{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Python Machine Learning 3rd Edition* by [Sebastian Raschka](https://sebastianraschka.com) & [Vahid Mirjalili](http://vahidmirjalili.com), Packt Publishing Ltd. 2019\n",
    "\n",
    "Code Repository: https://github.com/rasbt/python-machine-learning-book-3rd-edition\n",
    "\n",
    "Code License: [MIT License](https://github.com/rasbt/python-machine-learning-book-3rd-edition/blob/master/LICENSE.txt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Chapter 14: Going Deeper -- the Mechanics of TensorFlow (Part 2/3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that the optional watermark extension is a small IPython notebook plugin that I developed to make the code reproducible. You can just skip the following line(s)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sebastian Raschka & Vahid Mirjalili \n",
      "last updated: 2019-11-03 \n",
      "\n",
      "numpy 1.17.3\n",
      "scipy 1.3.1\n",
      "matplotlib 3.1.1\n",
      "tensorflow 2.0.0\n"
     ]
    }
   ],
   "source": [
    "%load_ext watermark\n",
    "%watermark -a \"Sebastian Raschka & Vahid Mirjalili\" -u -d -p numpy,scipy,matplotlib,tensorflow"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import pandas as pd\n",
    "\n",
    "from IPython.display import Image\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## TensorFlow Estimators\n",
    "\n",
    "##### Steps for using pre-made estimators\n",
    "\n",
    " * **Step 1:** Define the input function for importing the data   \n",
    " * **Step 2:**  Define the feature columns to bridge between the estimator and the data   \n",
    " * **Step 3:** Instantiate an estimator or convert a Keras model to an estimator   \n",
    " * **Step 4:** Use the estimator: train() evaluate() predict()   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "tf.random.set_seed(1)\n",
    "np.random.seed(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Working with feature columns\n",
    "\n",
    "\n",
    " * See definition: https://developers.google.com/machine-learning/glossary/#feature_columns\n",
    " * Documentation: https://www.tensorflow.org/api_docs/python/tf/feature_column"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 4,
     "metadata": {
      "image/png": {
       "width": 700
      }
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(filename='images/02.png', width=700)"
   ]
  },
  {
   "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>MPG</th>\n",
       "      <th>Cylinders</th>\n",
       "      <th>Displacement</th>\n",
       "      <th>Horsepower</th>\n",
       "      <th>Weight</th>\n",
       "      <th>Acceleration</th>\n",
       "      <th>ModelYear</th>\n",
       "      <th>Origin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>393</td>\n",
       "      <td>27.0</td>\n",
       "      <td>4</td>\n",
       "      <td>140.0</td>\n",
       "      <td>86.0</td>\n",
       "      <td>2790.0</td>\n",
       "      <td>15.6</td>\n",
       "      <td>82</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>394</td>\n",
       "      <td>44.0</td>\n",
       "      <td>4</td>\n",
       "      <td>97.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>2130.0</td>\n",
       "      <td>24.6</td>\n",
       "      <td>82</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>395</td>\n",
       "      <td>32.0</td>\n",
       "      <td>4</td>\n",
       "      <td>135.0</td>\n",
       "      <td>84.0</td>\n",
       "      <td>2295.0</td>\n",
       "      <td>11.6</td>\n",
       "      <td>82</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>396</td>\n",
       "      <td>28.0</td>\n",
       "      <td>4</td>\n",
       "      <td>120.0</td>\n",
       "      <td>79.0</td>\n",
       "      <td>2625.0</td>\n",
       "      <td>18.6</td>\n",
       "      <td>82</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>397</td>\n",
       "      <td>31.0</td>\n",
       "      <td>4</td>\n",
       "      <td>119.0</td>\n",
       "      <td>82.0</td>\n",
       "      <td>2720.0</td>\n",
       "      <td>19.4</td>\n",
       "      <td>82</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      MPG  Cylinders  Displacement  Horsepower  Weight  Acceleration  \\\n",
       "393  27.0          4         140.0        86.0  2790.0          15.6   \n",
       "394  44.0          4          97.0        52.0  2130.0          24.6   \n",
       "395  32.0          4         135.0        84.0  2295.0          11.6   \n",
       "396  28.0          4         120.0        79.0  2625.0          18.6   \n",
       "397  31.0          4         119.0        82.0  2720.0          19.4   \n",
       "\n",
       "     ModelYear  Origin  \n",
       "393         82       1  \n",
       "394         82       2  \n",
       "395         82       1  \n",
       "396         82       1  \n",
       "397         82       1  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset_path = tf.keras.utils.get_file(\"auto-mpg.data\", \n",
    "                                       (\"http://archive.ics.uci.edu/ml/machine-learning-databases\"\n",
    "                                        \"/auto-mpg/auto-mpg.data\"))\n",
    "\n",
    "column_names = ['MPG', 'Cylinders', 'Displacement', 'Horsepower',\n",
    "                'Weight', 'Acceleration', 'ModelYear', 'Origin']\n",
    "\n",
    "df = pd.read_csv(dataset_path, names=column_names,\n",
    "                 na_values = \"?\", comment='\\t',\n",
    "                 sep=\" \", skipinitialspace=True)\n",
    "\n",
    "df.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MPG             0\n",
      "Cylinders       0\n",
      "Displacement    0\n",
      "Horsepower      6\n",
      "Weight          0\n",
      "Acceleration    0\n",
      "ModelYear       0\n",
      "Origin          0\n",
      "dtype: int64\n"
     ]
    },
    {
     "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>MPG</th>\n",
       "      <th>Cylinders</th>\n",
       "      <th>Displacement</th>\n",
       "      <th>Horsepower</th>\n",
       "      <th>Weight</th>\n",
       "      <th>Acceleration</th>\n",
       "      <th>ModelYear</th>\n",
       "      <th>Origin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>387</td>\n",
       "      <td>27.0</td>\n",
       "      <td>4</td>\n",
       "      <td>140.0</td>\n",
       "      <td>86.0</td>\n",
       "      <td>2790.0</td>\n",
       "      <td>15.6</td>\n",
       "      <td>82</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>388</td>\n",
       "      <td>44.0</td>\n",
       "      <td>4</td>\n",
       "      <td>97.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>2130.0</td>\n",
       "      <td>24.6</td>\n",
       "      <td>82</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>389</td>\n",
       "      <td>32.0</td>\n",
       "      <td>4</td>\n",
       "      <td>135.0</td>\n",
       "      <td>84.0</td>\n",
       "      <td>2295.0</td>\n",
       "      <td>11.6</td>\n",
       "      <td>82</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>390</td>\n",
       "      <td>28.0</td>\n",
       "      <td>4</td>\n",
       "      <td>120.0</td>\n",
       "      <td>79.0</td>\n",
       "      <td>2625.0</td>\n",
       "      <td>18.6</td>\n",
       "      <td>82</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>391</td>\n",
       "      <td>31.0</td>\n",
       "      <td>4</td>\n",
       "      <td>119.0</td>\n",
       "      <td>82.0</td>\n",
       "      <td>2720.0</td>\n",
       "      <td>19.4</td>\n",
       "      <td>82</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      MPG  Cylinders  Displacement  Horsepower  Weight  Acceleration  \\\n",
       "387  27.0          4         140.0        86.0  2790.0          15.6   \n",
       "388  44.0          4          97.0        52.0  2130.0          24.6   \n",
       "389  32.0          4         135.0        84.0  2295.0          11.6   \n",
       "390  28.0          4         120.0        79.0  2625.0          18.6   \n",
       "391  31.0          4         119.0        82.0  2720.0          19.4   \n",
       "\n",
       "     ModelYear  Origin  \n",
       "387         82       1  \n",
       "388         82       2  \n",
       "389         82       1  \n",
       "390         82       1  \n",
       "391         82       1  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(df.isna().sum())\n",
    "\n",
    "df = df.dropna()\n",
    "df = df.reset_index(drop=True)\n",
    "df.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>MPG</td>\n",
       "      <td>313.0</td>\n",
       "      <td>23.404153</td>\n",
       "      <td>7.666909</td>\n",
       "      <td>9.0</td>\n",
       "      <td>17.5</td>\n",
       "      <td>23.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>46.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Cylinders</td>\n",
       "      <td>313.0</td>\n",
       "      <td>5.402556</td>\n",
       "      <td>1.701506</td>\n",
