7ffDBB6fy0KFDUzm2g3333TeVZ86cmdV16fK7P0WPSkh5e92wYUMqjx49Ojtu3bp1qextWpI++eSTVF6xYkVWF58m0X6tX78+lT06JSlbfWu//fbL6gYPHpzKO3fuTGW/70n5PXjOnDlZ3VtvvbUXV9y2eEIEAEB0iAAASKJDBABAUp3kEH3kXu/evbO6f/7nf05lz9/EHMqWLVtSOY6489xOrPOY+5FHHpnKMca+cuXK3b4nnt9j7JJ08sknp/LHH3+cyrfcckt23MKFC4Xa5G3Lv8OY7/Yc3zHHHJPVffbZZ6k8f/78rM5Hnfpx/rlS/ncSR7F6rj223QULFggdg48Q7devX1bneWwfcyHlIzM957x27dqK56hHPCECACA6RAAAJNVJyLRbt26p/Fd/9VdZ3ec7OEv5kPQ4+dVflybGenhTyocYn3baaakcJ3d6iDaGpPzz/BqlPMx14IEHpvIPf/jD7LhLL700ldvbppz1xr8zKV/MwadgzJgxIztu8eLFqRzboLe72La8zhcBiO3Aw/0eupWkvn37pnIM5fpweG/vaH883Pnhhx9mdR7ujFNxvL36BHu/N0v5AijxXloPeEIEAEB0iAAASKJDBABAUp3kEN0JJ5yQvfY8SsztVBKP85xNzKF4vqWUo/T4e8wh+jnjsGR/7ctrxcXDL7744lSeNm2aUD0xb+K5ZV+uyqdISPlQ9pi/8e++lOP29h6nCPnUCl+GS5L233//VI5Tl3zB8JjjRvsV26C3u7h0pN8zPTcY75ee4/bNFuoFT4gAAIgOEQAASXUSMvXQUFyBw4cHe3gzDl3313HYuYcDYhjBPzu+r9L5Y6jMz18Kh3m4IYbDjjrqqIrvQ9uK7WDIkCGp/O6776ZyDD/6+2Lo3NtM3MvTp1B4aDWew0OmMVzl74srKXkImJBpx7F06dLs9dlnn53Ksf14GqjSqjVS5dBqveAJEQAA0SECACCJDhEAAEk1mkOMeTbfQTzGpT2e7XHuuHSVHxfP39wcX6nOz1HKUcZz+HXFvKfzYfPx/DFnidYVd7T372348OGp/P7772fHrVq1KpVj/q+U//Y24+0/Tv/waR0xB+2v407pxx57bCr7ri2of6X7zXPPPZfVXXDBBakcx1L4dB+/35TGS8TdWOoBT4gAAIgOEQAASTUaMo18OHnc+NdDQR7Kiiso+ArtMSRVWiWk0uo3pd004nuau9OG70iwcePGrM5DbL6ivFSfoYl6Flci8s2bPWx5+umnZ8d5iCruVOFt13c9kfJwuX/X8Tp8BZrYfnwqh7d3KV8VydtnKYSP+lD6Dn33FUlqbGxM5QEDBmR13ma83cV7nd9n63GzYJ4QAQAQHSIAAJJqNGQaH/OPPPLIVPZHcikPjfr74mofpUXA/XVLjNjck3CqX7+HynxEopRffz2uANGeeAhTkpqamlLZ28+FF16YHeffdVwlxMNLcRTrIYccksqbNm1K5TgS0EPpflx8Hdug/z4e8o0jTtk8uH2JI/EXLVqUyt4OpLxd+0pHcTSz38PiIvL1gCdEAABEhwgAgCQ6RAAAJNVoDjHyuPTWrVuzOs9/eB4m5hB9uHrMofiQ4r0dal4aru51Mebu1+J5npiH9F0+4golaFtxOLm3Qf8OYxv06UM9evTI6rx9xjy25wYPOuigVI4rznhOMeaZ/fxxpaMzzjgjlX2Xg1/84hfZcfHz0L6sXbs2lX0HFym/R/br1y+VlyxZkh3nYx2GDh2a1fnfTbwP1gqeEAEAEB0iAACS6iRk+tRTT6VyHGp+/PHHp/L48eNTuaGhITtu3bp1qRxDUqWFv5sbQvXj4vD00rQOD6MdfvjhqRynXcyaNSuV4yokaH3+HQ4bNiyr82kRy5YtS+V58+Zlx3kYM7YzHwIfpxb5aw9pxukfpalFvrpIDNf6aw+VxeMImbYvsQ16WNTDp1K+Qpi3cS9L+apHPl1OkgYNGpTKvipOLeEJEQAA0SECACCJDhEAAEl1kkP03MVDDz2U1cXXn7v55puz11/84hdTOQ75LW3SWymHWFqeLeYQPW9YyvuMHj06lbdv354d57nTWIfW59/p2LFjs7oTTjghladMmZLKmzdvzo7ztuQ7m0jlHVfizhifizm+OCXJ+VQLHzYf3+fLclX6XLQPcfqWLxkY25K3f2/XcVcMvzcddthhWZ2P9yCHCABADaNDBABAdRIy9SHkzQ1pvvvuu9nrCRMmpHJcqaMUrqp0/tKuFaVzxFVOPEQVhzC70sbCaFsxnOShSx9qHjf69VB93NzXV6OJm2B7W6u0MpOUTy2Kdd5+Dj744KzOr8XDZnGlHbQvcaPxY445puKxPvXH20icBuf31pia8nvwww8/vGcX20Z4QgQAQHSIAABIokMEAEBSneQQSzt1V9plIuZhPO8T83+lnSoqHVfK6cXzeyw9nr+5Uyj2dhcOtLzFixdnrz35PuwjAAAc+0lEQVSv57uS+A4TUp6zibuVl3LQXuflOIWnlGv3ZePi+zyH6NfouSK0P7EN+rJrsc5zgz6NLObCK7VVKf/bqFU8IQIAIDpEAAAk1UnItKRS6PKDDz7IXvuw4dJw+LgbhYcK/LNiWDS+z/mKH/F6m5qaKr4PtSl+Z94WfANfnwYh5dMY4ga+XhenTHhY3dtSHDbv7TqGaz20FduuX4t/tu/Egvan1EbilJtKKadSeD/yEHxsW6tXr27GFbc+nhABABAdIgAAktpByLSSGA7wUFNp9Fx85I+r2lRS2mTYzxnDVZs2bWrW+VE74ubTvpnqiBEjUnnhwoXZcf7dx5Cph6viSjK+uH2lVWtiXQy7ujiC0M/jof94HWhf4khkD7nH+5Lfw7ytxtVoKo1GlfIw7NChQ7M6QqYAANQQOkQAAESHCACApHacQ/QdLKTfz5s4j3uXjiutFlMablw6R1wtHrUvri7k0zDGjRuXyrFN+Hcf26fX9ezZM6vzfKC/z6d4SHk+J+aqvV3HjX8r7V7ASjXtW2w/njeMmwd7ztvHZ8TjvH3G8Rf+dzNq1Kisbs6cOalczY2peUIEAEB0iAAASGrHIdMYriotsO3HNncR7dI5SuKUjErhgeZuVIy2F4ek+6pI3g569+6dHecLzsdwVen79eHwHoYqbeC7JwvM+6LOq1atSuVevXplxzV3EXzUhwEDBmSvfeWjmBbwlb68LcVpFy6mBby9HnHEEVmdT/Gp5updPCECACA6RAAAJNEhAgAgqR3nEOOyQR7rLg2HL9nbPGFpWTeP1Zc2HUbt8JX/pTy/4nmT4cOHZ8fNnDkzlT1vJ+Xts7QTRmlzVn8dpw95rjrmHv2zN2/enMpx6TafhhH/DVB/DjnkkOx1Kcft0ys2btyYyvGe5Tlu/7uQ8rYbc4i1kpPmCREAANEhAgAgqR2HTOMjeGkVj1Io1I/1sFMMVzV36kasqxR6qpUQAn5f3PjXQ0G+80XcjcLbjIcmpbxtxXBnpVBoDK36NIk4dN1DW/H8Hrb3c8RwmE//IGRa/+K0CH8d74k+tcjbe2nXlsjft3Tp0qxu7NixqexTf9oaT4gAAIgOEQAASXSIAABIagc5xEq5tvhzfx1ziB7bjjuN+7GeU4mxc98lIH62vy9OB4lDk1H7NmzYkL32/LTvbh9524q7nPiyWaXh6p5rjO3Mh9HHHI2L5/dz9u3bN5VjrtSH4lczz4OWEXPccecK16NHj1T2thvvZ35fjDlozy9efPHFWd0TTzzRjCtufTwhAgAgOkQAACS1g5BpJfFR3l+XpmTEVTx85YXS1A0/Z/xsD5XFujj0GfXHv1NfxSOuBFKamuPhqxh29VCWh6Fi6LN///4Vz++fHduch4A9rBWH3seNi1Hf4vfrbTDe37xdl1ZV8ik9Mf3k03veeOONrM5XdXr++ef/4LW3Fp4QAQAQHSIAAJLaccg0hoU8vBRDBc1dZcZDY6WFuOOmmR5WiJ8dVw1B/fEReL6Cy+DBg7PjtmzZkspx893169enctw02kOVfg4PY0nSySefnMpxxZA4Mtb5eTxsFv+G/PdE/YurJfm9L373PsLYw6m+go2Ut9V4H4yjWt2wYcN2e464GXdr4wkRAADRIQIAIIkOEQAASe04hxjj182d3hBzg34eryutRhPrPOYec4hMu6h/vrGq5xBPOumkiu+JK9V4+4lD3n2XCd9NI+aABgwYkMq9e/fO6jwXE/PWngfyPE88jhxi+xLzc34vilMmKq1OE4/zaWqRT7uI09saGhpS2dsxOUQAAKqADhEAALXjkGlpE9SoNNXCQ5w+faK0eHgMi3qIwRdxln4/5ID6s3r16lTu06dPKsfFkj3EGUNL3rbilAmfhuHt+K233sqO87Brv379srr33nuv4vn9WkaPHp3K77zzTnacD71H/YsrHXlI3Kf3SHnq6OCDD07luKqS33dj2N6nFsX276F6X2B+0aJFlX+BVsATIgAAokMEAEASHSIAAJLacQ4xxqg9bxLzf35snHbhOT6Po8c8jOdX4hJFngOKy3LNnz9/978A6obnOU455ZRUHjlyZHact4uVK1dmdd7uYq7O26dPyfCcjJTnruNnz5gxI5Xj9AnPL/oUkrjrQFwqDvUtTu/x8Q1xukOlpdvi9KHDDjssleO0IM9Zxvb/zDPPpHLcCaMt8YQIAIDoEAEAkCR12lWac9DSH1bYIaKlxXDA17/+9VSeMGFCVudh0dI0CB+KHFdaWLZsWSovXLgwq3v99ddTOQ6Vd3Hz4JbWhl91zWrLNhidc845qezTGyRp6NChqXzUUUdldR6u8rD9ihUrsuNOOOGEVI6/p7e7GGq9++67U7mxsTGVY4i0ubvClNAGq9sG3cCBA7PXX/3qV1M5htyXLFmSypMmTdrtzyXpzTffTOULL7wwq5szZ04qx7SS3z8XLFiQyo8//njlX2AvldogT4gAAIgOEQAASXSIAABIasfTLuLyadOmTUvluBzQaaedlsqjRo3K6nz3Zl9ey5cvisedfvrpWZ0vtxWHIt9www2p7PF4ci3tj+eWzz333KzOh6v7yv9SPsXHp+3EtuS7FcQctw95j0Plve36lIzSjuq0z/rX1NSUvf7Vr36Vytdcc01W5+3C89hxV6FLLrkklWMOesSIEans91Ip3+Gie/fuqdwaOcQSnhABABAdIgAAktp42gUAALWKJ0QAAESHCACAJDpEAAAk0SECACCJDhEAAEl0iAAASKJDBABAEh0iAACS6BABAJBEhwgAgCQ6RAAAJNEhAgAgiQ4RAABJdIgAAEiiQwQAQBIdIgAAkugQAQCQRIcIAIAkOkQAACTRIQIAIIkOEQAASXSIAABIokMEAEASHSIAAJLoEAEAkESHCACAJDpEAAAk0SECACCJDhEAAEl0iH/Qk08+qeOOO05du3bV0KFDddddd1X7ktDB0AZRS5577jl17txZI0aMqPaltDg6xILXXntNF1xwgSZOnKg5c+boBz/4gW6++WZNmTKl2peGDoI2iFqyevVqXXnllTr77LOrfSmtotOuXbt2VfsiatVll12mxsZGzZ49O/3shhtu0EMPPaQlS5ZU8crQUdAGUSt27typc845RxMmTNCOHTv0wAMPaPHixdW+rBbFE2LBrFmzNHHixOxnEydOVGNjo5YvX16lq0JHQhtErbj99tvVqVMn3XjjjdW+lFZDh1jQ1NSkhoaG7Gefv25qaqrGJaGDoQ2iFjz//POaMmWK7r//fu2zT/vtNtrvb9bKOnXqVO1LQAdHG0Rb+OCDD3T55Zfr3nvv/b3/OWtvulT7AmrZgAEDtGrVquxnq1evlqR23zBQG2iDqLb58+dr5cqVOv/889PPdu7cqV27dqlLly667777dNlll1XxClsOT4gF48aN01NPPZX9bMaMGRoyZIgGDRpUpatCR0IbRLWddNJJmjdvnubMmZP+mzx5sgYPHqw5c+boz/7sz6p9iS2GJ8SC6667TqeddppuueUWXXHFFXrllVd099136yc/+Um1Lw0dBG0Q1da9e3cdc8wx2c/69eun/fbb7/d+Xu94Qiw46aST9Mgjj2j69Ok69thjdeutt+qHP/yhJk+eXO1LQwdBGwTaDvMQAQAQT4gAAEiiQwQAQBIdIgAAkugQAQCQ1MbTLpq7skY8zl/v3Lkzq+vS5Xe/wr777lvxnEceeWQqjxw5Mqv74IMPUvnFF1/M6rp27brbsn+uJJ122mmp3Lt376zu7bffTuW33norq/v4449Tedu2ban86aefZsf5v8Fnn32mvcH4qdpd3aVz586p3LNnz6xu1KhRqextMLaD7t277/Y9krK5Yo8++mhW9+CDD6ayt5HNmzdnx8W/vb1BG6ydNtivX7/s9fjx41P5xBNPzOoOOOCAVH7nnXdSuW/fvtlx/r6tW7dmdZ8vKCFJN910U1b34YcfNvey/2ilNsgTIgAAokMEAEBSja5UsydhlT59+qTysGHDdvtzKQ8nxVDBRRddlMpnnXVWVuehgoMPPjiVP/nkk+w4Dyd5aEDKQ1sDBw7M6lauXJnKixYtSuU1a9aokj0JKaN2eEj/kEMOyeq8bR199NFZnYdT161bl8q9evXKjtu+fXsqb9iwIaubOXNmKs+YMSOri8ei/fC2I