lec_10_v1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "02092bef",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found cached dataset fashion_mnist (/Users/arun/.cache/huggingface/datasets/fashion_mnist/fashion_mnist/1.0.0/0a671f063342996f19779d38c0ab4abef9c64f757b35af8134b331c294d7ba48)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "60306f9f14a34137a26e2f925a331c58",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/2 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc:0.568, loss:1.415, training:True\n",
      "acc:0.639, loss:1.112, training:True\n",
      "acc:0.677, loss:0.974, training:True\n",
      "acc:0.681, loss:0.960, training:False\n",
      "{'accuracy': '0.592', 'loss': '1.355', 'epoch': 0, 'train': 'train'}\n",
      "{'accuracy': '0.666', 'loss': '0.911', 'epoch': 0, 'train': 'eval'}\n",
      "{'accuracy': '0.717', 'loss': '0.794', 'epoch': 1, 'train': 'train'}\n",
      "{'accuracy': '0.728', 'loss': '0.754', 'epoch': 1, 'train': 'eval'}\n",
      "{'accuracy': '0.755', 'loss': '0.689', 'epoch': 2, 'train': 'train'}\n",
      "{'accuracy': '0.743', 'loss': '0.696', 'epoch': 2, 'train': 'eval'}\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "<style>\n",
       "    /* Turns off some styling */\n",
       "    progress {\n",
       "        /* gets rid of default border in Firefox and Opera. */\n",
       "        border: none;\n",
       "        /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "        background-size: auto;\n",
       "    }\n",
       "    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>accuracy</th>\n",
       "      <th>loss</th>\n",
       "      <th>epoch</th>\n",
       "      <th>train</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0.598</td>\n",
       "      <td>1.182</td>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>0.684</td>\n",
       "      <td>0.883</td>\n",
       "      <td>0</td>\n",
       "      <td>eval</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "\n",
       "<style>\n",
       "    /* Turns off some styling */\n",
       "    progress {\n",
       "        /* gets rid of default border in Firefox and Opera. */\n",
       "        border: none;\n",
       "        /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "        background-size: auto;\n",
       "    }\n",
       "    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>accuracy</th>\n",
       "      <th>loss</th>\n",
       "      <th>epoch</th>\n",
       "      <th>train</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0.665</td>\n",
       "      <td>0.966</td>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>0.795</td>\n",
       "      <td>0.594</td>\n",
       "      <td>0</td>\n",
       "      <td>eval</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD8CAYAAABn919SAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAAsTAAALEwEAmpwYAAAoVElEQVR4nO3de5xdVX3//9fn3Oee24SEJCQBwl0QCOGmQPFSQCyVokV/giBtCgV+0ou21q/Qr36rtvUKqUTUqlTUL1VUwCgCNgpyTcLdIIZMQiLRTCaZ25k51/35/nHOjMNkJplJznXyfuaxH2efvdfs9ZmVeXxmzTpr72XujoiI1L9QtQMQEZHSUEIXEZkilNBFRKYIJXQRkSlCCV1EZIpQQhcRmSIi1ap41qxZvmjRompVLyJSl9auXbvD3dvHOle1hL5o0SLWrFlTrepFROqSmW0e75yGXEREpogJJXQz22Rmz5nZ02a2W7faCm42sw1m9qyZnVT6UEVEZE8mM+TyR+6+Y5xz5wNLitupwK3FVxERqZBSDblcBNzuBY8B08xsbomuLSIiEzDRhO7AT81srZktH+P8PGDLiPdbi8dew8yWm9kaM1vT2dk5+WhFRGRcE03oZ7r7SRSGVq41s7NGnbcxvma3xzi6+23uvtTdl7a3jznrRkRE9tGEErq7v1p83Q58H1g2qshWYMGI9/OBV0sRoIjIlPDqU7BxdVmr2GtCN7MmM2sZ2gfeCjw/qtjdwOXF2S6nAT3uvq3k0YqI1KvHboW7//+yVjGRWS4HAd83s6Hy33L3n5jZ1QDuvhJYBVwAbAAGgCvLE66ISJ3KJCHWXNYq9prQ3X0jcMIYx1eO2Hfg2tKGJiIyhWT6IdZU1ip0p6iISCVkkkroIiJTghK6iMgUkekv+xi6ErqISCWohy4iMkUooYuITAH5HORSEG8pazVK6CIi5ZbpL7yqhy4iUucyycKrErqISJ0bTuia5SIiUt805CIiMkVoyEVEZIpQQhcRmSKGh1w0hi4iUt/UQxcRmSKU0EVEpoham7ZoZmEze8rM7h3j3Dlm1mNmTxe3G0sbpohIHcv0QTgO4WhZq5nIEnRDPgCsB1rHOf+Qu1+4/yGJiEwxFXgwF0ywh25m84G3AV8pbzgiIlNQBdYThYkPuXwe+BAQ7KHM6Wb2jJn92MyO3e/IRESmigqsJwoTSOhmdiGw3d3X7qHYOmChu58A3AL8YJxrLTezNWa2prOzc1/iFRGpPzU05HIm8Cdmtgn4DnCumX1zZAF373X3/uL+KiBqZrNGX8jdb3P3pe6+tL29ff+jFxGpB7WS0N39w+4+390XAZcCP3P3944sY2ZzzMyK+8uK1+0qQ7wiIvWnQmPok5nl8hpmdjWAu68ELgGuMbMcMAhc6u5emhBFROpchcbQJ5XQ3X01sLq4v3LE8RXAilIGJiIyZdTKkIuIiOynTBLitTNtUURE9kWQh+wA+UgT5R6JVkIXESmn4nNcVv26lzf+2/+UtSoldBGRciom9N4gTmMsXNaqlNBFRMppKKHnYzTH93li4YQooYuIlFNxtaJduRjNifI+bVEJXUSknIo99F25GC3qoYuI1LFiQt+ZidIU1xi6iEj9Kg657MhGaY5ryEVEpH4Ve+g70hGaExpyERGpX8WE3k+DxtBFROpapg+AARLqoYuI1LVMEg9FyRLRPHQRkbqWSZKPNAKohy4iUtcySXLFhK4xdBGRepbpJxdWD11EpP5lkqRDDQC1M4ZuZmEze8rM7h3jnJnZzWa2wcyeNbOTShumiEidGpHQW2roxqIPAOvHOXc+sKS4LQdu3c+4RESmhkw/KUsA1Mat/2Y2H3gb8JVxilwE3O4FjwHTzGxuiWIUEalfmSQD1kBDNEwkXN5R7ole/fPAh4BgnPPzgC0j3m8tHhMRObBlkiS9/DcVwQQSupldCGx397V7KjbGsd0WzzOz5Wa2xszWdHZ2TiJMEZE6le4vJPQyfyAKE+uhnwn8iZltAr4DnGtm3xxVZiuwYMT7+cCroy/k7re5+1J3X9re3r6PIYuI1IkggGySvqD8qxXBBBK6u3/Y3ee7+yLgUuBn7v7eUcXuBi4vznY5Dehx922lD1dEpI5kB4DCeqKVSOj7XIOZXQ3g7iuBVcAFwAZgALiyJNGJiNSz4pMWe/LxioyhT6oGd18NrC7urxxx3IFrSxmYiEjdKy5u0Z2Llv22f9CdoiIi5TO0/Fw2WhuzXEREZB+NTOjqoYuI1LFiQu+t0Bi6ErqISLkUx9ArsfwcKKGLiJRPsYdeieXnQAldRKR8ij30pMdpLj5p8Z8f+Wfu23RfWapTQhcRKZdiQh+gcOt/Lshx12/uYkP3hrJUp4QuIlIumSSBhUkTpSURYVdqF44zKzGrLNUpoYuIlEsmWVx+zmiKR9gxuAOAmQ0zy1KdErqISLlk+smG/7D8XFeqC1BCFxGpP5kkmVBhgeiWRISuwWJCTyihi4jUl0ySdChBJGTEIyH10EVE6lYmyaA10JyIYGZ0DXaRCCdojDSWpToldBGRcsn0D09ZBOhKdTGzYSZmYy3ytv+U0EVEymVoPdFiQt8xuKNswy2ghC4iUj7pfvo9Tkvxtv+uwa6yfSAKSugiIuWTSdI3Yvm5namd1e2hm1nCzJ4ws2fM7AUz+99jlDnHzHrM7OnidmN5whURqRPukOkvPjo3Si7IsSu1q6w99Ik8/isNnOvu/WYWBR42sx+7+2Ojyj3k7heWPkQRkTqUHQSc7nyM5niE7nQ3jpe1h77XhF5cL7S/+DZa3LxsEYmITAXFR+d252KvualoVkN5nuMCExxDN7OwmT0NbAfud/fHxyh2enFY5sdmduw411luZmvMbE1nZ+e+Ry0iUuuGF4gu9NCHn+NS7Q9F3T3v7q8H5gPLzOy4UUXWAQvd/QTgFuAH41znNndf6u5L29vb9z1qEZFaV+yhJ0nQVIHnuMAkZ7m4ezewGjhv1PFed+8v7q8ComZWvr8rRERq3fBqRXFa4uV/jgtMbJZLu5lNK+43AG8GXhxVZo4Vb30ys2XF63aVPFoRkXoxtJ6oF2797xrsIh6O0xRtKluVE5nlMhf4hpmFKSTqO939XjO7GsDdVwKXANeYWQ4YBC4tfpgqInJgGrmeaDxCV3fhpqJy3fYPE5vl8ixw4hjHV47YXwGsKG1oIiJ1bGg9UeLDPfRyznAB3SkqIlIeQz10TxTG0FNdzGiYUdYqldBFRMphuIeeoDlRmLZYzg9EQQldRKQ8MkkcI0WMhqjRne4u65RFUEIXESmPTJJMcYHorPcTeKAeuohIXcr0kwk10ByPsDNd/puKQAldRKQ8MklSVpyymCr/c1xgYvPQRURksobWE63QXaKgHrqISHlkkn+4qWhQQy4iIvUr0z+8nmhXqotYKEZztLmsVSqhi4iUQyZJn8eHe+gzG8p72z8ooYuIlEe6v7CeaKLQQy/3+DkooYuIlEcmSW9x+blKPMcFlNBFRErPHc/0/2H5uVRX2T8QBSV0EZHSy6Uxz5P0BI0xY2dqJzMS5X0wFyihi4iU3ojl58LRwcJt/+qhi4jUoeKTFgeI46E+oPxz0EEJXUSk9IZ66J4gCPUC5b9LFCa2pmjCzJ4ws2fM7AUz+99jlDEzu9nMNpjZs2Z2UnnCFRGpA8NDLg1kvdBDr8Qsl4k8yyUNnOvu/WYWBR42sx+7+2MjypwPLClupwK3Fl9FRA48Q4tbeJyU9wA1MuTiBf3Ft9HiNnoB6IuA24tlHwOmmdnc0oYqIlInRiwQPZjvJhqK0hJtKXu1ExpDN7OwmT0NbAfud/fHRxWZB2wZ8X5r8djo6yw3szVmtqazs3MfQxYRqXEjlp/ry+2syG3/MMGE7u55d389MB9YZmbHjSoyVqSje/G4+23uvtTdl7a3t086WBGRujA0y8UT9GZ2VeQDUZjkLBd37wZWA+eNOrUVWDDi/Xzg1f0JTESkbhWHXLKRRnald1Zk/BwmNsul3cymFfcbgDcDL44qdjdweXG2y2lAj7tvK3WwIiJ1oZjQI7HGij3HBSY2y2Uu8A0zC1P4BXCnu99rZlcDuPtKYBVwAbABGACuLFO8IiK1L5MkbQkaGyLsTO2s2JDLXhO6uz8LnDjG8ZUj9h24trShiYjUqUw/qVADjYkMuzxfO0MuIiIySZkkgySIxwtDLzX5oaiIiExAJkmSBmJDCV09dBGROpXpJ+lxwpHC9EX10EVE6lUmSV8Qx4YSunroIiJ1Kt1PbxAnCPURDUVpjbVWpFoldBGREvNMP/1BnLz1MSMxoyK3/YMSuohIyXkmSZIEae+u2HALKKGLiJScZZIMkCAV9FTsA1FQQhcRKa1cBguyJD3BQF49dBGR+jX86NwY/bnuij3HBZTQRURKq/hgru5QmMDzGnIREalbQwk9XJjZoiEXEZF6VUzofeEAqNxdoqCELiJSWsUx9P5wHlAPXUSkfhUT+mA0B6iHLiJSv4pDLkEsRyQUoTVemdv+QQldRKS0ij30fCzLjMQMQla5NDuRNUUXmNn/mNl6M3vBzD4wRplzzKzHzJ4ubjeWJ1wRkRo31EOPpCo63AITW1M0B/ydu68zsxZgrZnd7+6/GlXuIXe/sPQhiojUkaGEHh5gZsMhFa16rz10d9/m7uuK+33AemBeuQMTEalLmX7SxMmH+ireQ5/U4I6ZLaKwYPTjY5w+3cyeMbMfm9mx43z9cjNbY2ZrOjs7Jx+tiEityyRJWoKc9VZ0yiJMIqGbWTPwPeAGd+8ddXodsNDdTwBuAX4w1jXc/TZ3X+ruS9vb2/cxZBGRGpZJ8ntL4OQr+hwXmGBCN7MohWR+h7vfNfq8u/e6e39xfxUQNbPKficiIrUgk+TVUByo7Bx0mNgsFwO+Cqx398+OU2ZOsRxmtqx43a5SBioiUg883c+2UBSo7F2iMLFZLmcClwHPmdnTxWP/BBwC4O4rgUuAa8wsBwwCl7q7lz5cEZHaFqT76AxHgGztTVt094eBPS6I5+4rgBWlCkpEpF4F6X52hAuptWY/FBURkQnIJNkZNkKEaIu3VbTqiQy5yFQXBJAbhMwAZJMQaYDGmRDWj4fIZFkmSU/TdFqi0yt62z9MsYTu7qTzaRwnGooStjDFz2prjruTC3Kk82lyQY5oOEosFCMSiuw15q5X1pO58yryFnnNliNK3sLkLUrYoCGUJxHKEbccMXKEyTDgGfrIsis3QDY3SHMmRVNukKbAaQoCGtyH/2zLRFvJxGeQic8gm5hJLj6DTLiR/GAPPtiDZXoJZ/qI5fpJBEkafIAcUdKhBnKRRoJII8SaCCVaiCaaCSVaSIWbGQi3kLQm+mmml0a6vZHeIEE2myafTuHZFJ4dxLODkEthuRQRC0hEQjREQySiIRKRP7zGY1ESTS00tkyjuWU6bdOmkWiaBvEWiDVDJA4WhlAYim3r7mTzTi4IiIVDRML6Y1VKI5RN0hsJmBafUfG66y6hP7HtCW595lZSuRSDuUFS+cLrYG6QVC6F89rPYiOhCNFQlIhFiIQihENhAg9wdwICAg/AGd6PhWMkwgkaIg0kIgkS4cTwazgUJhfkyAZZckHuNfvZIIu7M/yv+Jnw0H7e86TzadL5NNl8dvgXz2iGEQ1FiYVjw1s8HCcRThCPFF7zqQz5hhRxd2LuYMU6LSjUaE4OyGH0hUL0haAvDAMh8Nf8rmgobtNfE0M4CBMLjKhDxJ04XcS8k8RgnhgB5iEsEiIUiRBqDBO2ZsI2k0g4RjhwwtkcsXyORC5LU24XLX3bae3JMs3TzPRB8iEvxhWiNxSiL2TD+2kzwkDYnXAEImGIxL3wHmgInIaU0+ABDYHT6E5D8RdRLzBgxmAoxIAZAyFj0EIMhIy0GY2BFzdoDJxEYMSDMPl8A7mgiZy3kPFpJENt9IenMRBuJRIJkwg58bCTCDvxUEAsFBAPOdFYA9HpB9M0cz7TDlpI+8GH0NrUWLKfdalD+SyhIEN/OGBBhT8QhTpM6GaG47Ql2pgbmfua5Dv0athuiTfnhde85wkRwswIWQij+GqGYWTyGVL5FOlcmsF84ZdEKpeiO91d6EmHooVfEMVfFA2RhsJfA6Hw8J9Xhg1fzzAwhr8uHo4TD8eHE3UsXOiVZ/NZMkGGTD7zh9fils6nh2NK5VOkQmlS8+aTzqXJBlnCoTBhCxMJRQhZqLBvEcxCNIWamRlqJmothL0JzzdhQROtsTaaYnFC4TShcAYsTWApAtKk8oMM5gbI5LNkgyzZIEcuyBY2z4HlccuTCQrnR8Y89D6dT4/6n4sWt+bd/k8jFqY1nKA10kA8FCMP5HHyBOTdyXlA3gNyniOVT5Pa7dp7FvUwETeyFpCzYJxSqeLWSWMQ0FLcWoOAlsBpDAKag4CmrNPoAc3Fv2iatgc0BAGhfJ5QLs+OoJXu8AwGYjOxUJhICMIG4ZAVXg3CIcglZpGZcQTefhSRucfS0H4obY1xEtFQzf5VKRNQfI7LYDjHrAp/IAp1mNBPmXMKXz/v69UOQ/YiHxT+IhnIDQz/BTX0V1Q8HKc11kpLrIWWWAvxcHxSSSzwgFQuVbh2dnC4DoCGSAON0cbCa6SRRCTxmnHMTD5DMpukP9NPf7a4Ffd7M730ZnrpS/fSN7iTvtRO+rJJtueSJHMp+nMDDOQG9/gLJeYhpuVhZr6TaTmjISg83S479FeTQR4I5bYQ63yCOdtyLFqbY04WLDODZHYeXeFDiDa00tjcQnNzK22tbUxva2Xm9Om0z5hOfPrB0NQ+PHwkNSSTxIFUOMPsxsrfW1l3CV3qQzgUpjHUSGO09EMQIQvRGC1eu2FyXzs0jDU9MX3vhceRC3Iks0kGsgP0Z/vZmdpJ52AnOwZ20DnYWdgf3MG2gU6S2eQfhv1CEcIWIUQEszADuRy/Gfgtd+eHnqQRYP4Kc3MdLMxlac/liXZDtLswtBZzJ+oQcyfiIZq8hdZQO22JBSTaFhGZuZCmOYcza+GxtEzTjdpVkUnSGzKCkDO3ufKPN1FCF5mkSChCW7ytZFPSBrIDbO7dzKbeTWzq3cTm7o1s6unglXR3cfgqOzy8lWf0kNEOYAfTUmuYtznH/A055q3OMTcTYWa+nUR0Edm2Q2HWEhrnHsmSJUcxc0blhwIOGJl+usJhANrVQxc58DRGGzl65tEcPfPovZZ1d7JBlmQ2yavJV/lt32/Z3LuFl3dsZMuul3lq8FUeyHeTx4EBmoMXODa9jtdtTnPcSxlyP83Qn4+TjLdDy8E0ty+gceYCrHUutB8FC06FSKz83/RUNSKhV/qmIlBCF6krZvaaYaNjZ+7+pOpckGNjz0ae3/F8Yet8hq91v0zeC6vQt+ajLEiFODL9O07q2MDpv97F7KCwoHEQacQWvxE77Fw47FyYtURj9ZORSf4hoWuWi4jsr0gowhHTj+CI6Udw8ZKLAUjlUvx616+Hk/wLO17g+72buKs4dTWca6N1sJnj0mnO3/Icb9rwUxrdyTUfTGTJm+DQc2DO62DGoRCOVvX7q2mZJL8rJvTZjbMrXr0SusgBIBFJcEL7CZzQfsLwsf5MP+t3rudXXb/i+R3P8/T253hoYAcPEQE/hNZ0E0en87zlN/ey7Plvsyibg1AEm3k4tB9ZGKKZdUThtf0o3VkMkOmnIxYl6k0Vv+0flNBFDljNsWZOmXMKp8w5ZfhYT7qHZzuf5ZnOZ3jst0/x9K4XeDxoAVqI5KMcMtjIH+fC/Nmmp5i9/h7Mix/Sts6DU/8KTnofNEyryvdTEzJJOqIRmphbleqV0EVkWFu8jTfOfyNvnP9GrjuxMOe/o6eDdb9/ip91PMnazse4NdjJF4MoQd95nBBewnunt/GmgQdJ3H8jrP5XOOkyOO0amL6o2t9O5WWSdESjNEfmV6V6JXQRGVfIQhw27TAOm3YY7zzyEgIPWPv7tfzfX93Dz3/7IM8Hz/MPqTi5gWN5yyH/wEejLzH7ya/AE7fBURfC6dfBIadW+9uomJ2DXXSHwxwSXVCV+pXQRWTCQhYaHqbJBTfyxLYnuOule/nZlgdZ7ev4WbqVQw5+Dzc1BZzacQ+2/m6YtxSW/SUc86cQTVT7WyirjYOdALQ31GhCN7MFwO3AHCAAbnP3L4wqY8AXgAuAAeAKd19X+nBFpFZEQhHOmHcGZ8w7g0w+w30dP+O2df+XTQM/5y8zTmLWyVzRPIertj9O4vt/Bff9E5x4GSy9csoOx2xM7wRgbuPCqtQ/kWeG5oC/c/ejgdOAa83smFFlzgeWFLflwK0ljVJEalosHOPth5/HPe/6Gqsuvo8/mn052dAuVqYf5pSWCMuPeAe/mXciPHIzfOH18K0/h9/cX3gW/xSyMdNDLHDmNlXnQ9G9JnR33zbU23b3PmA9MG9UsYuA273gMWCamVXnOxKRqlrQOpebz/8ga694kGuP/nfa/DgeST/NO7Iv894jrmD7KdfBb9fBHZfALSfC41+C7GC1wy6JTfkkc7IhWhPVudt2Uk/1N7NFwInA46NOzQO2jHi/ld2TvogcQMKhMFcvO4+H3/+ffOmcHzA3fAbPZB7gj7b9gvfO/SCvnLsCmg+CH38IPn88PPx5SPdVO+z9sok0szNhmuPV+XhywgndzJqB7wE3uHvv6NNjfMluqzeY2XIzW2Nmazo7OycXqYjUrTMXL+L+y77Ip9/wJaYnmngmfwtvffYe/iz/ATZccCccdCw8cBN87jhY/SkY2FntkCctlUvxqgXMyMRpTtRwQjezKIVkfoe73zVGka3AyI915wOvji7k7re5+1J3X9reXvlHS4pIdf3xYWfwP++5m78+4QM0tG7kpehHufAXD/Afh3yK4KoHYeEZsPqT8PnXwf03QX/9dPw2927GDdqyidrtoRdnsHwVWO/unx2n2N3A5VZwGtDj7ttKGKeITBHRUJRrXv8XrLr4Hs5a8Aai7T/hP166lkvu62D7274GV/8SlrwVfvkF+OKp8PsXqh3yhHT0dgDQnG6q3YQOnAlcBpxrZk8XtwvM7Gozu7pYZhWwEdgAfBn46/KEKyJTxdzmufzHm29mxbkraGvK81LkX3jLVz/Bg7tmwTu/Btc8AuEYfONPYPv6aoe7Vx09HZg78WxL7Q65uPvD7m7ufry7v764rXL3le6+sljG3f1adz/M3V/n7mvKH7qITAVnLzibVX/2fZbOPpVgxve59sEP8JEfPk565pHwvnshFIFvvB22v1jtUPeoo3sjB+fyZILGmu6hi4iU1fTEdL52/kr+5qS/J9bya37Q+Xecv/K/2BAcBO+7B7BCUu98qdqhjqujewOLs1kGQo3EI9VJrUroIlITzIz3v+59fPttdzC7uYnO5s/xjjs+xjOpWXDFvYAXkvqODdUOdTeBB2zq3czibJYtkYWTWvS8lJTQRaSmHDvrWO65+HucPf/NMOMnvPdHV/FoMlboqQc5+MaF0PVytcN8jd8lf0cqyLI4m+WVxBFVi0MJXURqTnOsmVve9GluOOEjEN/M8geu5LHBKLzvbshnCj31nRurHeawjp7CDJfpQRuemFG1OJTQRaQmmRlXvf5S/u3ML0I4yfIHruKJdBwu/yFkB+Drb4fuV6odJvCHhO6+iJYqfSAKSugiUuPOP+I0PnXGCjw0wF/cfxVrcg1w+d2FxwTc8S5I9VQ7RDZ2ractn2crS6o2ZRGU0EWkDrztyFP5xOkrcBvk/T99P+u8Ef78duj6Ddx5OeSzVY2vY8evWJzN8XxwaNWmLIISuojUibcfdSr/57QVBKS58idX8lTDPHj7zbBxNdx7A/huj4+qmI7kVhZns6zLLlIPXURkIi46ZhkfW3YLebJc8ZMrWTt7KZz1QXjqm/DQZ6oSU0+6h658isWRFrZl4hpDFxGZqIuPW8ZNS28hH+S44r4r+X77BfC6d8LPPg7Pfbfi8Wzq3QTAorZDSWUDDbmIiEzGO48/hc+ffRtm8NEnruNzrVfih5wOP7gGNj9a0Vg2bn8OgJltxwEwqyVe0fpHUkIXkbr05sOP5/bzv0o4kuHLGz7JDfY35Fvnw3feU9Ebjzq2PUnUnU47HoClC6dXrO7RlNBFpG69fs7R/OtZ/4dw42Z+2nMXl6c+SC5wuOOdFVsko2PXb1iYzbK652CmN0Y5fHZzReodixK6iNS18xafx+XHXE5k+iO8GNvCZckbyHdvKUxnDPJlr3/T4HYWE+MXW7KcsmhG1Z7jAkroIjIF3HDyDZx80MlY+3+TWjiff0xfAZsewss88yWbz7IlSLMgcRCbuwZYtrh6t/2DErqITAHRUJRPn/1pWmOtpGd+jdgpl/DD/Bn4/3wKXnmsbPW+sv1Z8gaNkUUASugiIqUwq2EWnznnM2zrf5Welv/iyWP+ia3BDPq/dQUMdpelzo5Xfg7AzuwSmmJhjpnbWpZ6Jmoia4r+p5ltN7Pnxzl/jpn1jFie7sbShykisncnzj6Rvz/l7/n51p8z/+hfcfu8G4kPbmfbHX9VljtJO37/NACPdx3JSQunEwlXt488kdq/Dpy3lzIPjVie7mP7H5aIyL55z1Hv4W2Hvo0vPvMfnHneYr7TfBlzt/6El396a8nr6ujp4KAAnv59lGWLqjvcAhNbU/QXQGXm/4iI7Ccz48bTbuTw6Yfz0Uc/zNLLb2Bd+AQOfvQmXv7V2pLW1ZHZxTwK0xSrPX4OpRtDP93MnjGzH5vZsSW6pojIPmmMNvL5cz5PPsjz8TUfYdb7vkKKBMGdV7Jle2n6p97fSUfIaaKdWDjECQumleS6+6MUCX0dsNDdTwBuAX4wXkEzW25ma8xsTWdnZwmqFhEZ2yGth/DxMz/Oczue49u/+z4DF9zCEjbz5JevY2cys9/X377p5yRDIXoGF3DCgjYS0XAJot4/+53Q3b3X3fuL+6uAqJnNGqfsbe6+1N2Xtre372/VIiJ79OaFb+a9R7+Xb67/Jr86qInfHX0lF2d/xBe/tIJUdv9uOurY+ggAz+9cVBPDLVCChG5mc6x4a5SZLStes2t/rysiUgp/e/Lfcvys47nxlzeSeet19E47mr/u+Qz/dd8v9+u6HV0vANCbms8pNfCBKExs2uK3gUeBI81sq5ldZWZXm9nVxSKXAM+b2TPAzcCl7lV80ryIyAjRcJR/P/vfCVmIv3v4w8Tf/VUaws4ZT17Ppm37PvTb0beVBg9h+RZOruIDuUaayCyXd7v7XHePuvt8d/+qu69095XF8yvc/Vh3P8HdT3P3R8oftojIxB3cfDCfeMMneHHni/zrxu+RvuhLHM1mfnf7VXgQTP6C/Z10kKYt38wxB7fRkoiWPuh9oDtFReSAcPaCs3n/ce/nv1/6bx5uMZ464npOG/w5L37v45O/2Lan2RiNkh5oZ9mimaUPdh8poYvIAeP6E6/npNkn8bFHP0br+e/j57GzOPKFzzHw/L2Tuk5y6xNsj0TYmVrIssW1MdwCSugicgCJhCL821n/RkOkgQ899EES7/oCLwQLCd/1l9D56wlfZ9OrTwKQyhxcMx+IghK6iBxgDmo6iE++8ZO83P0yd736JVYd8+/05SOkv/nnMLhrQtfYuOslAOY1LWRmc/WWnBtNCV1EDjhnHHwG1594Pas6VjEw/3E+FP57Qj1b8O9etfdFMfp+T0e+H3M4bcGRlQl4gpTQReSA9Bev+wuuOPYK7nr5TqIn7+J/Zd+HvfwgPHDTnr9w29N0RKNEM62cfujsygQ7QUroInJAMjP+9uS/5dIjL+WRHd/jiUPjfIc/hkdugYc/B/ncbl/j7tzx4nf4RUMD6dQhnFIjd4gOUUIXkQOWmfHhUz/MOw5/Bzsi9/Kx5gU83/JGeOCf4bazYfOjw2V3DO7gmgev4VO71nBS2mhJvZt50xqqF/wYItUOQESkmkIW4qbTbyKdT7OKVbzjdxfyo7PfzRFPfQK+dh4cfymrjzufG9d9loHcAB/pz9OyYzHNRx1S7dB3ox66iBzwwqEw//KGf+Gc+eeSmHMvf7rmN/zo7B8yeOYH+PirD3D9ox/lIMLcec7NXNr5W9ZmFtfMA7lGUg9dRITCHPXPnvNprr7/ep7wu/ibn/Uya84L9LU0coVN5/oX1xF75TIAngsW854amn8+RD10EZGiaDjKF9/yBZbNXUZ89v30pPtZkL6Byy66j9il34JoI2lrYFvDEg5rb6p2uLtRQhcRGSEejnPLuTfz0dM+yv96/Vfo2DKPt634JU/ET4frnuDS+M0ct3gexaeG1xQNuYiIjNIYbeRdR74LgJPmz+fqb67l3V9+jL8661Ce6m7ixjfU3nALqIcuIrJHR85p4YfXncmbjprNF1e/DNTGgtBjUQ9dRGQvWhNRvnTZyXz5oY2s29zN0XNbqx3SmJTQRUQmwMxYftZh1Q5jjyayBN1/mtl2M3t+nPNmZjeb2QYze9bMTip9mCIisjcTGUP/OnDeHs6fDywpbsuBW/c/LBERmayJrCn6C2DnHopcBNzuBY8B08xsbqkCFBGRiSnFLJd5wJYR77cWj+3GzJab2RozW9PZue+rbYuIyO5KkdDHml3vYxV099vcfam7L21vby9B1SIiMqQUCX0rsGDE+/nAqyW4roiITEIpEvrdwOXF2S6nAT3uvq0E1xURkUnY6zx0M/s2cA4wy8y2AjcBUQB3XwmsAi4ANgADwJXlClZERMZn7mMOd5e/YrNOoBvoGXG4bQ/vR+7PAnaUMJzR9e5v+fHOj3V8T9/z3t6Xsh1qqQ1GH1Mb7P5+KrbBWMfrISfsaxuMd25vbbDE3dvGvJq7V20Dbpvo+1H7a8oZx/6WH+/8WMcn0wblbIdaaoM9/d8fqG2wpzaZKm2wt++5Um0w2XbY1zbYl//3vV2v2g/numcS70efK2cc+1t+vPNjHZ9MG0yk7n1VS20w+pjaYPf3U7ENxjpeDzlhX9tgvHP7nAOqNuSyP8xsjbsvrXYc1aZ2UBuA2gDUBkOq3UPfV7dVO4AaoXZQG4DaANQGQJ320EVEZHf12kMXEZFRlNBFRKYIJXQRkSliyiV0MzvHzB4ys5Vmdk6146kWM2sys7VmdmG1Y6kGMzu6+DPwXTO7ptrxVIuZ/amZfdnMfmhmb612PNVgZoea2VfN7LvVjqXcaiqhj7c6kpmdZ2a/Lq6K9I97uYwD/UCCwoPD6kqJ2gDgH4A7yxNleZWiDdx9vbtfDbwLqMvpbCVqhx+4+18CVwB/XsZwy6JEbbDR3a8qb6S1oaZmuZjZWRSS8e3uflzxWBh4CXgLhQT9JPBuIAx8ctQl3g/scPfAzA4CPuvu/1+l4i+FErXB8RRuhU5QaI97KxN9aZSiDdx9u5n9CfCPwAp3/1al4i+VUrVD8es+A9zh7usqFH5JlLgNvuvul1Qq9mqoqUWi3f0XZrZo1OFlwAZ33whgZt8BLnL3TwJ7Gk7YBcTLEmgZlaINzOyPgCbgGGDQzFa5e1DeyEunVD8H7n43cLeZ/Qiou4Reop8FAz4F/LjekjmUPCdMeTWV0Mcx1opIp45X2MwuBv4YmAasKGtklTOpNnD3jwCY2RUU/2Ipa3SVMdmfg3OAiyn8Ul9VzsAqbFLtAFwPvBloM7PDvfCE1Ho32Z+FmcC/ACea2YeLiX9KqoeEPuEVkQDc/S7grvKFUxWTaoPhAu5fL30oVTPZn4PVwOpyBVNFk22Hm4GbyxdOVUy2DbqAq8sXTu2oqQ9Fx6EVkdQGoDYYonZQG4yrHhL6k8ASM1tsZjHgUgqrJB1I1AZqgyFqB7XBuGoqoRdXR3oUONLMtprZVe6eA64D7gPWA3e6+wvVjLOc1AZqgyFqB7XBZNXUtEUREdl3NdVDFxGRfaeELiIyRSihi4hMEUroIiJThBK6iMgUoYQuIjJFKKGLiEwRSugiIlOEErqIyBTx/wAMcZCo5HmF/AAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "from torch.utils.data import DataLoader, default_collate\n",
    "from datasets import load_dataset\n",
    "import matplotlib.pyplot as plt\n",
    "from functools import partial\n",
    "import fastcore.all as fc\n",
    "import torchvision.transforms.functional as TF\n",
    "import torch.nn.functional as F\n",
    "from operator import itemgetter, attrgetter\n",
    "import torch.optim as optim\n",
    "from functools import partial\n",
    "from copy import copy\n",
    "from torcheval.metrics import MulticlassAccuracy, Mean\n",
    "from fastprogress import master_bar, progress_bar\n",
    "from miniai.lec_9_v4 import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "403c2ab3",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found cached dataset fashion_mnist (/Users/arun/.cache/huggingface/datasets/fashion_mnist/fashion_mnist/1.0.0/0a671f063342996f19779d38c0ab4abef9c64f757b35af8134b331c294d7ba48)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "329f56ab06734584969ee036b6da978f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/2 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "x,y = 'image','label'\n",
    "name = \"fashion_mnist\"\n",
    "dsd = load_dataset(name)\n",
    "bs = 1024\n",
    "\n",
    "def inplace(f):\n",
    "    def _f(b):\n",
    "        f(b)\n",
    "        return b\n",
    "    return _f\n",
    "\n",
    "def get_dls(train_ds, valid_ds, batch_size, **kwargs):\n",
    "    return (\n",
    "        DataLoader(train_ds, batch_size=batch_size, shuffle=True, **kwargs),\n",
    "        DataLoader(valid_ds, batch_size=batch_size*2, shuffle=False, **kwargs)\n",
    "    )\n",
    "\n",
    "def collate_dict(x):\n",
    "    g = itemgetter(*x.features)\n",
    "    def _f(b):\n",
    "        return g(default_collate(b))\n",
    "    return _f\n",
    "\n",
    "class DataLoaders:\n",
    "    def __init__(self, *ds):\n",
    "        self.train,self.valid = ds[:2]\n",
    "    \n",
    "    @classmethod\n",
    "    def from_dd(cls, dd, batch_size, as_tuple=True, **kwargs):\n",
    "        f = collate_dict(dd['train'])\n",
    "        return cls(*get_dls(*dd.values(), batch_size=batch_size, collate_fn=f, **kwargs))    \n",
    "\n",
    "@inplace\n",
    "def transformi(b): b[x] = [TF.to_tensor(o) for o in b[x]]\n",
    "\n",
    "tds = dsd.with_transform(transformi)\n",
    "dls = DataLoaders.from_dd(tds, bs, num_workers=4)\n",
    "dt = dls.train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "fae8e5f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "def conv(ni, nf, ks=3, act=True):\n",
    "    res = nn.Conv2d(ni, nf, stride=2, kernel_size=ks, padding=ks//2)\n",
    "    if act: res = nn.Sequential(res, nn.ReLU())\n",
    "    return res\n",
    "\n",
    "def cnn_layers():\n",
    "    return [\n",