       "      <td>3.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>8.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Displacement</td>\n",
       "      <td>313.0</td>\n",
       "      <td>189.512780</td>\n",
       "      <td>102.675646</td>\n",
       "      <td>68.0</td>\n",
       "      <td>104.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>260.0</td>\n",
       "      <td>455.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Horsepower</td>\n",
       "      <td>313.0</td>\n",
       "      <td>102.929712</td>\n",
       "      <td>37.919046</td>\n",
       "      <td>46.0</td>\n",
       "      <td>75.0</td>\n",
       "      <td>92.0</td>\n",
       "      <td>120.0</td>\n",
       "      <td>230.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Weight</td>\n",
       "      <td>313.0</td>\n",
       "      <td>2961.198083</td>\n",
       "      <td>848.602146</td>\n",
       "      <td>1613.0</td>\n",
       "      <td>2219.0</td>\n",
       "      <td>2755.0</td>\n",
       "      <td>3574.0</td>\n",
       "      <td>5140.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Acceleration</td>\n",
       "      <td>313.0</td>\n",
       "      <td>15.704473</td>\n",
       "      <td>2.725399</td>\n",
       "      <td>8.5</td>\n",
       "      <td>14.0</td>\n",
       "      <td>15.5</td>\n",
       "      <td>17.3</td>\n",
       "      <td>24.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ModelYear</td>\n",
       "      <td>313.0</td>\n",
       "      <td>75.929712</td>\n",
       "      <td>3.675305</td>\n",
       "      <td>70.0</td>\n",
       "      <td>73.0</td>\n",
       "      <td>76.0</td>\n",
       "      <td>79.0</td>\n",
       "      <td>82.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Origin</td>\n",
       "      <td>313.0</td>\n",
       "      <td>1.591054</td>\n",
       "      <td>0.807923</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              count         mean         std     min     25%     50%     75%  \\\n",
       "MPG           313.0    23.404153    7.666909     9.0    17.5    23.0    29.0   \n",
       "Cylinders     313.0     5.402556    1.701506     3.0     4.0     4.0     8.0   \n",
       "Displacement  313.0   189.512780  102.675646    68.0   104.0   140.0   260.0   \n",
       "Horsepower    313.0   102.929712   37.919046    46.0    75.0    92.0   120.0   \n",
       "Weight        313.0  2961.198083  848.602146  1613.0  2219.0  2755.0  3574.0   \n",
       "Acceleration  313.0    15.704473    2.725399     8.5    14.0    15.5    17.3   \n",
       "ModelYear     313.0    75.929712    3.675305    70.0    73.0    76.0    79.0   \n",
       "Origin        313.0     1.591054    0.807923     1.0     1.0     1.0     2.0   \n",
       "\n",
       "                 max  \n",
       "MPG             46.6  \n",
       "Cylinders        8.0  \n",
       "Displacement   455.0  \n",
       "Horsepower     230.0  \n",
       "Weight        5140.0  \n",
       "Acceleration    24.8  \n",
       "ModelYear       82.0  \n",
       "Origin           3.0  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import sklearn\n",
    "import sklearn.model_selection\n",
    "\n",
    "\n",
    "df_train, df_test = sklearn.model_selection.train_test_split(df, train_size=0.8)\n",
    "train_stats = df_train.describe().transpose()\n",
    "train_stats"
   ]
  },
  {
   "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>MPG</th>\n",
       "      <th>Cylinders</th>\n",
       "      <th>Displacement</th>\n",
       "      <th>Horsepower</th>\n",
       "      <th>Weight</th>\n",
       "      <th>Acceleration</th>\n",
       "      <th>ModelYear</th>\n",
       "      <th>Origin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>203</td>\n",
       "      <td>28.0</td>\n",
       "      <td>-0.824303</td>\n",
       "      <td>-0.901020</td>\n",
       "      <td>-0.736562</td>\n",
       "      <td>-0.950031</td>\n",
       "      <td>0.255202</td>\n",
       "      <td>76</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>255</td>\n",
       "      <td>19.4</td>\n",
       "      <td>0.351127</td>\n",
       "      <td>0.413800</td>\n",
       "      <td>-0.340982</td>\n",
       "      <td>0.293190</td>\n",
       "      <td>0.548737</td>\n",
       "      <td>78</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>72</td>\n",
       "      <td>13.0</td>\n",
       "      <td>1.526556</td>\n",
       "      <td>1.144256</td>\n",
       "      <td>0.713897</td>\n",
       "      <td>1.339617</td>\n",
       "      <td>-0.625403</td>\n",
       "      <td>72</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>235</td>\n",
       "      <td>30.5</td>\n",
       "      <td>-0.824303</td>\n",
       "      <td>-0.891280</td>\n",
       "      <td>-1.053025</td>\n",
       "      <td>-1.072585</td>\n",
       "      <td>0.475353</td>\n",
       "      <td>77</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>14.0</td>\n",
       "      <td>1.526556</td>\n",
       "      <td>1.563051</td>\n",
       "      <td>1.636916</td>\n",
       "      <td>1.470420</td>\n",
       "      <td>-1.359240</td>\n",
       "      <td>71</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      MPG  Cylinders  Displacement  Horsepower    Weight  Acceleration  \\\n",
       "203  28.0  -0.824303     -0.901020   -0.736562 -0.950031      0.255202   \n",
       "255  19.4   0.351127      0.413800   -0.340982  0.293190      0.548737   \n",
       "72   13.0   1.526556      1.144256    0.713897  1.339617     -0.625403   \n",
       "235  30.5  -0.824303     -0.891280   -1.053025 -1.072585      0.475353   \n",
       "37   14.0   1.526556      1.563051    1.636916  1.470420     -1.359240   \n",
       "\n",
       "     ModelYear  Origin  \n",
       "203         76       3  \n",
       "255         78       1  \n",
       "72          72       1  \n",
       "235         77       1  \n",
       "37          71       1  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "numeric_column_names = ['Cylinders', 'Displacement', 'Horsepower', 'Weight', 'Acceleration']\n",
    "\n",
    "df_train_norm, df_test_norm = df_train.copy(), df_test.copy()\n",
    "\n",
    "for col_name in numeric_column_names:\n",
    "    mean = train_stats.loc[col_name, 'mean']\n",
    "    std  = train_stats.loc[col_name, 'std']\n",
    "    df_train_norm.loc[:, col_name] = (df_train_norm.loc[:, col_name] - mean)/std\n",
    "    df_test_norm.loc[:, col_name] = (df_test_norm.loc[:, col_name] - mean)/std\n",
    "    \n",
    "df_train_norm.tail()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Numeric Columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[NumericColumn(key='Cylinders', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None),\n",
       " NumericColumn(key='Displacement', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None),\n",
       " NumericColumn(key='Horsepower', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None),\n",
       " NumericColumn(key='Weight', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None),\n",
       " NumericColumn(key='Acceleration', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None)]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "numeric_features = []\n",
    "\n",
    "for col_name in numeric_column_names:\n",
    "    numeric_features.append(tf.feature_column.numeric_column(key=col_name))\n",
    "    \n",
    "numeric_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[BucketizedColumn(source_column=NumericColumn(key='ModelYear', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), boundaries=(73, 76, 79))]\n"
     ]
    }
   ],
   "source": [
    "feature_year = tf.feature_column.numeric_column(key=\"ModelYear\")\n",
    "\n",
    "bucketized_features = []\n",
    "\n",
    "bucketized_features.append(tf.feature_column.bucketized_column(\n",
    "    source_column=feature_year,\n",
    "    boundaries=[73, 76, 79]))\n",
    "\n",
    "print(bucketized_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[IndicatorColumn(categorical_column=VocabularyListCategoricalColumn(key='Origin', vocabulary_list=(1, 2, 3), dtype=tf.int64, default_value=-1, num_oov_buckets=0))]\n"
     ]
    }
   ],
   "source": [
    "feature_origin = tf.feature_column.categorical_column_with_vocabulary_list(\n",
    "    key='Origin',\n",
    "    vocabulary_list=[1, 2, 3])\n",
    "\n",
    "categorical_indicator_features = []\n",
    "categorical_indicator_features.append(tf.feature_column.indicator_column(feature_origin))\n",
    "\n",
    "print(categorical_indicator_features)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Machine learning with pre-made Estimators"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Keys: dict_keys(['Cylinders', 'Displacement', 'Horsepower', 'Weight', 'Acceleration', 'ModelYear', 'Origin'])\n",
      "Batch Model Years: tf.Tensor([82 78 76 72 78 73 70 78], shape=(8,), dtype=int32)\n"
     ]
    }
   ],
   "source": [
    "def train_input_fn(df_train, batch_size=8):\n",
    "    df = df_train.copy()\n",