+X3sxhWHzBgQCrHe4yvp+upnQMPPDA7burUqam8fv36rM7bZwyR+udVM6zOEyIAAKJDBABAEh0iAACSajSHuP/++2evPecX838+xcHrDjrooOw4z+PF4eSNjY2pvGnTpqzOY+Ret3z58uy4jRs3prJPpZDyKRQ+bF6Shg8fnspDhw5NZY/TS9LixYtT+Y033sjqGMpeXZ7/8LzeEUcckR3nbXK//fbL6jx/E/PTc+fOTeUxY8ak8je+8Y3sOM/zxHbm+cv/+7//y+reffddoX2KYwq8fY4dOzar8/ENp5xySlbnYzAWLlyYyp7TlqSjjjoqleNYiueee67iddbKPYwnRAAARIcIAICkGgqZ7rPP7/rmbt26ZXWHH354KnuIUcqHEXt4YMeOHdlxHoKMIdNly5bt9jokae3atan80UcfpXIcUuziijkeAovhVP9de/Tokcox5HXooYem8pYtW7I6D6ei7fXv3z+VzzvvvFT2FZCk8gpD3p5iuHzcuHGp7FMkYjueMGFCKi9dujSre+SRR1I5hrl8dRGf+hOvA/UnhiJXrFiRynElmXPPPTeVfRqQlN+bRowYkcrxPuuf558l5emtWbNmZXXxWqqFJ0QAAESHCACAJDpEAAAk1VAO0Yf1HnvssVndoEGDUjnGxD334uUYv660a4WU5/zikkU+JN3zlXFJJJ9aEYfN+84Ycek2z0t6fijmUf2zfXqGlMfffeg9Wkf87n25P8/Pef5Zyr/D2I69Dcbzey7Pv19fji0e9/rrr2d1fi1x2SxfRu7MM89M5TfffDM7zpcZZInA+hCn91x88cWp/JWvfCWr8+lucaxDpdxg3O3Cx2DEPKQvpxlzj7WCJ0QAAESHCACApCqGTGNo0sOMPoxdyh/D46O2r6DuYacLL7wwO86HoftqClIeCoor3AwZMiSVfepD3Ej43nvvTeW4QbBvvPnFL34xq/MVb2bPnp3KvvKNlK8wEc/vrwmZtr7479/Q0JDKvpqRr4Ak5d9hDKt76DyGU73t+mpM7733Xnact5/49zV69OhUjrtpeDjV/4bi34mHSZuamrK6WllpBLl4v/na176Wyscdd1xW56H0+H36ffc3v/lNKsewqK9wE+9Fnuq58cYbs7o77rhjt9ff1nhCBABAdIgAAEiiQwQAQFINTbvw3EgcKux5Dc8ZSvl0B8+7+a4AUj6tY8mSJVmdr8oeY+c+RcPj5XHqRmm3ad8xw69Dyoe8+/SS+Ht6HjUOy4//Xmh53gaHDRuW1fnUHM8z+3JXUt5WI28zcYk3n67hQ+NjO/b8ZRwOP2/evN2eQ8rbj/8NlZZQjMu/eQ4UtSPuuOLLrsWpFX7PifdBb5M+zSje6/wc3m7j62uvvTare+mll1K5tCtGa+MJEQAA0SECACCpiiHT+EgeN/51HiKM4UEPJc6fP7/icT40Pm6QGncNcD7U3Ifbx+kZHl6KoQgPmcYdEHznCj/OQ2/xnDH05v+WMYTBcPiW4WFGX3FDysNEHn6P04e8Lu6IEr8351M0vK3GnVk8ROurI0l5+4khd78WH14fpzj5vwHtqnZ5u/BUjpS3kdJ9JE4L8hSRt5dSO4jtzMPqMWx/9dVXpzIhUwAAqowOEQAA1dAoU3+8jo/y/vgeR2n6o7cvnB3DPb5KyJFHHpnV+WK1HtaS8gXDfQRePL+HaN9///2szleA+MIXvpDV+WojHk6NI7Q85BVDZaWFy2t1Ed164yFNH80p5f/mHgKP4Spv4zHU5OeP4VP/vn10Zwz1e5uJbcTbri/wLOUbTPvGrXFFHg/DlkbMorp85LqnYaT8fhDD6vG189SRpwhiW/XjYhv388e0ki8q738npU21WwNPiAAAiA4RAABJdIgAAEiqoRyix55j3NvjyHEHAZ+u4av4x5UzPL9y2GGHZXUe247D4T3m/vbbb6dy3I3CdwaIuaNf//rXqRzzPr46jedDY+zcf58Yt/dcTxzqjJbh7aK0MpB/N6UcSmnIe+Rtwd8Xv2v/G4o5Gr/+mCuqtHFxbINeF/8NyFXXjpEjR6Zy3JDcv/vY5vZm+lbMVfvr2H7ivdV5+/ExHgsWLKj4ntbAEyIAAKJDBABAUhVDpnHhY1cadh7Dnb5YrU/PWLhwYXacb7Iap0WMHTs2lY899tiszh/l/bpiWNcX/o6rMMyZMyeVp0+fntUdc8wxqeyr38SQr392DNf6UHxCV63Dwz/btm3L6rwteLg8hjQ97BTDln7+GJ7y8KfXxXbm4dS4OLyHOOP1+/v8nPHvsNJxEu2ulgwePDiV4yLvHraM95jSVJrmrk7jYvvxNh5D7v73cPLJJ6cyIVMAAKqADhEAANEhAgAgqYo5xBhD9iXZ4nBdzyHG3IUvi7Z169ZUbmpqyo7z5dniBq+XX355KsfcoC/T5dccr9Ffx+XTPJbuUzDi+956661UXrlyZcXriL8bWp/nbV999dWszr9fz2n7VBwpzynGIeieJ4xTJjxn43WlZa18CoaU52g8py3lGw2XdpYpTbtA7fBpWD179szqvB3EHS38+y1N3/LjYj6xue+L/J7p00baGk+IAACIDhEAAElVDJnGYeE+bDtOi/CVNGIIwHeI8J0GYmjV3xeHvPvq8HEDXw9VeoiqtNNA3BjWPy+GsjyM4P8m7777bnZcDG84DynH86P1+Xe4fPnyVB41alR2nIcZY8gxDoF3laZkxBCUt4M45N3P/9vf/rbidfmuMDF94KF6T1WgtvjqVz4dTMpDmvFe4W2muStelTa2Lu2EUVq1Jk6ta0s8IQIAIDpEAAAk0SECACCphnKIL730UirH+HVDQ0Mqx6WIPLfmORXf+ULK49kbNmzI6jyHEvMmjzzySCr/13/9127fI0nXXHNNKnsMP15XaUk2/zcpxfBjPpG8Ye3w76Jbt25ZnX9v8fv1PGH8Ppv7/ZZ2Ky/lh7zNz507N5XXrFmTHed5n9IyX2h7vuyj3y/jFDBvZzEP7Dno0rSa0vSJ0n3LPzvew3zMh4/HaGs8IQIAIDpEAAAk1dAGwS6uwOGrtqxatSqr88d8nz4Rh/X667gKjG+iGVcJ8aHmvhJODCn4NcdNM0srfHiYy6d87G3YDNVV2o3Cw5al77f0XZemXVQ6Xzy2tJuMT12KSlN/UF2+e5CHHOP9xnc6KYW949S0Sm0ytiW/z5amdcSUmf+txM9uSzwhAgAgOkQAACTVaMi0FNKJo5j8tS/AHEOfvnqML2YsSRMnTkxlX5lGyle4ueiii1I5hsN8Ief4yF8a4edhC//sUsgLtau0OLaH1eOGut4OWuK7L40yjSF9H/