    "        conv(1 ,8, ks=5),        #14x14\n",
    "        conv(8 ,16),             #7x7\n",
    "        conv(16,32),             #4x4\n",
    "        conv(32,64),             #2x2\n",
    "        conv(64,10, act=False),  #1x1\n",
    "        nn.Flatten()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "f226a9c7",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "<style>\n",
       "    /* Turns off some styling */\n",
       "    progress {\n",
       "        /* gets rid of default border in Firefox and Opera. */\n",
       "        border: none;\n",
       "        /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "        background-size: auto;\n",
       "    }\n",
       "    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>accuracy</th>\n",
       "      <th>loss</th>\n",
       "      <th>epoch</th>\n",
       "      <th>train</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0.149</td>\n",
       "      <td>2.328</td>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>0.100</td>\n",
       "      <td>2.303</td>\n",
       "      <td>0</td>\n",
       "      <td>eval</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from torcheval.metrics import MulticlassAccuracy\n",
    "\n",
    "metrics = MetricsCB(accuracy=MulticlassAccuracy())\n",
    "cbs = [TrainerCB(1), metrics, ProgressCB(plot=True)]\n",
    "\n",
    "def fit(model, epochs=1, xtra_cbs=None):\n",
    "    learn = Learner(model, copy(dls), loss_func=F.cross_entropy, lr=0.6, cbs=cbs+fc.L(xtra_cbs))\n",
    "    learn.fit(epochs)\n",
    "    return learn\n",
    "\n",
    "learn = fit(nn.Sequential(*cnn_layers()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "e437a27b",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "<style>\n",
       "    /* Turns off some styling */\n",
       "    progress {\n",
       "        /* gets rid of default border in Firefox and Opera. */\n",
       "        border: none;\n",
       "        /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "        background-size: auto;\n",
       "    }\n",
       "    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>accuracy</th>\n",
       "      <th>loss</th>\n",
       "      <th>epoch</th>\n",
       "      <th>train</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0.148</td>\n",
       "      <td>2.308</td>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>0.111</td>\n",
       "      <td>2.275</td>\n",
       "      <td>0</td>\n",
       "      <td>eval</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD7CAYAAABkO19ZAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAAsTAAALEwEAmpwYAAAnCElEQVR4nO3dfXRc9X3n8fd3njR68pMsP2Bb2AEDIcTYoDoGEwLkYQ1tQrpLGlOapN2k3hByApykG5qek5bTZjdtzslu2BCot2GTbE0oC6EkxBBIGuoQ82QT2/gBgiEGGxtsC/wgW9I8ffePe0ceSTPSSBZId/R5nTNnZu793dHvZ+t+56vv/d17zd0REZHaFRvrDoiIyFtLgV5EpMYp0IuI1DgFehGRGqdALyJS4xToRURqXGKoBmaWBtYBdWH7e9z9r/u1+QvgmpLPfCfQ6u5vmNku4CiQB3Lu3j563RcRkaHYUPPozcyARnfvNLMk8Bhwvbs/UaH9h4Eb3f2y8P0uoN3dD45qz0VEpCpDZvQefBN0hm+T4WOwb4ergR+eTKemT5/u8+fPP5mPEBGZUDZu3HjQ3VvLrRsy0AOYWRzYCJwO3OruT1Zo1wCsAD5fstiBh83MgX9099VD/bz58+ezYcOGaromIiKAmb1caV1VB2PdPe/ui4G5wFIzO6dC0w8Dv3b3N0qWLXf384DLgevM7OIKnVxlZhvMbMOBAweq6ZaIiFRhWLNu3P0Q8ChB1l7OSvqVbdx9b/i8H7gPWFrhs1e7e7u7t7e2lv3rQ0RERmDIQG9mrWY2JXxdD3wAeK5Mu8nA+4D7S5Y1mllz8TXwIWDrqPRcRESqUk2Nfjbw/bBOHwPudvcHzOyzAO5+e9juD4GH3f1YybYzgfuCiTskgDvd/aFR672ICJDNZtmzZw/d3d1j3ZW3XDqdZu7cuSSTyaq3GXJ65Vhob293HYwVkWr97ne/o7m5mZaWFsLEsia5Ox0dHRw9epQFCxb0WWdmGyudp6QzY0Uk8rq7u2s+yAOYGS0tLcP+y0WBXkRqQq0H+aKRjFOBXiLJ3bl34x66Mvmx7ooIhw4d4jvf+c6wt7viiis4dOjQ6HeoHwV6iaRdHcf54v/bzMPbXxvrrohUDPT5/OCJyNq1a5kyZcpb1KsTqjozVmS8KWbyx3qU0cvYu+mmm3jxxRdZvHgxyWSSpqYmZs+ezaZNm9i+fTsf/ehH2b17N93d3Vx//fWsWrUKOHEVgM7OTi6//HIuuugi1q9fz5w5c7j//vupr68flf4po5dIyuQLAHRnFehl7H3961/ntNNOY9OmTXzjG9/gqaee4mtf+xrbt28H4I477mDjxo1s2LCBW265hY6OjgGf8cILL3Ddddexbds2pkyZwr333jtq/VNGL5GUyYWBPqdAL33d/JNtbN97ZFQ/8+xTJvHXH35X1e2XLl3aZ/rjLbfcwn333QfA7t27eeGFF2hpaemzzYIFC1i8eDEA559/Prt27Trpfhcp0EskZXsz+sIY90RkoMbGxt7Xjz76KD//+c95/PHHaWho4JJLLik7PbKurq73dTwep6ura9T6o0AvkVTM6HtUupF+hpN5j5bm5maOHj1adt3hw4eZOnUqDQ0NPPfcczzxRNlbebylFOglknpyqtHL+NHS0sLy5cs555xzqK+vZ+bMmb3rVqxYwe23386iRYs488wzWbZs2dvePwV6iSSVbmS8ufPOO8sur6ur48EHHyy7rliHnz59Olu3nrje45e+9KVR7Ztm3Ugk6WCsSPUU6CWSNL1SpHoK9BJJvRm9SjciQ1Kgl0jKKqMXqZoCvURS76ybnDJ6kaEo0EskaR69SPUU6CWSVLqRqGtqagJg7969XHXVVWXbXHLJJYzG3fYU6CWSdDBWasUpp5zCPffc85b+jCEDvZmlzewpM9tsZtvM7OYybS4xs8Nmtil8fLVk3Qoze97MdprZTaM9AJmYeqdXah69jBNf/vKX+1yT/m/+5m+4+eabef/73895553Hu9/9bu6///4B2+3atYtzzjkHgK6uLlauXMmiRYv4+Mc/PmrXu6nmzNge4DJ37zSzJPCYmT3o7v0v2PArd/+D0gVmFgduBT4I7AGeNrMfu/v20ei8TFwq3ch4s3LlSm644QY+97nPAXD33Xfz0EMPceONNzJp0iQOHjzIsmXL+MhHPlLxdoC33XYbDQ0NbNmyhS1btnDeeeeNSt+GDPTu7kBn+DYZPrzKz18K7HT3lwDM7C7gSkCBXk5KT0npxt0nzP1CpQoP3gSvPTu6nznr3XD51wdtsmTJEvbv38/evXs5cOAAU6dOZfbs2dx4442sW7eOWCzGq6++yuuvv86sWbPKfsa6dev4whe+AMCiRYtYtGjRqHS/qmvdhJn5RuB04FZ3f7JMswvMbDOwF/iSu28D5gC