    "    train_x, train_y = df, df.pop('MPG')\n",
    "    dataset = tf.data.Dataset.from_tensor_slices((dict(train_x), train_y))\n",
    "\n",
    "    # shuffle, repeat, and batch the examples\n",
    "    return dataset.shuffle(1000).repeat().batch(batch_size)\n",
    "\n",
    "## inspection\n",
    "ds = train_input_fn(df_train_norm)\n",
    "batch = next(iter(ds))\n",
    "print('Keys:', batch[0].keys())\n",
    "print('Batch Model Years:', batch[0]['ModelYear'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[NumericColumn(key='Cylinders', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), NumericColumn(key='Displacement', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), NumericColumn(key='Horsepower', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), NumericColumn(key='Weight', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), NumericColumn(key='Acceleration', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), BucketizedColumn(source_column=NumericColumn(key='ModelYear', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), boundaries=(73, 76, 79)), IndicatorColumn(categorical_column=VocabularyListCategoricalColumn(key='Origin', vocabulary_list=(1, 2, 3), dtype=tf.int64, default_value=-1, num_oov_buckets=0))]\n"
     ]
    }
   ],
   "source": [
    "all_feature_columns = (numeric_features + \n",
    "                       bucketized_features + \n",
    "                       categorical_indicator_features)\n",
    "\n",
    "print(all_feature_columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Using default config.\n",
      "INFO:tensorflow:Using config: {'_model_dir': 'models/autompg-dnnregressor/', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true\n",
      "graph_options {\n",
      "  rewrite_options {\n",
      "    meta_optimizer_iterations: ONE\n",
      "  }\n",
      "}\n",
      ", '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': 100, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7f47c054d650>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}\n"
     ]
    }
   ],
   "source": [
    "regressor = tf.estimator.DNNRegressor(\n",
    "    feature_columns=all_feature_columns,\n",
    "    hidden_units=[32, 10],\n",
    "    model_dir='models/autompg-dnnregressor/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Steps: 40000\n",
      "WARNING:tensorflow:From /home/vahid/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow_core/python/ops/resource_variable_ops.py:1630: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "If using Keras pass *_constraint arguments to layers.\n",
      "WARNING:tensorflow:From /home/vahid/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow_core/python/training/training_util.py:236: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.\n",
      "INFO:tensorflow:Calling model_fn.\n",
      "WARNING:tensorflow:Layer dnn is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.\n",
      "\n",
      "If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.\n",
      "\n",
      "To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.\n",
      "\n",
      "WARNING:tensorflow:From /home/vahid/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow_core/python/feature_column/feature_column_v2.py:4276: IndicatorColumn._variable_shape (from tensorflow.python.feature_column.feature_column_v2) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "The old _FeatureColumn APIs are being deprecated. Please use the new FeatureColumn APIs instead.\n",
      "WARNING:tensorflow:From /home/vahid/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow_core/python/feature_column/feature_column_v2.py:4331: VocabularyListCategoricalColumn._num_buckets (from tensorflow.python.feature_column.feature_column_v2) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "The old _FeatureColumn APIs are being deprecated. Please use the new FeatureColumn APIs instead.\n",
      "WARNING:tensorflow:From /home/vahid/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow_estimator/python/estimator/head/regression_head.py:156: to_float (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use `tf.cast` instead.\n",
      "WARNING:tensorflow:From /home/vahid/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow_core/python/keras/optimizer_v2/adagrad.py:108: calling Constant.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n",
      "INFO:tensorflow:Done calling model_fn.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n",
      "INFO:tensorflow:Saving checkpoints for 0 into models/autompg-dnnregressor/model.ckpt.\n",
      "INFO:tensorflow:loss = 610.1284, step = 0\n",
      "INFO:tensorflow:global_step/sec: 448.416\n",
      "INFO:tensorflow:loss = 712.06726, step = 100 (0.225 sec)\n",
      "INFO:tensorflow:global_step/sec: 585.712\n",
      "INFO:tensorflow:loss = 822.8959, step = 200 (0.170 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.546\n",
      "INFO:tensorflow:loss = 710.2063, step = 300 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.579\n",
      "INFO:tensorflow:loss = 583.29407, step = 400 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 643.959\n",
      "INFO:tensorflow:loss = 582.77167, step = 500 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.776\n",
      "INFO:tensorflow:loss = 649.7673, step = 600 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 572.456\n",
      "INFO:tensorflow:loss = 396.41064, step = 700 (0.175 sec)\n",
      "INFO:tensorflow:global_step/sec: 566.38\n",
      "INFO:tensorflow:loss = 669.36633, step = 800 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 566.658\n",
      "INFO:tensorflow:loss = 632.7246, step = 900 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 617.634\n",
      "INFO:tensorflow:loss = 758.34875, step = 1000 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 620.319\n",
      "INFO:tensorflow:loss = 392.08755, step = 1100 (0.161 sec)\n",
      "INFO:tensorflow:global_step/sec: 564.039\n",
      "INFO:tensorflow:loss = 476.3385, step = 1200 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 572.083\n",
      "INFO:tensorflow:loss = 506.0124, step = 1300 (0.175 sec)\n",
      "INFO:tensorflow:global_step/sec: 567.277\n",
      "INFO:tensorflow:loss = 577.36475, step = 1400 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 565.847\n",
      "INFO:tensorflow:loss = 393.88156, step = 1500 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 566.604\n",
      "INFO:tensorflow:loss = 635.9873, step = 1600 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 586.154\n",
      "INFO:tensorflow:loss = 423.8427, step = 1700 (0.171 sec)\n",
      "INFO:tensorflow:global_step/sec: 647.064\n",
      "INFO:tensorflow:loss = 498.0647, step = 1800 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.755\n",
      "INFO:tensorflow:loss = 283.3813, step = 1900 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 666.887\n",
      "INFO:tensorflow:loss = 405.86353, step = 2000 (0.150 sec)\n",
      "INFO:tensorflow:global_step/sec: 652.271\n",
      "INFO:tensorflow:loss = 501.8705, step = 2100 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.585\n",
      "INFO:tensorflow:loss = 304.44452, step = 2200 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 649.773\n",
      "INFO:tensorflow:loss = 422.43158, step = 2300 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 653.582\n",
      "INFO:tensorflow:loss = 365.81732, step = 2400 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 655.929\n",
      "INFO:tensorflow:loss = 402.35364, step = 2500 (0.152 sec)\n",
      "INFO:tensorflow:global_step/sec: 657.293\n",
      "INFO:tensorflow:loss = 479.02838, step = 2600 (0.152 sec)\n",
      "INFO:tensorflow:global_step/sec: 612.069\n",
      "INFO:tensorflow:loss = 361.15582, step = 2700 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.744\n",
      "INFO:tensorflow:loss = 379.43427, step = 2800 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 568.614\n",
      "INFO:tensorflow:loss = 320.28763, step = 2900 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 638.986\n",
      "INFO:tensorflow:loss = 385.37427, step = 3000 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 656.466\n",
      "INFO:tensorflow:loss = 512.6961, step = 3100 (0.152 sec)\n",
      "INFO:tensorflow:global_step/sec: 568.927\n",
      "INFO:tensorflow:loss = 316.60834, step = 3200 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 564.392\n",
      "INFO:tensorflow:loss = 273.2796, step = 3300 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.287\n",
      "INFO:tensorflow:loss = 502.52917, step = 3400 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 651.162\n",
      "INFO:tensorflow:loss = 349.2741, step = 3500 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 649.318\n",
      "INFO:tensorflow:loss = 139.09305, step = 3600 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 653.045\n",
      "INFO:tensorflow:loss = 198.39453, step = 3700 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.472\n",