nc3BVKUFt8lKmPbo73qbjClivdZyuF6uNxpQ2IK51PKm+O0JZ4QgQAQHSIAABIokMEAEBSjeYQS/HlOCXD83Weh4nDxwcMGJDKcYWGv/iLv0jlM888M6vr06dPKnteMuYJfcpEKc8Z80Mecy+tHE8OsT74kPdSPi4OeffXpfxfyd62Ec/1sAJNffIN1j0XWNpsOvL2GttcpRxiS+yYEfk54u5GcYxHS+MJEQAA0SECACCpRkOme8sfteOqGj7c+Igjjsjqjj322FT24cuStH79+lSutFGrlIcm4gLhHl6N7/Owgg83jsPyUbu83fl3X1qxKLZPb1t7G64qrWLj54znL4VrUR/83uH3ophi8vtN/K5LbcvbSKktlRaYL/Hr8uv3FXgkQqYAALQJOkQAAESHCACApHaQQ6y0CnscPu5TLeKUidLGvP4+3/li2bJl2XG+08aQIUOyOo+zx5h+peNQn3z4eylHE5e18nbc3GWzYlvyc5R2DIjn93w1u6rUJ18msJS7K42DKO3GUmm3lFK+u7m7tsRjvdzWmwXzhAgAgOgQAQCQ1A5Cps4f32NIqjQlw3fJiCutv/HGG6k8derUVF6wYEF23O23396sz44rR5R2IXAMh68PvtNAacWl2Aa9HZRCrSXNbT+l3S4ImdYn39XCv+u4Ea9/v6VwZwzHt8QKNCX+9+Crd8VpcK2NJ0QAAESHCACAJDpEAAAkteMcYoyde9w7xsfXrFmTyjGH0rt371T2ZYRi/N3zMKXl2Uo5mtJq86hd/n2Xhon7VKA4Laj0fVfKDcb3NDcfHadkeP6GHGJ98rybi+2gNL2nJXZVKS0fWLoPep3fj+OuRa2NJ0QAAESHCACApHYWMvXwQHzU9kfyGCrwR/Q4LcKHM/tqNEcffXR2nIdWY0jWw2hxWkdpODzqjw8T35NdJZobriqtEuLtp7TaTaxjt4v69/LLL6fy2LFjU/m8887Ljtu2bVsq721aphR29ftnqS3F++zMmTNT+YknnkjlxYsX79U17i3uwAAAiA4RAABJ7SBkWmlUnG8ILOWhytLmwfF8gwYNSuVvf/vbqezhU0nq27dvKsdQxOrVq1M5hkVLYS7Un9Io09JozlJI05VGkpZWEykt/B3/HlB/fCPz+++/P5WnT5+eHeebof/d3/1dVlcaYVwpVB/vdZ4uiu3MQ/Pxs5566qlU/uUvf1nxOlobT4gAAIgOEQAASXSIAABIagc5ROdx6Q0bNmR1voFmzK+sW7culWPusVevXqm8adOmVF66dGl2nMfwY+z83/7t31K5f//+WZ3njnyKB1Mw6pOvGFLKr5TE777S8PVSDrF0bGyfzb0u1C7Pz/m9aeXKldlxvnl5XC3J20jMK1e6H+3J9CFvZ3H6md9nSznK1l5JibsuAACiQwQAQFI7C5mW+OLbMWS0ZMmSVJ42bVpW56s++KoJO3bsyI7zR/m33347q/MQbQwVeJ2HTFlkuT75BsGxjXz44YepHL9fD0mVQpj+vhiuKoVFnacPJKb7tAf+3Xu7iGHRUkjf21YpZePnKG1mHVejcbENenstTS0iZAoAQBugQwQAQHSIAABIagc5xEqx89JOA5FPhXjzzTezuueffz6Vu3fvnspxQ84DDzxwt8dJeSw9xsD9PP4+ltOqH543POyww1J57dq12XE9e/bc7XukvO16vluq3HZj/sbblucrpXzD7NgGaWv1z++DpVxg6R5WymP7/dTvZ6UdV2Ieu9I1StLGjRtTubTJcGvjCREAANEhAgAgqR2ETCuFSeOwXt8Y0zdxjcf61AdJOvTQQ1PZQ0sxjFV6tPc6D11J+Yo6visG6odPW5g7d24qx7ZU2jy4UkhKysNLflwpJBVDsv467tSyatWqVI4rNaE+VAozxjbiq23Nnj07q/OdfRoaGrI6bzPePktt0O+5Ur6Czs9+9rOs7tlnn93t9bf19DOeEAEAEB0iAACS6BABAJDUDnKIlXacj3kSz9/47vaS1NTUlMpbtmzJ6rZu3ZrKHjuPOcPS8mw+vNmH3sfXnk9sbGzMjmMpt9rVp0+fVPbdBGI+znMqzd2ZQsrblrelOGzezxmXjfM84QsvvJDVrVixouK1oD5UWu4v5qNfeumlVI555kmTJqVy3JXH25bfi0pTduI9q1+/fqn84x//OKubMmVKKt95552pHNt4aTm4lsATIgAAokMEAECS1GlXWy8FAABADeIJEQAA0SECACCJDhEAAEl0iAAASKJDBABAEh0iAACS6BABAJBEhwgAgCQ6RAAAJNEhAgAgiQ4RAABJdIgAAEiiQwQAQBIdIgAAkugQAQCQRIcIAIAkOkQAACTRIQIAIIkOEQAASXSIAABIokMEAEASHSIAAJLoEAEAkCT9fzx3TYbKDO5YAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 504x432 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Validation set\n",
    "databunch.show_batch(rows=3, figsize=(7,6), is_train=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data labels and details of the final DataBunch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "databunch.classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LabelLists;\n",
       "\n",
       "Train: LabelList (54000 items)\n",
       "x: ImageList\n",
       "Image (3, 28, 28),Image (3, 28, 28),Image (3, 28, 28),Image (3, 28, 28),Image (3, 28, 28)\n",
       "y: CategoryList\n",
       "9,0,0,3,0\n",
       "Path: /home/jupyter;\n",
       "\n",
       "Valid: LabelList (6000 items)\n",
       "x: ImageList\n",
       "Image (3, 28, 28),Image (3, 28, 28),Image (3, 28, 28),Image (3, 28, 28),Image (3, 28, 28)\n",
       "y: CategoryList\n",
       "3,7,2,9,3\n",