7S9rsAd5zcl0WOVGjhyDop5PxMeyNSOCqq67innvu4bXXXmPlypWsWbOGAwcOsHHjRpLJJPPnzy97ieJSb0XSUlWgd/c8sNjMpgD3mdk57r61pMkzwKlheecK4F+BhUC5Hpf9a8DMVgGrANra2qoegExMxdINQE9WgV5KDJF5v5VWrlzJn//5n3Pw4EH+/d//nbvvvpsZM2aQTCb55S9/ycsvvzzo9hdffDFr1qzh0ksvZevWrWzZsmVU+jWsWTfufgh4FFjRb/kRd+8MX68FkmY2nSCDn1fSdC5Bxl/us1e7e7u7t7e2tg6nWzIBlWb0OiAr48W73vUujh49ypw5c5g9ezbXXHMNGzZsoL29nTVr1nDWWWcNuv21115LZ2cnixYt4h/+4R9YunTpqPRryIzezFqBrLsfMrN64APA3/drMwt43d3dzJYSfIF0AIeAhWa2AHgVWAn88aj0XCa0TElGrwOyMp48++yJ4wPTp0/n8ccfL9uuszM49Dl//vzeSxTX19dz1113jXqfqindzAa+H9bpY8Dd7v6AmX0WwN1vB64CrjWzHNAFrAwP4ubM7PPAz4A4cEdYuxc5KdnciQqg5tKLDK6aWTdbgCVllt9e8vrbwLcrbL8WWHsSfRQZoEcZvUjVdGasRFImV6AhFRyAVaAXGZwCvURSNl9gUjoJ6AqWEgiqxbVvJONUoJdIyuQKTKoPKo/K6CWdTtPR0VHzwd7d6ejoIJ1OD2s73RxcIimTK9DaXAco0AvMnTuXPXv2cODAgbHuylsunU4zd+7cYW2jQC+RFJRugl/fHs26mfCSySQLFiwY626MWyrdSCQFpZtijV4ZvchgFOglknryBZrTqtGLVEOBXiLH3cnmCzQXZ92odCMyKAV6iZxcwXGHhmScRMyU0YsMQYFeIqd4QbNUIkY6GVdGLzIEBXqJnGKgT8bDQK+DsSKDUqCXyCleiz7I6GMq3YgMQYFeIqdnQOlGgV5kMAr0EjnFa9Gn4sWMXjV6kcEo0Evk9CndJJTRiwxFgV4ip3fWTVylG5FqKNBL5PTOukmodCNSDQV6iZzSGn2dpleKDEmBXiKnzwlTibiuXikyhCEDvZmlzewpM9tsZtvM7OYyba4xsy3hY72ZnVuybpeZPWtmm8xsw2gPQCaeYqCv0zx6kapUcz36HuAyd+80syTwmJk96O5PlLT5HfA+d3/TzC4HVgPvKVl/qbsfHL1uy0SWzQd3EUrqYKxIVYYM9B7cm6szfJsMH96vzfqSt08Aw7v9icgwZPJBYO89M1b3jBUZVFU1ejOLm9kmYD/wiLs/OUjzTwMPlrx34GEz22hmq0bcU5FQ/xp9vuC9c+tFZKCqbiXo7nlgsZlNAe4zs3PcfWv/dmZ2KUGgv6hk8XJ332tmM4BHzOw5d19XZttVwCqAtra24Y9EJoxMb+nGSCfjQHDzkWRccwtEyhnWnuHuh4BHgRX915nZIuCfgCvdvaNkm73h837gPmBphc9e7e7t7t7e2to6nG7JBNN7MDYeJ50MfoU1l16ksmpm3bSGmTxmVg98AHiuX5s24EfAJ9z9tyXLG82sufga+BAw4C8BkeEoLd3UlWT0IlJeNaWb2cD3zSxO8MVwt7s/YGafBXD324GvAi3Ad8wMIOfu7cBMglJP8Wfd6e4Pjf4wZCIp1uNLSzc9OmlKpKJqZt1sAZaUWX57yevPAJ8p0+Yl4Nz+y0VORiZXIGaQiMdIJ1S6ERmKjl5J5GTyBVJhgE+rdCMyJAV6iZxMrtA7w+ZEoFdGL1KJAr1ETiZfoK43oy+WbpTRi1SiQC+Rk8kVSPXP6HUwVqQiBXqJnEyuQLKY0SdUuhEZigK9RE42X5rRq3QjMhQFeomcTO7ErBudMCUyNAV6iZxMvnTWTfDcoytYilSkQC+RU5rRp+IxzJTRiwxGgV4ip3R6pZmRTujmIyKDUaCXyCmdXgmEtxNU6UakEgV6iZxsSY0e0O0ERYagQC+RU1qjhzDQ62CsSEUK9BI5/QN9XSKmjF5kEAr0EjmZvKt0IzIMCvQSOZlcvnfWDQQHY3t0MFakIgV6iZzS69FDsUavjF6kEgV6iZxs3knGrfe95tGLDK6am4OnzewpM9tsZtvM7OYybczMbjGznWa2xczOK1m3wsyeD9fdNNoDkIklX3DyBScVj/cuq0/FNY9eZBDVZPQ9wGXufi6wGFhhZsv6tbkcWBg+VgG3AYQ3FL81XH82cLWZnT06XZeJKBNOo0wl+p8wpYxepJIhA70HOsO3yfDh/ZpdCfwgbPsEMMXMZgNLgZ3u/pK7Z4C7wrYiI5LJB4G+tHRTp9KNyKCqqtGbWdzMNgH7gUfc/cl+TeYAu0ve7wmXVVouMiLFjL5OJ0yJVK2qQO/ueXdfDMwFlprZOf2a2MCt8EGWD2Bmq8xsg5ltOHDgQDXdkgmomNH3L91kcgXyhbK/WiIT3rBm3bj7IeBRYEW/VXuAeSXv5wJ7B1le7rNXu3u7u7e3trYOp1sygWRzxdJN34weoEdTLEXKqmbWTauZTQlf1wMfAJ7r1+zHwCfD2TfLgMPuvg94GlhoZgvMLAWsDNuKjEjZjD5RvJ2gyjci5SSqaDMb+H44gyYG3O3uD5jZZwHc/XZgLXAFsBM4DvxZuC5nZp8HfgbEgTvcfdvoD0Mmit5ZN2Uyeh2QFSlvyEDv7luAJWWW317y2oHrKmy/luCLQOSk9c66SSjQi1RLZ8ZKpPTOuul34xFQ6UakEgV6iZRyJ0zVFTN6HYwVKUuBXiKl7JmxCZVuRAajQC+Rks2Xm14ZvNalikXKU6CXSCl/wpQyepHBKNBLpPQMNr1SNXqRshToJVKyFS6BAJp1I1KJAr1EStkTpnQwVmRQCvQSKeWvR18M9MroRcpRoJdIKTfrpq73WjfK6EXKUaCXSMnkBt54JBYzUomYDsaKVKBAL5HSky+QSsQw63urg3Qipnn0IhUo0EukZHPe50BsUTqp2wmKVKJAL5GSyef7HIgtUqAXqUyBXiIlkytUyOhjmnUjUoECvURKNu8kEwNvRRzcIFwZvUg5CvQSKRUz+oRKNyKVKNBLpPTkCqTCM2FL1al0I1KRAr1ESjZfIBWvULpRRi9S1pD3jDWzecAPgFlAAVjt7t/q1+YvgGtKPvOdQKu7v2Fmu4CjQB7IuXv76HVfJppMrlBx1k3xypYi0teQgR7IAV9092fMrBnYaGaPuPv2YgN3/wbwDQAz+zBwo7u/UfIZl7r7wdHsuExMmXyh92qVpdKJmDJ6kQqGLN24+z53fyZ8fRTYAcwZZJOrgR+OTvdE+gpKN5pHLzIcw6rRm9l8YAnwZIX1DcAK4N6SxQ48bGYbzWzVCPspAgSlm6Tm0YsMSzWlGwDMrIkggN/g7kcqNPsw8Ot+ZZvl7r7XzGYAj5jZc+6+rsznrwJWAbS1tVU9AJlYBqvRd+fyuPuA6+CITHRVZfRmliQI8mvc/UeDNF1Jv7KNu+8Nn/cD9wFLy23o7qvdvd3d21tbW6vplkxAPYMEevcT95QVkROGDPQWpEffBXa4+zcHaTcZeB9wf8myxvAALmbWCHwI2HqynZaJq1KN/sQ16RXoRfqrpnSzHPgE8KyZbQqXfQVoA3D328Nlfwg87O7HSradCdwX/imdAO5094dGod8yQWXylTN6gJ5sHuqTb3e3RMa1IQO9uz8GDFn0dPfvAd/rt+wl4NwR9k1kgMoXNdPtBEUq0ZmxEinZfIFkmYy+vhjodWEzkQEU6CUyCgUnm6904xHdN1akEgV6iYzijJrBavQq3YgMpEAvkZEtBnpl9CLDokAvkZHJVc7o6xLFjF6BXqQ/BXqJjGpKN10K9CIDKNBLZGRzDlDxWjcAParRiwygQC+RkckH2fqgB2M1vVJkAAV6iYzijUUGP2FKgV6kPwV6iYxsPijdpBJlbiWoa92IVKRAL5HRO+smPvDm4Il4jETMlNGLlKFAL5Ex2PRKKN5lShm9SH8K9BIZ2UGmV0J4lykdjBUZQIFeIqN4MDYZL38x1bqE7hsrUo4CvURG8YSpukEyes2jFxlIgV4iIzvIwVgo1uiV0Yv0p0AvkVHM6JNlplfCiRuEi0hfCvQSGZlBTpiC8GCsSjciAyjQS2QMOetGB2NFyhoy0JvZPDP7pZntMLNtZnZ9mTaXmNlhM9sUPr5asm6FmT1vZjvN7KbRHoBMHCdm3Qw2j16BXqS/IW8ODuSAL7r7M2bWDGw0s0fcfXu/dr9y9z8oXWBmceBW4IPAHuBpM/txmW1FhjRU6aZOpRuRsobM6N19n7s/E74+CuwA5lT5+UuBne7+krtngLuAK0faWZnYMvkCybgRi1U+GNujg7EiAwyrRm9m84ElwJNlVl9gZpvN7EEze1e4bA6wu6TNHqr/khDpI5srVCzbQLFGr4xepL9qSjcAmFkTcC9wg7sf6bf6GeBUd+80syuAfwUWAuVSL6/w+auAVQBtbW3VdksmkEy+UPFALBRn3SijF+mvqozezJIEQX6Nu/+o/3p3P+LuneHrtUDSzKYTZPDzSprOBfaW+xnuvtrd2929vbW1dZjDkIkgkytUrM9DULrJFZxcXlm9SKlqZt0Y8F1gh7t/s0KbWWE7zGxp+LkdwNPAQjNbYGYpYCXw49HqvEwsQY1+8IweoDunQC9SqprSzXLgE8CzZrYpXPYVoA3A3W8HrgKuNbMc0AWsdHcHcmb2eeBnQBy4w923je4QZKLI5AoVr3MDfe8y1VRXdVVSpOYNuTe4+2OUr7WXtvk28O0K69YCa0fUO5ESmdwQNfqEbicoUo7OjJXIyA5Ruqkrlm4080akDwV6iYyhZ90ooxcpR4FeIqOaWTeATpoS6UeBXiIjk/chavQq3YiUo0AvkZEZ6sxYlW5EylKgl8jI5PJVTq9URi9SSoFeIiM7VOmmd9aNMnqRUgr0EhlB6abyKR29Gb0Oxor0oUAvkVH99EqVbkRKKdBLZGRzBVLxeMX1Kt2IlKdAL5HRky+QTFQu3aTiMeqTcX6986CuYClSQoFeIsHdg4uaDTK90sz46ofPZv2LHdz8k+0E19UTEV3iTyIhVwiC9mA1eoCrl7ax6+Ax/nHdS8yf3sinL1rwdnRPZFxToJdIKN4YfLATpoq+vOIsXu44zt/9dDtt0xr44Nkz3+ruiYxrKt1IJBQD/VAZPUAsZvyPjy9m0ZzJfOGHv2Hrq4d71x3uynL/ple57s5n+NQdT3Hboy+yZc8h8gWVeaR2KaOXSMjmqw/0APWpOP/7U+384a3r+c/fe5r/8r7TePT5/Tz+Yge5gtPaXMeU+iR//9BzAExKJ1j2jhbOmNlMx7EeXjvczetHeth/tJvDXVkm1yeZ2pBiWmPwmNKQJJNzurI5jvXk6crk6c7laUwlaGlKMb2pjmmNKVqaUkxKJ2mqS9CQitNYl6AxvClKZ3eOzp7iI0su7zSnEzTVJWlKJ8LXwXYNqQTxWOUD0cUvqpgFxyqKenJ5Dh3P8ubxDG8eC57rEjHapjUwb1pD75RUqW01FejP+9tHONaTwwkO3rkHdyKPm5GIG8l4jGQ8RipuxGKG9dtvYta3TTIeIx6zPp9XcKfgEI8ZiVjQJhE3ErEYqUTJ9olY75UWs/kC+YKTzTv5QoGYGXXJOPXJOPWpGOlEnFQi1tsfC+/zYgaJmJGIx0jGiz8rRiJmxMx6+xCPGU3pBJPrk0ypTzK5PkmiihJHlPQMo3RTNKM5zR1/+nv8p9vW87cPbOcd0xv5zHvfwYfeNZPFc6cQixn7j3bz+IsdrN/Zwa9fPMgjO16npbGOmZPqmDU5zbnzpjC5PsmR7ixvdGZ443iGnfs7OdSVDWb5pOI0puLUp+JMa0xxrCfH9r1HONjZw5Hu3Kj+G6STMRpTCdLJONl8gUy+QE82eC79i8Qs+F2OWXA28WBmTUrT1tJAa1MdTrjPhL/nMTMa6uI0phI01MVpKv7swomf25Mt0JPLc7gry6HjWQ51BV8oh7uyGJBOhb/nyTjpVJy4Qd6hUHAK7uQLTjxmNKYSNNbFaahL0Bh+sdUlg30jnYxTn4yRTsZ7lxWfU4lYn350h8/uwb+DmfX+W6STcSalkzSHX6KT6pMk4zGy+ULwyDnZQvB7Nikd7EfVJhbjXU0F+quXziOXd8yCIG4E/9kFD+Zg5wpOJl8gmyuU/VM9704uH7bpfXiYJcWIxYIdCIIdIZt3jmdy5MIgngt3vmyuQCbvZHJ5zCwM1sGXQSJu5Ase/EJm8xzP5t+SskFTXfDL3BjuOMVMsrkuQVOYKRafm9MJptSnmBR+SUxpCJ6HE1Tfapkwox/sWjflnDmrmZ9+4SKyeef0GU0D1s9oTnPl4jlcuXgOQG/gGQ2ZXIE3jmU42p3lWCbPsZ5c8MgEXwBNdUGmX/y/SMSMzp4cR7uDDP9omPEf78lzPJPneCbYtitTIJUwUmFCUZeI9/5fBYlIMYhCYyrO1MYUUxtSTG0M/irpyuZ5peM4r7xxnJc7jvPKG8d47rUjwX5D8Dse7DfO8WK/M/ne8llRMm7UhcG2+HszoznNGTObmVKfwnG6s0Hw7coEv+vuwf4ZtyBZipmRKwT7UcexDK+8cbz3Z3bnCgN+5tstnQzG1pxOhn9ZBV9C9angi2/W5DRzp9Yzd2oDc6fWM2tyelztN0U1Fej/4j+cNdZdGJFs/sQvdGnIL4RfPLl8gWzBwy+rAvkC5AoFCuFzvuAc7clx+Hi2T2bV2R0Ehs6eYMd549jxE6WC7lzvTJZKmtOJ3lLFtIagHHH+qVO54LQW5k1reAv/RQbqrdGPYCc6taWx6rajFeQhKDPNmpxm1uT0qH3maDmvbeqwt8nmC3Rn8+FfvDFio/hvVUmh4PTkCnRl83Rn8/TkCn2ei1c0TRcz/kSMumScWJjgFf8Szxecrmyeo905jnRlOdqT5UhXjmx4tnXxL/Fk3HCHo91ZjnTnONyV5UhXliPd2fDLNs/+o90czwSfdbCzh9JZvGbBLS2TcevzuTMn1XH6jCZOa23i9BnB45TJ9W/LvyFUEejNbB7wA2AWUABWu/u3+rW5Bvhy+LYTuNbdN4frdgFHgTyQc/f2Uet9jSj+Mryd3IMd6Gh3jiPdwRdE8Zf6cFe2t577xrEMbx7P8NqRbp555U3+ZcNuANqmNXDhaS1ccFoLl5w5g8n1ybe0v8Ua/XjMliaKsfg9jcWM+rA0Nh5lcgX2He5iz5td7HnzOK8e6u79AipWBXpyBfYe6uKhra/x5vFs77btp07lnmsvfFv6WU1GnwO+6O7PmFkzsNHMHnH37SVtfge8z93fNLPLgdXAe0rWX+ruB0ev23KyzIx0Mqh/tjbXVbWNu/PC/k7W7zzI+hc7WPvsPu56ejepeIxLz2rlo4vncOlZM96SA3zDmXUj8nZJJWKc2tJY9V+NHZ09vLC/k537O2l4G7+8hgz07r4P2Be+PmpmO4A5wPaSNutLNnkCmDvK/ZRxwMw4Y2YzZ8xs5k+XLyBfcDbvOcQDm/fxky17+dm212muS/Chd83ij9rnsnTBtD4zQE6GAr3UgpamOlqa6lj2jpa39ecOq0ZvZvOBJcCTgzT7NPBgyXsHHjYzB/7R3VcPt5MyPsVjxnltUzmvbSp/9fvv5ImXOrh/06s8uPU17n1mD2fNauaTF8zno0tOoSF1coeDMsOcXikiJ1S995lZE3AvcIO7H6nQ5lKCQH9RyeLl7r7XzGYAj5jZc+6+rsy2q4BVAG1tbcMYgowH8Zix/PTpLD99Ojd/5Bx+snkv31u/i6/c9yz//cEdfOz8eXzmvQs4ZUr9iD7/ZA7Gikx0Ve01ZpYkCPJr3P1HFdosAv4JuNLdO4rL3X1v+LwfuA9YWm57d1/t7u3u3t7a2jq8Uci4Up+K80e/N4+ffuEi7r32Ai49cwb/94ldfOz2x3nzWGZEn6mMXmTkhtxrLCiyfhfY4e7frNCmDfgR8Al3/23J8sbwAC5m1gh8CNg6Gh2X8c/MOP/Uadxy9RLu+eyFHDjaw/X/smlE5w30nhmrjF5k2KrZa5YDnwAuM7NN4eMKM/usmX02bPNVoAX4Trh+Q7h8JvCYmW0GngJ+6u4PjfYgZPw7d94U/vojZ7Putwf4X//2wrC3772omTJ6kWGrZtbNY8CgUyfc/TPAZ8osfwk4d8S9k5ryx0vb2Pjym3zrFy+wpG0q7zuj+hKdavQiI6e9Rt42ZsbXPvpuzpzZzPV3/YZXD3VVvW0mX9316EVkIO018raqT8W57U/OJ593PrfmGXpy1d3fVRm9yMhpr5G33YLpjXzjY+eyefch/ttPd1S1jU6YEhk57TUyJlacM4tPLDuVf37yFfZWUcLJ5gvEw0syi8jwKNDLmFl18Ttwd/75iZeHbJvJF0jGFeRFRkKBXsbMvGkNvP+dM7nr6d10Zwev1WdyBdXnRUZIe46MqT+9cD5vHMvwk817B22XyRdIJcbnpWpFxjsFehlTF57WwsIZTXz/8V24Vz5jNsjoVboRGQkFehlTZsYnL5zP1leP8Mwrb1Zsl8kVNONGZIS058iY+49L5tCcTvC99ZUPyhZv+SYiw6c9R8ZcY12Cj50/jwef3cfrR7rLtineG1REhk97jowLn7zgVPLurHnylbLrM8roRUZMe46MC/OnN3LJGa3c+eQrvWfBlurR9EqREdOeI+PGpy6cz8HOHtY+u2/AOtXoRUZOe46MGxcvbGXB9Ea+t37XgHU6YUpk5LTnyLgRixnXvKeNTbsP8UrH8T7rNL1SZOS058i4culZMwBY98KBPstVuhEZOe05Mq68Y3ojc6bUs+63fQO9pleKjJz2HBlXzIyLz5jO+hc7em8IDppeKXIyhtxzzGyemf3SzHaY2TYzu75MGzOzW8xsp5ltMbPzStatMLPnw3U3jfYApPZcvLCVzp4cv3nlUO8yHYwVGblq9pwc8EV3fyewDLjOzM7u1+ZyYGH4WAXcBmBmceDWcP3ZwNVlthXp48LTpxOPGb8qqdMroxcZuSH3HHff5+7PhK+PAjuAOf2aXQn8wANPAFPMbDawFNjp7i+5ewa4K2wrUtHk+iSL503pU6dXRi8ycsPac8xsPrAEeLLfqjnA7pL3e8JllZaX++xVZrbBzDYcOHCgXBOZQN67cDpbXj3MG8cy5AtOwXW/WJGRqnrPMbMm4F7gBnc/0n91mU18kOUDF7qvdvd2d29vbW2ttltSoy4+oxV3eGznwd5LImjWjcjIVLXnmFmSIMivcfcflWmyB5hX8n4usHeQ5SKDOnfuFCbXJ/nVbw/0Bnpl9CIjU82sGwO+C+xw929WaPZj4JPh7JtlwGF33wc8DSw0swVmlgJWhm1FBhWPGRedPp11LxygJx/cT1aBXmRkElW0WQ58AnjWzDaFy74CtAG4++3AWuAKYCdwHPizcF3OzD4P/AyIA3e4+7bRHIDUrvcunM5Pn93Htr1BpVC3EhQZmSEDvbs/Rvlae2kbB66rsG4twReByLBcfEZwrObn218HlNGLjJT2HBm3TplSz+kzmvjFjv0ApOLxMe6RSDQp0Mu4dvHCVl4Lby+YVOlGZEQU6GVce+8Z03tfq3QjMjLac2RcW7agpTfAK9CLjIz2HBnX6lNxls6fBkCdAr3IiGjPkXHvvQuD8o3OjBUZmWrm0YuMqY+1z+NgZw9nzZo01l0RiSQFehn3pjWm+Kvf19WtRUZKfwuLiNQ4BXoRkRqnQC8iUuMU6EVEapwCvYhIjVOgFxGpcQr0