      "INFO:tensorflow:loss = 331.9128, step = 3800 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 643.694\n",
      "INFO:tensorflow:loss = 345.54785, step = 3900 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 639.938\n",
      "INFO:tensorflow:loss = 434.36136, step = 4000 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 654.364\n",
      "INFO:tensorflow:loss = 271.73187, step = 4100 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 659.689\n",
      "INFO:tensorflow:loss = 509.27518, step = 4200 (0.152 sec)\n",
      "INFO:tensorflow:global_step/sec: 657.273\n",
      "INFO:tensorflow:loss = 317.50986, step = 4300 (0.152 sec)\n",
      "INFO:tensorflow:global_step/sec: 553.198\n",
      "INFO:tensorflow:loss = 186.29042, step = 4400 (0.181 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.877\n",
      "INFO:tensorflow:loss = 368.4395, step = 4500 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 556.365\n",
      "INFO:tensorflow:loss = 193.78143, step = 4600 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 570.239\n",
      "INFO:tensorflow:loss = 441.1708, step = 4700 (0.175 sec)\n",
      "INFO:tensorflow:global_step/sec: 567.788\n",
      "INFO:tensorflow:loss = 240.40808, step = 4800 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 587.168\n",
      "INFO:tensorflow:loss = 319.34958, step = 4900 (0.170 sec)\n",
      "INFO:tensorflow:global_step/sec: 613.08\n",
      "INFO:tensorflow:loss = 270.3896, step = 5000 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 615.585\n",
      "INFO:tensorflow:loss = 154.97989, step = 5100 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 618.903\n",
      "INFO:tensorflow:loss = 253.60646, step = 5200 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 613.373\n",
      "INFO:tensorflow:loss = 250.14478, step = 5300 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 618.895\n",
      "INFO:tensorflow:loss = 314.1364, step = 5400 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 615.247\n",
      "INFO:tensorflow:loss = 287.53287, step = 5500 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 579.559\n",
      "INFO:tensorflow:loss = 292.2387, step = 5600 (0.173 sec)\n",
      "INFO:tensorflow:global_step/sec: 564.426\n",
      "INFO:tensorflow:loss = 224.04553, step = 5700 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 568.456\n",
      "INFO:tensorflow:loss = 353.1428, step = 5800 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 563.157\n",
      "INFO:tensorflow:loss = 335.3949, step = 5900 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 569.556\n",
      "INFO:tensorflow:loss = 187.25581, step = 6000 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 560.625\n",
      "INFO:tensorflow:loss = 207.89288, step = 6100 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 564.964\n",
      "INFO:tensorflow:loss = 352.81, step = 6200 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 601.182\n",
      "INFO:tensorflow:loss = 200.35771, step = 6300 (0.166 sec)\n",
      "INFO:tensorflow:global_step/sec: 624.884\n",
      "INFO:tensorflow:loss = 148.6445, step = 6400 (0.160 sec)\n",
      "INFO:tensorflow:global_step/sec: 617.641\n",
      "INFO:tensorflow:loss = 98.60468, step = 6500 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 565.054\n",
      "INFO:tensorflow:loss = 272.97253, step = 6600 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.052\n",
      "INFO:tensorflow:loss = 189.83186, step = 6700 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 559.404\n",
      "INFO:tensorflow:loss = 177.83542, step = 6800 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 616.915\n",
      "INFO:tensorflow:loss = 154.5396, step = 6900 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 586.968\n",
      "INFO:tensorflow:loss = 312.20593, step = 7000 (0.170 sec)\n",
      "INFO:tensorflow:global_step/sec: 584.733\n",
      "INFO:tensorflow:loss = 146.29216, step = 7100 (0.171 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.786\n",
      "INFO:tensorflow:loss = 175.46495, step = 7200 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 652.056\n",
      "INFO:tensorflow:loss = 225.34903, step = 7300 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 642.788\n",
      "INFO:tensorflow:loss = 203.46408, step = 7400 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 657.435\n",
      "INFO:tensorflow:loss = 115.84112, step = 7500 (0.152 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.413\n",
      "INFO:tensorflow:loss = 222.01236, step = 7600 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 566.584\n",
      "INFO:tensorflow:loss = 195.82758, step = 7700 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 635.592\n",
      "INFO:tensorflow:loss = 135.50404, step = 7800 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 581.135\n",
      "INFO:tensorflow:loss = 128.616, step = 7900 (0.172 sec)\n",
      "INFO:tensorflow:global_step/sec: 565.747\n",
      "INFO:tensorflow:loss = 142.26901, step = 8000 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 564.354\n",
      "INFO:tensorflow:loss = 186.27264, step = 8100 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 579.482\n",
      "INFO:tensorflow:loss = 285.70624, step = 8200 (0.172 sec)\n",
      "INFO:tensorflow:global_step/sec: 565.625\n",
      "INFO:tensorflow:loss = 119.08911, step = 8300 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 556.787\n",
      "INFO:tensorflow:loss = 122.49934, step = 8400 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 615.835\n",
      "INFO:tensorflow:loss = 179.74419, step = 8500 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 618.261\n",
      "INFO:tensorflow:loss = 101.167206, step = 8600 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 616.308\n",
      "INFO:tensorflow:loss = 233.63202, step = 8700 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 624.669\n",
      "INFO:tensorflow:loss = 69.64491, step = 8800 (0.160 sec)\n",
      "INFO:tensorflow:global_step/sec: 618.563\n",
      "INFO:tensorflow:loss = 127.40744, step = 8900 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 620.277\n",
      "INFO:tensorflow:loss = 145.69806, step = 9000 (0.161 sec)\n",
      "INFO:tensorflow:global_step/sec: 579.132\n",
      "INFO:tensorflow:loss = 172.43636, step = 9100 (0.173 sec)\n",
      "INFO:tensorflow:global_step/sec: 575.305\n",
      "INFO:tensorflow:loss = 207.42215, step = 9200 (0.174 sec)\n",
      "INFO:tensorflow:global_step/sec: 613.089\n",
      "INFO:tensorflow:loss = 132.4537, step = 9300 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 619.394\n",
      "INFO:tensorflow:loss = 70.08789, step = 9400 (0.161 sec)\n",
      "INFO:tensorflow:global_step/sec: 613.773\n",
      "INFO:tensorflow:loss = 62.70175, step = 9500 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 612.9\n",
      "INFO:tensorflow:loss = 132.35649, step = 9600 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 615.735\n",
      "INFO:tensorflow:loss = 109.410255, step = 9700 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 612.842\n",
      "INFO:tensorflow:loss = 166.54138, step = 9800 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 589.04\n",
      "INFO:tensorflow:loss = 97.47385, step = 9900 (0.170 sec)\n",
      "INFO:tensorflow:global_step/sec: 559.295\n",
      "INFO:tensorflow:loss = 103.19119, step = 10000 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 560.623\n",
      "INFO:tensorflow:loss = 85.13874, step = 10100 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 569.201\n",
      "INFO:tensorflow:loss = 106.877075, step = 10200 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 581.087\n",
      "INFO:tensorflow:loss = 176.70938, step = 10300 (0.172 sec)\n",
      "INFO:tensorflow:global_step/sec: 590.159\n",
      "INFO:tensorflow:loss = 87.25897, step = 10400 (0.169 sec)\n",
      "INFO:tensorflow:global_step/sec: 640.17\n",
      "INFO:tensorflow:loss = 92.061295, step = 10500 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 635.601\n",
      "INFO:tensorflow:loss = 122.7544, step = 10600 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 642.358\n",
      "INFO:tensorflow:loss = 58.609745, step = 10700 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 593.496\n",
      "INFO:tensorflow:loss = 130.04077, step = 10800 (0.168 sec)\n",
      "INFO:tensorflow:global_step/sec: 591.825\n",
      "INFO:tensorflow:loss = 139.86115, step = 10900 (0.169 sec)\n",
      "INFO:tensorflow:global_step/sec: 576.898\n",
      "INFO:tensorflow:loss = 128.72388, step = 11000 (0.173 sec)\n",
      "INFO:tensorflow:global_step/sec: 567.129\n",
      "INFO:tensorflow:loss = 82.31069, step = 11100 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 555.931\n",
      "INFO:tensorflow:loss = 77.19307, step = 11200 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 569.253\n",
      "INFO:tensorflow:loss = 84.63773, step = 11300 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 572.477\n",