       "Path: /home/jupyter;\n",
       "\n",
       "Test: LabelList (10000 items)\n",
       "x: ImageList\n",
       "Image (3, 28, 28),Image (3, 28, 28),Image (3, 28, 28),Image (3, 28, 28),Image (3, 28, 28)\n",
       "y: EmptyLabelList\n",
       ",,,,\n",
       "Path: /home/jupyter"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "databunch.label_list"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Looking at a few individual sample"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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\n",
      "image/png": "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\n",
      "text/plain": [
       "Image (3, 28, 28)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "databunch.train_ds[0][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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\n",
      "image/png": "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\n",
      "text/plain": [
       "Image (3, 28, 28)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "databunch.valid_ds[0][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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\n",
      "image/png": "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\n",
      "text/plain": [
       "Image (3, 28, 28)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "databunch.test_ds[0][0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Class distribution in the training and the validation splits (with percentages)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(Category 0, 5442),\n",
       " (Category 3, 5431),\n",
       " (Category 9, 5409),\n",
       " (Category 7, 5402),\n",
       " (Category 4, 5400),\n",
       " (Category 2, 5399),\n",
       " (Category 8, 5399),\n",
       " (Category 1, 5384),\n",
       " (Category 6, 5375),\n",
       " (Category 5, 5359)]"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_classes = Counter(databunch.train_ds.y)\n",
    "train_classes.most_common()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{Category 9: 0.16694444444444442,\n",
       " Category 0: 0.16796296296296295,\n",
       " Category 3: 0.16762345679012344,\n",
       " Category 2: 0.1666358024691358,\n",
       " Category 5: 0.1654012345679012,\n",
       " Category 7: 0.1667283950617284,\n",
       " Category 6: 0.16589506172839508,\n",
       " Category 4: 0.16666666666666669,\n",
       " Category 1: 0.16617283950617284,\n",
       " Category 8: 0.1666358024691358}"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_classes = {k: (v/54000)*100 for k, v in train_classes.items()}\n",
    "train_classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(Category 5, 641),\n",
       " (Category 6, 625),\n",
       " (Category 1, 616),\n",
       " (Category 2, 601),\n",
       " (Category 8, 601),\n",
       " (Category 4, 600),\n",
       " (Category 7, 598),\n",
       " (Category 9, 591),\n",
       " (Category 3, 569),\n",
       " (Category 0, 558)]"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "valid_classes = Counter(databunch.valid_ds.y)\n",
    "valid_classes.most_common()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{Category 3: 9.483333333333334,\n",
       " Category 7: 9.966666666666667,\n",
       " Category 2: 10.016666666666667,\n",