IiI1ToFeRKTGWXDPkPHFzA4AL49w8+nAwVHszliqpbGAxjOe1dJYoLbGU+1YTnX31nIrxmWgPxlmtsHd28e6H6OhlsYCGs94Vktjgdoaz2iMRaUbEZEap0AvIlLjajHQrx7rDoyiWhoLaDzjWS2NBWprPCc9lpqr0YuISF+1mNGLiEiJmgn0ZrbCzJ43s51mdtNY92e4zOwOM9tvZltLlk0zs0fM7IXweepY9rFaZjbPzH5pZjvMbJuZXR8uj+p40mb2lJltDsdzc7g8kuMBMLO4mf3GzB4I30d5LLvM7Fkz22RmG8JlUR7PFDO7x8yeC/ehC052PDUR6M0sDtwKXA6cDVxtZlG7U8X3gBX9lt0E/MLdFwK/CN9HQQ74oru/E1gGXBf+f0R1PD3AZe5+LrAYWGFmy4jueACuB3aUvI/yWAAudffFJdMQozyebwEPuftZwLkE/08nNx53j/wDuAD4Wcn7vwT+cqz7NYJxzAe2lrx/Hpgdvp4NPD/WfRzhuO4HPlgL4wEagGeA90R1PMDcMFhcBjwQLovkWML+7gKm91sWyfEAk4DfER4/Ha3x1ERGD8wBdpe83xMui7qZ7r4PIHyeMcb9GTYzmw8sAZ4kwuMJSx2bgP3AI+4e5fH8T+C/AoWSZVEdC4ADD5vZRjNbFS6L6njeARwA/k9YWvsnM2vkJMdTK4HeyizTdKIxZmZNwL3ADe5+ZKz7czLcPe/uiwmy4aVmds4Yd2lEzOwPgP3uvnGs+zKKlrv7eQSl2+vM7OKx7tBJSADnAbe5+xLgGKNQdqqVQL8HmFfyfi6wd4z6MppeN7PZAOHz/jHuT9XMLEkQ5Ne4+4/CxZEdT5G7HwIeJTieEsXxLAc+Yma7gLuAy8zsn4nmWABw973h837gPmAp0R3PHmBP+BcjwD0Egf+kxlMrgf5pYKGZLTCzFLAS+PEY92k0/Bj4VPj6UwS17nHPzAz4LrDD3b9Zsiqq42k1synh63rgA8BzRHA87v6X7j7X3ecT7Cf/5u5/QgTHAmBmjWbWXHwNfAjYSkTH4+6vAbvN7Mxw0fuB7ZzseMb64MMoHsS4Avgt8CLwV2PdnxH0/4fAPiBL8K3+aaCF4KDZC+HztLHuZ5VjuYigdLYF2BQ+rojweBYBvwnHsxX4arg8kuMpGdclnDgYG8mxENS0N4ePbcV9P6rjCfu+GNgQ/r79KzD1ZMejM2NFRGpcrZRuRESkAgV6EZEap0AvIlLjFOhFRGqcAr2ISI1ToBcRqXEK9CIiNU6BXkSkxv1/uwEtyalaafgAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "class SequentialModel(nn.Module):\n",
    "    def __init__(self, *layers):\n",
    "        super().__init__()\n",
    "        self.layers = nn.ModuleList(layers)\n",
    "        self.act_means = [[] for _ in layers]\n",
    "        self.act_stds = [[] for _ in layers]\n",
    "    \n",
    "    def __call__(self, x):\n",
    "        for i,l in enumerate(self.layers):\n",
    "            x = l(x)\n",
    "            self.act_means[i].append(x.mean().detach())\n",
    "            self.act_stds[i].append(x.std().detach())\n",
    "        return x\n",
    "    \n",
    "    def __iter__(self):\n",
    "        return iter(self.layers)\n",
    "    \n",
    "model = SequentialModel(*cnn_layers())\n",
    "learn = fit(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "5853b308",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x180ad8450>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for l in model.act_means:\n",
    "    plt.plot(l)\n",
    "plt.legend(range(5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "1feee31d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1808ba250>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for l in model.act_stds:\n",
    "    plt.plot(l)\n",
    "plt.legend(range(5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "a76256ce",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "<style>\n",
       "    /* Turns off some styling */\n",
       "    progress {\n",
       "        /* gets rid of default border in Firefox and Opera. */\n",
       "        border: none;\n",
       "        /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "        background-size: auto;\n",
       "    }\n",
       "    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>accuracy</th>\n",
       "      <th>loss</th>\n",
       "      <th>epoch</th>\n",
       "      <th>train</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0.146</td>\n",
       "      <td>2.316</td>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>0.100</td>\n",
       "      <td>2.304</td>\n",
       "      <td>0</td>\n",
       "      <td>eval</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n",
      "[W ParallelNative.cpp:230] Warning: Cannot set number of intraop threads after parallel work has started or after set_num_threads call when using native parallel backend (function set_num_threads)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<miniai.lec_9_v4.Learner at 0x18082fd50>"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = nn.Sequential(*cnn_layers())\n",
    "act_means = [[] for _ in model]\n",
    "act_stds = [[] for _ in model]\n",
    "\n",
    "def append_stats(i, mod, inp, outp):\n",
    "    act_means[i].append(outp.mean().detach())\n",
    "    act_stds[i].append(outp.std().detach())\n",
    "\n",
    "for i,m in enumerate(model):\n",
    "    m.register_forward_hook(partial(append_stats, i))\n",
    "\n",
    "fit(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "9c371bbe",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x186d0ca10>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for o in act_means:\n",
    "    plt.plot(o)\n",
    "plt.legend(range(5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "77feca89",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "79416089",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e8e81b2a",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "python_main",
   "language": "python",
   "name": "python_main"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}