      "INFO:tensorflow:loss = 91.148834, step = 11400 (0.175 sec)\n",
      "INFO:tensorflow:global_step/sec: 571.016\n",
      "INFO:tensorflow:loss = 41.38865, step = 11500 (0.175 sec)\n",
      "INFO:tensorflow:global_step/sec: 573.122\n",
      "INFO:tensorflow:loss = 99.62012, step = 11600 (0.174 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.055\n",
      "INFO:tensorflow:loss = 55.67815, step = 11700 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 654.057\n",
      "INFO:tensorflow:loss = 59.75948, step = 11800 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 654.278\n",
      "INFO:tensorflow:loss = 60.011417, step = 11900 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.568\n",
      "INFO:tensorflow:loss = 115.91528, step = 12000 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 568.138\n",
      "INFO:tensorflow:loss = 73.505295, step = 12100 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.474\n",
      "INFO:tensorflow:loss = 88.06746, step = 12200 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 560.304\n",
      "INFO:tensorflow:loss = 85.41072, step = 12300 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 587.284\n",
      "INFO:tensorflow:loss = 56.19171, step = 12400 (0.170 sec)\n",
      "INFO:tensorflow:global_step/sec: 637.918\n",
      "INFO:tensorflow:loss = 64.24606, step = 12500 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 601.848\n",
      "INFO:tensorflow:loss = 52.26082, step = 12600 (0.166 sec)\n",
      "INFO:tensorflow:global_step/sec: 664.047\n",
      "INFO:tensorflow:loss = 52.31598, step = 12700 (0.151 sec)\n",
      "INFO:tensorflow:global_step/sec: 647.043\n",
      "INFO:tensorflow:loss = 99.53073, step = 12800 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 649.28\n",
      "INFO:tensorflow:loss = 85.73153, step = 12900 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 660.834\n",
      "INFO:tensorflow:loss = 59.555412, step = 13000 (0.151 sec)\n",
      "INFO:tensorflow:global_step/sec: 653.213\n",
      "INFO:tensorflow:loss = 69.10744, step = 13100 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 661.769\n",
      "INFO:tensorflow:loss = 66.7493, step = 13200 (0.151 sec)\n",
      "INFO:tensorflow:global_step/sec: 653.19\n",
      "INFO:tensorflow:loss = 50.434845, step = 13300 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 662.245\n",
      "INFO:tensorflow:loss = 119.52582, step = 13400 (0.151 sec)\n",
      "INFO:tensorflow:global_step/sec: 651.11\n",
      "INFO:tensorflow:loss = 39.486324, step = 13500 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 572.621\n",
      "INFO:tensorflow:loss = 36.93599, step = 13600 (0.175 sec)\n",
      "INFO:tensorflow:global_step/sec: 556.868\n",
      "INFO:tensorflow:loss = 35.120228, step = 13700 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 566.421\n",
      "INFO:tensorflow:loss = 28.110342, step = 13800 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 570.946\n",
      "INFO:tensorflow:loss = 50.28643, step = 13900 (0.175 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.298\n",
      "INFO:tensorflow:loss = 34.63925, step = 14000 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 661.018\n",
      "INFO:tensorflow:loss = 75.51688, step = 14100 (0.151 sec)\n",
      "INFO:tensorflow:global_step/sec: 652.163\n",
      "INFO:tensorflow:loss = 58.08398, step = 14200 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.281\n",
      "INFO:tensorflow:loss = 43.38926, step = 14300 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 659.09\n",
      "INFO:tensorflow:loss = 31.798458, step = 14400 (0.152 sec)\n",
      "INFO:tensorflow:global_step/sec: 642.84\n",
      "INFO:tensorflow:loss = 50.548183, step = 14500 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 632.044\n",
      "INFO:tensorflow:loss = 52.485527, step = 14600 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 615.95\n",
      "INFO:tensorflow:loss = 46.583717, step = 14700 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 584.977\n",
      "INFO:tensorflow:loss = 57.482048, step = 14800 (0.171 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.703\n",
      "INFO:tensorflow:loss = 28.527182, step = 14900 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 559.545\n",
      "INFO:tensorflow:loss = 24.597378, step = 15000 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 611.333\n",
      "INFO:tensorflow:loss = 67.47382, step = 15100 (0.164 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.873\n",
      "INFO:tensorflow:loss = 9.842392, step = 15200 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.772\n",
      "INFO:tensorflow:loss = 38.127098, step = 15300 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.294\n",
      "INFO:tensorflow:loss = 32.482162, step = 15400 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 647.809\n",
      "INFO:tensorflow:loss = 21.738888, step = 15500 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.159\n",
      "INFO:tensorflow:loss = 47.9902, step = 15600 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 640.936\n",
      "INFO:tensorflow:loss = 71.79877, step = 15700 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.737\n",
      "INFO:tensorflow:loss = 53.093693, step = 15800 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 566.742\n",
      "INFO:tensorflow:loss = 35.7934, step = 15900 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 626.839\n",
      "INFO:tensorflow:loss = 48.071213, step = 16000 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 654.178\n",
      "INFO:tensorflow:loss = 23.8972, step = 16100 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 562.791\n",
      "INFO:tensorflow:loss = 47.391075, step = 16200 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 653.972\n",
      "INFO:tensorflow:loss = 33.310158, step = 16300 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 656.621\n",
      "INFO:tensorflow:loss = 36.17897, step = 16400 (0.152 sec)\n",
      "INFO:tensorflow:global_step/sec: 639.331\n",
      "INFO:tensorflow:loss = 29.826387, step = 16500 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.783\n",
      "INFO:tensorflow:loss = 39.875584, step = 16600 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.618\n",
      "INFO:tensorflow:loss = 62.336967, step = 16700 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 633.634\n",
      "INFO:tensorflow:loss = 25.50624, step = 16800 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 613.129\n",
      "INFO:tensorflow:loss = 26.36065, step = 16900 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 617.887\n",
      "INFO:tensorflow:loss = 22.199059, step = 17000 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 636.057\n",
      "INFO:tensorflow:loss = 35.419243, step = 17100 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 567.684\n",
      "INFO:tensorflow:loss = 16.926636, step = 17200 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 552.403\n",
      "INFO:tensorflow:loss = 41.73528, step = 17300 (0.181 sec)\n",
      "INFO:tensorflow:global_step/sec: 554.903\n",
      "INFO:tensorflow:loss = 60.46621, step = 17400 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 606.698\n",
      "INFO:tensorflow:loss = 9.7304735, step = 17500 (0.165 sec)\n",
      "INFO:tensorflow:global_step/sec: 629.775\n",
      "INFO:tensorflow:loss = 23.861652, step = 17600 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 580.814\n",
      "INFO:tensorflow:loss = 44.88912, step = 17700 (0.172 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.919\n",
      "INFO:tensorflow:loss = 21.361008, step = 17800 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 553.493\n",
      "INFO:tensorflow:loss = 33.784424, step = 17900 (0.181 sec)\n",
      "INFO:tensorflow:global_step/sec: 560.444\n",
      "INFO:tensorflow:loss = 20.45843, step = 18000 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 565.069\n",
      "INFO:tensorflow:loss = 31.591064, step = 18100 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 562.811\n",
      "INFO:tensorflow:loss = 13.793056, step = 18200 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 556.338\n",
      "INFO:tensorflow:loss = 45.03756, step = 18300 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 556.655\n",
      "INFO:tensorflow:loss = 11.519857, step = 18400 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 643.54\n",
      "INFO:tensorflow:loss = 34.568317, step = 18500 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 649.191\n",
      "INFO:tensorflow:loss = 44.755013, step = 18600 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 576.328\n",
      "INFO:tensorflow:loss = 14.43598, step = 18700 (0.174 sec)\n",
      "INFO:tensorflow:global_step/sec: 564.586\n",
      "INFO:tensorflow:loss = 10.982545, step = 18800 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.083\n",
      "INFO:tensorflow:loss = 30.731411, step = 18900 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 643.407\n",