       " Category 9: 9.85,\n",
       " Category 8: 10.016666666666667,\n",
       " Category 4: 10.0,\n",
       " Category 1: 10.266666666666667,\n",
       " Category 6: 10.416666666666668,\n",
       " Category 5: 10.683333333333334,\n",
       " Category 0: 9.3}"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "valid_classes = {k: (v/6000)*100 for k, v in valid_classes.items()}\n",
    "valid_classes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Starting with a baseline (ResNet34)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner = cnn_learner(databunch, models.resnet34, \n",
    "                      metrics=[error_rate, accuracy], model_dir=\"/tmp/model/\").to_fp16()\n",
    "learner.lr_find();\n",
    "learner.recorder.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>error_rate</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.569271</td>\n",
       "      <td>0.563735</td>\n",
       "      <td>0.206833</td>\n",
       "      <td>0.793167</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.487702</td>\n",
       "      <td>0.525217</td>\n",
       "      <td>0.196667</td>\n",
       "      <td>0.803333</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.416583</td>\n",
       "      <td>0.403819</td>\n",
       "      <td>0.153500</td>\n",
       "      <td>0.846500</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.370473</td>\n",
       "      <td>0.320558</td>\n",
       "      <td>0.118167</td>\n",
       "      <td>0.881833</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.344508</td>\n",
       "      <td>0.289453</td>\n",
       "      <td>0.107667</td>\n",
       "      <td>0.892333</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.fit_one_cycle(5, max_lr=slice(3e-02, 5e-03))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.recorder.plot_losses()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>error_rate</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.560680</td>\n",
       "      <td>0.527295</td>\n",
       "      <td>0.166833</td>\n",
       "      <td>0.833167</td>\n",
       "      <td>00:38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.514742</td>\n",
       "      <td>0.372615</td>\n",
       "      <td>0.130667</td>\n",
       "      <td>0.869333</td>\n",
       "      <td>00:37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.406393</td>\n",
       "      <td>0.290933</td>\n",
       "      <td>0.106667</td>\n",
       "      <td>0.893333</td>\n",
       "      <td>00:37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.338881</td>\n",
       "      <td>0.245159</td>\n",
       "      <td>0.089833</td>\n",
       "      <td>0.910167</td>\n",
       "      <td>00:37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.281849</td>\n",
       "      <td>0.232122</td>\n",
       "      <td>0.084167</td>\n",
       "      <td>0.915833</td>\n",
       "      <td>00:37</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.unfreeze()\n",
    "learner.fit_one_cycle(5, max_lr=slice(3e-02))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Validation accuracy reported: **91.58%**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "learner.save('stage-1-rn34')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Looking at model's predictions (on the validation set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x864 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.show_results(rows=3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Where model is incurring the top losses?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interp = ClassificationInterpretation.from_learner(learner)\n",
    "losses,idxs = interp.top_losses()\n",
    "len(databunch.valid_ds)==len(losses)==len(idxs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x720 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "interp.plot_top_losses(9, figsize=(12,10), heatmap=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 600x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "interp.plot_confusion_matrix(figsize=(10,10), dpi=60)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The model is getting confused at distinguishing between the shapes. Also there is an amount of label noise present in the dataset. A bit of manual relabelling and clearing unnecessary images from the data _might be helpful_."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ea6307bcdd77463aab4904ec391063ad",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(VBox(children=(Image(value=b'\\xff\\xd8\\xff\\xe0\\x00\\x10JFIF\\x00\\x01\\x01\\x01\\x00d\\x00d\\x00\\x00\\xff…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c4724178fac441d7bb405b5d3495d052",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Button(button_style='primary', description='Next Batch', layout=Layout(width='auto'), style=ButtonStyle())"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from fastai.widgets import *\n",
    "\n",
    "ds, idxs = DatasetFormatter().from_toplosses(learner, ds_type=DatasetType.Valid)\n",
    "ImageCleaner(ds, idxs, main_path/'fashion_mnist')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Preparation of a new DataBunch with the cleaned set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "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>name</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>train/1.png</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>train/2.png</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>train/3.png</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>train/4.png</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>train/5.png</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          name  label\n",
       "0  train/1.png      9\n",
       "1  train/2.png      0\n",
       "2  train/3.png      0\n",
       "3  train/4.png      3\n",
       "4  train/5.png      0"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cleaned = pd.read_csv('/home/jupyter/fashion_mnist/cleaned.csv')\n",
    "cleaned.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "\n",
    "def clean_string(m):\n",
    "\n",
    "    re1='.*?'\t# Non-greedy match on filler\n",
    "    re2='(\\\\d+)'\t# Integer Number 1\n",
    "    re3='(\\\\.)' \t# Any Single Character 1\n",
    "    re4='(png)' \t# Word 1\n",
    "\n",
    "    rg = re.compile(re1+re2+re3+re4,re.IGNORECASE|re.DOTALL)\n",
    "    clean = rg.search(m)\n",
    "    if m:\n",
    "        int1=clean.group(1)\n",
    "        c1=clean.group(2)\n",
    "        word1=clean.group(3)\n",