      "INFO:tensorflow:loss = 19.440567, step = 19000 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 607.951\n",
      "INFO:tensorflow:loss = 63.62726, step = 19100 (0.164 sec)\n",
      "INFO:tensorflow:global_step/sec: 573.549\n",
      "INFO:tensorflow:loss = 13.900019, step = 19200 (0.175 sec)\n",
      "INFO:tensorflow:global_step/sec: 580.514\n",
      "INFO:tensorflow:loss = 47.006104, step = 19300 (0.172 sec)\n",
      "INFO:tensorflow:global_step/sec: 642.58\n",
      "INFO:tensorflow:loss = 12.699298, step = 19400 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.932\n",
      "INFO:tensorflow:loss = 20.420689, step = 19500 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.287\n",
      "INFO:tensorflow:loss = 11.957592, step = 19600 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 635.855\n",
      "INFO:tensorflow:loss = 20.387909, step = 19700 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.552\n",
      "INFO:tensorflow:loss = 63.668373, step = 19800 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 564.799\n",
      "INFO:tensorflow:loss = 29.952385, step = 19900 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 561.545\n",
      "INFO:tensorflow:loss = 3.468977, step = 20000 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 564.833\n",
      "INFO:tensorflow:loss = 14.017977, step = 20100 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 559.31\n",
      "INFO:tensorflow:loss = 23.940512, step = 20200 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 561.595\n",
      "INFO:tensorflow:loss = 15.872161, step = 20300 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.326\n",
      "INFO:tensorflow:loss = 41.10446, step = 20400 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 597.31\n",
      "INFO:tensorflow:loss = 40.06671, step = 20500 (0.167 sec)\n",
      "INFO:tensorflow:global_step/sec: 615.561\n",
      "INFO:tensorflow:loss = 40.576504, step = 20600 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 599.323\n",
      "INFO:tensorflow:loss = 9.951186, step = 20700 (0.167 sec)\n",
      "INFO:tensorflow:global_step/sec: 564.476\n",
      "INFO:tensorflow:loss = 26.887535, step = 20800 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 558.206\n",
      "INFO:tensorflow:loss = 22.223, step = 20900 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 558.813\n",
      "INFO:tensorflow:loss = 26.191727, step = 21000 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 556.229\n",
      "INFO:tensorflow:loss = 18.055063, step = 21100 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 554.705\n",
      "INFO:tensorflow:loss = 17.920183, step = 21200 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 562.668\n",
      "INFO:tensorflow:loss = 26.513557, step = 21300 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 640.564\n",
      "INFO:tensorflow:loss = 21.903809, step = 21400 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 647.2\n",
      "INFO:tensorflow:loss = 30.5953, step = 21500 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 638.928\n",
      "INFO:tensorflow:loss = 27.62001, step = 21600 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.709\n",
      "INFO:tensorflow:loss = 20.175642, step = 21700 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.56\n",
      "INFO:tensorflow:loss = 4.968134, step = 21800 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 639.462\n",
      "INFO:tensorflow:loss = 51.621918, step = 21900 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 640.555\n",
      "INFO:tensorflow:loss = 32.733124, step = 22000 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 633.889\n",
      "INFO:tensorflow:loss = 11.014718, step = 22100 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 630.786\n",
      "INFO:tensorflow:loss = 14.58586, step = 22200 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 563.225\n",
      "INFO:tensorflow:loss = 8.891371, step = 22300 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 560.618\n",
      "INFO:tensorflow:loss = 38.58745, step = 22400 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 556.76\n",
      "INFO:tensorflow:loss = 22.485962, step = 22500 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 563.099\n",
      "INFO:tensorflow:loss = 15.741159, step = 22600 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.673\n",
      "INFO:tensorflow:loss = 15.188747, step = 22700 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 589.91\n",
      "INFO:tensorflow:loss = 22.645355, step = 22800 (0.169 sec)\n",
      "INFO:tensorflow:global_step/sec: 556.956\n",
      "INFO:tensorflow:loss = 12.435162, step = 22900 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 558.367\n",
      "INFO:tensorflow:loss = 28.399435, step = 23000 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 554.499\n",
      "INFO:tensorflow:loss = 29.847202, step = 23100 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.706\n",
      "INFO:tensorflow:loss = 24.096184, step = 23200 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 633.393\n",
      "INFO:tensorflow:loss = 29.605507, step = 23300 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 638.07\n",
      "INFO:tensorflow:loss = 6.675656, step = 23400 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.546\n",
      "INFO:tensorflow:loss = 8.849547, step = 23500 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.03\n",
      "INFO:tensorflow:loss = 22.545252, step = 23600 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.982\n",
      "INFO:tensorflow:loss = 32.837273, step = 23700 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 652.554\n",
      "INFO:tensorflow:loss = 23.164417, step = 23800 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.183\n",
      "INFO:tensorflow:loss = 20.617609, step = 23900 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.653\n",
      "INFO:tensorflow:loss = 14.767559, step = 24000 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 635.294\n",
      "INFO:tensorflow:loss = 38.283543, step = 24100 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.853\n",
      "INFO:tensorflow:loss = 44.3748, step = 24200 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 637.927\n",
      "INFO:tensorflow:loss = 40.130806, step = 24300 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 636.607\n",
      "INFO:tensorflow:loss = 13.896637, step = 24400 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 640.791\n",
      "INFO:tensorflow:loss = 15.867412, step = 24500 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 643.12\n",
      "INFO:tensorflow:loss = 27.252453, step = 24600 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 637.837\n",
      "INFO:tensorflow:loss = 10.193189, step = 24700 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 558.504\n",
      "INFO:tensorflow:loss = 7.39489, step = 24800 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 552.473\n",
      "INFO:tensorflow:loss = 43.868538, step = 24900 (0.181 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.42\n",
      "INFO:tensorflow:loss = 39.280296, step = 25000 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 557.644\n",
      "INFO:tensorflow:loss = 32.257843, step = 25100 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 552.62\n",
      "INFO:tensorflow:loss = 15.869023, step = 25200 (0.181 sec)\n",
      "INFO:tensorflow:global_step/sec: 596.984\n",
      "INFO:tensorflow:loss = 14.115913, step = 25300 (0.167 sec)\n",
      "INFO:tensorflow:global_step/sec: 562.627\n",
      "INFO:tensorflow:loss = 10.347841, step = 25400 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 555.116\n",
      "INFO:tensorflow:loss = 28.828302, step = 25500 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 562.795\n",
      "INFO:tensorflow:loss = 21.246841, step = 25600 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 626.387\n",
      "INFO:tensorflow:loss = 9.311673, step = 25700 (0.160 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.753\n",
      "INFO:tensorflow:loss = 24.5037, step = 25800 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.638\n",
      "INFO:tensorflow:loss = 16.885513, step = 25900 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 647.599\n",
      "INFO:tensorflow:loss = 17.163336, step = 26000 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 637.261\n",
      "INFO:tensorflow:loss = 19.075056, step = 26100 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 630.96\n",
      "INFO:tensorflow:loss = 15.663348, step = 26200 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.268\n",
      "INFO:tensorflow:loss = 18.120064, step = 26300 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 647.007\n",
      "INFO:tensorflow:loss = 41.735718, step = 26400 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.607\n",
      "INFO:tensorflow:loss = 17.466406, step = 26500 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.538\n",
      "INFO:tensorflow:loss = 27.561075, step = 26600 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.795\n",