    "        return str(int1) + str(c1) + str(word1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "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>name</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.png</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2.png</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3.png</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4.png</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.png</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    name  label\n",
       "0  1.png      9\n",
       "1  2.png      0\n",
       "2  3.png      0\n",
       "3  4.png      3\n",
       "4  5.png      0"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cleaned['name'] = cleaned.name.apply(clean_string)\n",
    "cleaned.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "databunch = ImageList.from_df(cleaned, path=main_path, folder='train')\\\n",
    "            .split_by_rand_pct(0.1)\\\n",
    "            .label_from_df()\\\n",
    "            .add_test(test_img)\\\n",
    "            .transform(get_transforms(do_flip=True), size=28)\\\n",
    "            .databunch(path='.', bs=128, device= torch.device('cuda:0'))\\\n",
    "            .normalize(mnist_stats)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>error_rate</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.292104</td>\n",
       "      <td>0.214714</td>\n",
       "      <td>0.084768</td>\n",
       "      <td>0.915232</td>\n",
       "      <td>00:26</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner = cnn_learner(databunch, models.resnet34, \n",
    "                      metrics=[error_rate, accuracy], model_dir=\"/tmp/model/\").to_fp16()\n",
    "\n",
    "learner.load('stage-1-rn34')\n",
    "\n",
    "learner.fit_one_cycle(1, max_lr = slice(3e-02))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I did not trained the network from scratch in this case. Instead, I used the previously saved weights from the earlier model. It can be seen that the score has not improved much. The model is failing to learn the shapes properly for certain categories, I think. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "learner.save('stage-2-rn34')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x720 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "interp = ClassificationInterpretation.from_learner(learner)\n",
    "interp.plot_top_losses(9, figsize=(12,10), heatmap=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The confusion is still there. However, I decided to make a quick submission to [this practice problem](https://datahack.analyticsvidhya.com/contest/practice-problem-identify-the-apparels/) at this stage. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "labels = []\n",
    "\n",
    "for i in range(len(databunch.test_ds)):\n",
    "    pred_class, _, _ = learner.predict(databunch.test_ds[i][0])\n",
    "    labels.append(pred_class)\n",
    "    \n",
    "labels = list(map(lambda x: str(x).replace('Category', ''), labels))\n",
    "\n",
    "sub_file['label'] = labels\n",
    "sub_file.to_csv('submission.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "id,label\r\n",
      "60001,0\r\n",
      "60002,0\r\n",
      "60003,0\r\n",
      "60004,0\r\n"
     ]
    }
   ],
   "source": [
    "!head -5 submission.csv "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}