      "INFO:tensorflow:loss = 9.69916, step = 26700 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 631.119\n",
      "INFO:tensorflow:loss = 83.92375, step = 26800 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 639.714\n",
      "INFO:tensorflow:loss = 16.178406, step = 26900 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 637.172\n",
      "INFO:tensorflow:loss = 14.685428, step = 27000 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 637.822\n",
      "INFO:tensorflow:loss = 16.487114, step = 27100 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 626.523\n",
      "INFO:tensorflow:loss = 6.233879, step = 27200 (0.160 sec)\n",
      "INFO:tensorflow:global_step/sec: 562.631\n",
      "INFO:tensorflow:loss = 25.59382, step = 27300 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 588.092\n",
      "INFO:tensorflow:loss = 7.877035, step = 27400 (0.170 sec)\n",
      "INFO:tensorflow:global_step/sec: 625.747\n",
      "INFO:tensorflow:loss = 26.259216, step = 27500 (0.160 sec)\n",
      "INFO:tensorflow:global_step/sec: 607.397\n",
      "INFO:tensorflow:loss = 18.831963, step = 27600 (0.165 sec)\n",
      "INFO:tensorflow:global_step/sec: 614.569\n",
      "INFO:tensorflow:loss = 10.519957, step = 27700 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 610.519\n",
      "INFO:tensorflow:loss = 29.63565, step = 27800 (0.164 sec)\n",
      "INFO:tensorflow:global_step/sec: 617.845\n",
      "INFO:tensorflow:loss = 15.680868, step = 27900 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 610.635\n",
      "INFO:tensorflow:loss = 31.339779, step = 28000 (0.164 sec)\n",
      "INFO:tensorflow:global_step/sec: 610.08\n",
      "INFO:tensorflow:loss = 16.819326, step = 28100 (0.164 sec)\n",
      "INFO:tensorflow:global_step/sec: 616.965\n",
      "INFO:tensorflow:loss = 1.2390225, step = 28200 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 614.248\n",
      "INFO:tensorflow:loss = 18.569372, step = 28300 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 631.205\n",
      "INFO:tensorflow:loss = 69.872086, step = 28400 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 647.207\n",
      "INFO:tensorflow:loss = 7.713742, step = 28500 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.569\n",
      "INFO:tensorflow:loss = 8.96863, step = 28600 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 635.145\n",
      "INFO:tensorflow:loss = 8.195501, step = 28700 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 639.996\n",
      "INFO:tensorflow:loss = 7.895544, step = 28800 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 634.209\n",
      "INFO:tensorflow:loss = 9.781753, step = 28900 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 642.806\n",
      "INFO:tensorflow:loss = 23.838917, step = 29000 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 639.762\n",
      "INFO:tensorflow:loss = 13.926859, step = 29100 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 635.524\n",
      "INFO:tensorflow:loss = 20.532545, step = 29200 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.009\n",
      "INFO:tensorflow:loss = 15.29386, step = 29300 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 636.193\n",
      "INFO:tensorflow:loss = 9.65447, step = 29400 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 636.497\n",
      "INFO:tensorflow:loss = 17.778759, step = 29500 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 638.647\n",
      "INFO:tensorflow:loss = 53.59935, step = 29600 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.218\n",
      "INFO:tensorflow:loss = 1.7887146, step = 29700 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 632.812\n",
      "INFO:tensorflow:loss = 9.185707, step = 29800 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 636.626\n",
      "INFO:tensorflow:loss = 2.0348744, step = 29900 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.384\n",
      "INFO:tensorflow:loss = 39.938457, step = 30000 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 636.928\n",
      "INFO:tensorflow:loss = 35.079205, step = 30100 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 631.889\n",
      "INFO:tensorflow:loss = 11.830448, step = 30200 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 629.569\n",
      "INFO:tensorflow:loss = 7.7222857, step = 30300 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 629.26\n",
      "INFO:tensorflow:loss = 5.0043783, step = 30400 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 633.129\n",
      "INFO:tensorflow:loss = 28.834631, step = 30500 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 560.8\n",
      "INFO:tensorflow:loss = 49.048584, step = 30600 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 630.476\n",
      "INFO:tensorflow:loss = 12.857588, step = 30700 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 576.142\n",
      "INFO:tensorflow:loss = 18.584694, step = 30800 (0.174 sec)\n",
      "INFO:tensorflow:global_step/sec: 554.342\n",
      "INFO:tensorflow:loss = 30.525942, step = 30900 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 567.578\n",
      "INFO:tensorflow:loss = 33.63379, step = 31000 (0.176 sec)\n",
      "INFO:tensorflow:global_step/sec: 560.022\n",
      "INFO:tensorflow:loss = 14.771801, step = 31100 (0.178 sec)\n",
      "INFO:tensorflow:global_step/sec: 558.288\n",
      "INFO:tensorflow:loss = 3.7000563, step = 31200 (0.179 sec)\n",
      "INFO:tensorflow:global_step/sec: 625.088\n",
      "INFO:tensorflow:loss = 3.9122221, step = 31300 (0.160 sec)\n",
      "INFO:tensorflow:global_step/sec: 619.433\n",
      "INFO:tensorflow:loss = 13.096632, step = 31400 (0.161 sec)\n",
      "INFO:tensorflow:global_step/sec: 630.211\n",
      "INFO:tensorflow:loss = 35.163918, step = 31500 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.753\n",
      "INFO:tensorflow:loss = 18.955883, step = 31600 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 642.018\n",
      "INFO:tensorflow:loss = 24.371445, step = 31700 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 654.459\n",
      "INFO:tensorflow:loss = 10.686918, step = 31800 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.877\n",
      "INFO:tensorflow:loss = 10.00495, step = 31900 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 636.379\n",
      "INFO:tensorflow:loss = 34.196156, step = 32000 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 639.306\n",
      "INFO:tensorflow:loss = 17.267265, step = 32100 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 628.14\n",
      "INFO:tensorflow:loss = 3.5982966, step = 32200 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 654.144\n",
      "INFO:tensorflow:loss = 22.826725, step = 32300 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 656.323\n",
      "INFO:tensorflow:loss = 18.253302, step = 32400 (0.152 sec)\n",
      "INFO:tensorflow:global_step/sec: 652.179\n",
      "INFO:tensorflow:loss = 19.89624, step = 32500 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.347\n",
      "INFO:tensorflow:loss = 26.507126, step = 32600 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 639.534\n",
      "INFO:tensorflow:loss = 9.670689, step = 32700 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 628.019\n",
      "INFO:tensorflow:loss = 19.151632, step = 32800 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.71\n",
      "INFO:tensorflow:loss = 11.301746, step = 32900 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 632.64\n",
      "INFO:tensorflow:loss = 39.349358, step = 33000 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.028\n",
      "INFO:tensorflow:loss = 32.2932, step = 33100 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 627.79\n",
      "INFO:tensorflow:loss = 14.889015, step = 33200 (0.159 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.24\n",
      "INFO:tensorflow:loss = 41.704803, step = 33300 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 635.648\n",
      "INFO:tensorflow:loss = 15.027176, step = 33400 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.35\n",
      "INFO:tensorflow:loss = 12.243839, step = 33500 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 651.522\n",
      "INFO:tensorflow:loss = 5.87178, step = 33600 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.649\n",
      "INFO:tensorflow:loss = 16.51314, step = 33700 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.37\n",
      "INFO:tensorflow:loss = 15.912634, step = 33800 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.818\n",
      "INFO:tensorflow:loss = 6.8702297, step = 33900 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 637.876\n",
      "INFO:tensorflow:loss = 9.3833275, step = 34000 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 635.339\n",
      "INFO:tensorflow:loss = 6.315841, step = 34100 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.389\n",
      "INFO:tensorflow:loss = 23.294544, step = 34200 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 626.389\n",
      "INFO:tensorflow:loss = 14.866375, step = 34300 (0.160 sec)\n",
      "INFO:tensorflow:global_step/sec: 636.443\n",
      "INFO:tensorflow:loss = 23.482327, step = 34400 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.709\n",
      "INFO:tensorflow:loss = 5.9481287, step = 34500 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.761\n",
      "INFO:tensorflow:loss = 10.6761265, step = 34600 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 638.862\n",
      "INFO:tensorflow:loss = 4.912558, step = 34700 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 642.065\n",
      "INFO:tensorflow:loss = 25.11901, step = 34800 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 612.919\n",
      "INFO:tensorflow:loss = 8.14024, step = 34900 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 556.115\n",
      "INFO:tensorflow:loss = 17.55474, step = 35000 (0.180 sec)\n",
      "INFO:tensorflow:global_step/sec: 553.651\n",
      "INFO:tensorflow:loss = 5.6831026, step = 35100 (0.181 sec)\n",
      "INFO:tensorflow:global_step/sec: 563.169\n",
      "INFO:tensorflow:loss = 12.501162, step = 35200 (0.177 sec)\n",
      "INFO:tensorflow:global_step/sec: 643.02\n",
      "INFO:tensorflow:loss = 25.764713, step = 35300 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.011\n",
      "INFO:tensorflow:loss = 10.363308, step = 35400 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 638.251\n",
      "INFO:tensorflow:loss = 23.721325, step = 35500 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 640.312\n",
      "INFO:tensorflow:loss = 13.199163, step = 35600 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 651.36\n",
      "INFO:tensorflow:loss = 31.616188, step = 35700 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 613.686\n",
      "INFO:tensorflow:loss = 5.4822397, step = 35800 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 618.484\n",
      "INFO:tensorflow:loss = 9.069538, step = 35900 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 617.06\n",
      "INFO:tensorflow:loss = 8.100636, step = 36000 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 615.627\n",
      "INFO:tensorflow:loss = 16.85956, step = 36100 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 612.812\n",
      "INFO:tensorflow:loss = 18.85649, step = 36200 (0.163 sec)\n",
      "INFO:tensorflow:global_step/sec: 616.103\n",
      "INFO:tensorflow:loss = 13.850594, step = 36300 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 625.165\n",
      "INFO:tensorflow:loss = 51.77982, step = 36400 (0.160 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.928\n",
      "INFO:tensorflow:loss = 21.499207, step = 36500 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.116\n",
      "INFO:tensorflow:loss = 9.992126, step = 36600 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.039\n",
      "INFO:tensorflow:loss = 11.250916, step = 36700 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 647.816\n",
      "INFO:tensorflow:loss = 9.30033, step = 36800 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.269\n",
      "INFO:tensorflow:loss = 4.329362, step = 36900 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 640.78\n",
      "INFO:tensorflow:loss = 10.897452, step = 37000 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.487\n",
      "INFO:tensorflow:loss = 9.148363, step = 37100 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 651.223\n",
      "INFO:tensorflow:loss = 30.404154, step = 37200 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.637\n",
      "INFO:tensorflow:loss = 14.910101, step = 37300 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.827\n",
      "INFO:tensorflow:loss = 15.732323, step = 37400 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 652.423\n",
      "INFO:tensorflow:loss = 12.93858, step = 37500 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.518\n",
      "INFO:tensorflow:loss = 31.476284, step = 37600 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.811\n",
      "INFO:tensorflow:loss = 12.837719, step = 37700 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 635.591\n",
      "INFO:tensorflow:loss = 6.9836364, step = 37800 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 640.438\n",
      "INFO:tensorflow:loss = 3.7009566, step = 37900 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 652.57\n",
      "INFO:tensorflow:loss = 15.457409, step = 38000 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 641.224\n",
      "INFO:tensorflow:loss = 15.640904, step = 38100 (0.156 sec)\n",
      "INFO:tensorflow:global_step/sec: 652.626\n",
      "INFO:tensorflow:loss = 3.6458635, step = 38200 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 654.56\n",
      "INFO:tensorflow:loss = 7.718712, step = 38300 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.262\n",
      "INFO:tensorflow:loss = 17.353992, step = 38400 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.266\n",
      "INFO:tensorflow:loss = 10.383542, step = 38500 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 646.131\n",
      "INFO:tensorflow:loss = 19.498116, step = 38600 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 650.171\n",
      "INFO:tensorflow:loss = 50.384224, step = 38700 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 649.759\n",
      "INFO:tensorflow:loss = 5.805867, step = 38800 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 634.323\n",
      "INFO:tensorflow:loss = 7.7724056, step = 38900 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 633.254\n",
      "INFO:tensorflow:loss = 7.2014084, step = 39000 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 648.197\n",
      "INFO:tensorflow:loss = 17.135563, step = 39100 (0.154 sec)\n",
      "INFO:tensorflow:global_step/sec: 651.45\n",
      "INFO:tensorflow:loss = 26.166622, step = 39200 (0.153 sec)\n",
      "INFO:tensorflow:global_step/sec: 644.187\n",
      "INFO:tensorflow:loss = 16.454807, step = 39300 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 645.437\n",
      "INFO:tensorflow:loss = 25.390144, step = 39400 (0.155 sec)\n",
      "INFO:tensorflow:global_step/sec: 637.175\n",
      "INFO:tensorflow:loss = 17.874966, step = 39500 (0.157 sec)\n",
      "INFO:tensorflow:global_step/sec: 632.669\n",
      "INFO:tensorflow:loss = 11.231724, step = 39600 (0.158 sec)\n",
      "INFO:tensorflow:global_step/sec: 617.176\n",
      "INFO:tensorflow:loss = 16.137482, step = 39700 (0.162 sec)\n",
      "INFO:tensorflow:global_step/sec: 609.228\n",
      "INFO:tensorflow:loss = 20.8172, step = 39800 (0.164 sec)\n",
      "INFO:tensorflow:global_step/sec: 587.507\n",
      "INFO:tensorflow:loss = 24.547935, step = 39900 (0.170 sec)\n",
      "INFO:tensorflow:Saving checkpoints for 40000 into models/autompg-dnnregressor/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 43.661736.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow_estimator.python.estimator.canned.dnn.DNNRegressorV2 at 0x7f47cde76a10>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "EPOCHS = 1000\n",
    "BATCH_SIZE = 8\n",
    "total_steps = EPOCHS * int(np.ceil(len(df_train) / BATCH_SIZE))\n",
    "print('Training Steps:', total_steps)\n",
    "\n",
    "regressor.train(\n",
    "    input_fn=lambda:train_input_fn(df_train_norm, batch_size=BATCH_SIZE),\n",
    "    steps=total_steps)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Using default config.\n",
      "INFO:tensorflow:Using config: {'_model_dir': 'models/autompg-dnnregressor/', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true\n",
      "graph_options {\n",
      "  rewrite_options {\n",
      "    meta_optimizer_iterations: ONE\n",
      "  }\n",
      "}\n",
      ", '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': 100, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7f47c054ddd0>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}\n"
     ]
    }
   ],
   "source": [
    "reloaded_regressor = tf.estimator.DNNRegressor(\n",
    "    feature_columns=all_feature_columns,\n",
    "    hidden_units=[32, 10],\n",
    "    warm_start_from='models/autompg-dnnregressor/',\n",
    "    model_dir='models/autompg-dnnregressor/')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling model_fn.\n",
      "WARNING:tensorflow:Layer dnn is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.\n",
      "\n",
      "If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.\n",
      "\n",
      "To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.\n",
      "\n",
      "INFO:tensorflow:Done calling model_fn.\n",
      "INFO:tensorflow:Starting evaluation at 2019-11-03T11:17:46Z\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Restoring parameters from models/autompg-dnnregressor/model.ckpt-40000\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n",
      "INFO:tensorflow:Finished evaluation at 2019-11-03-11:17:47\n",
      "INFO:tensorflow:Saving dict for global step 40000: average_loss = 19.54804, global_step = 40000, label/mean = 23.611393, loss = 19.48917, prediction/mean = 21.74643\n",
      "INFO:tensorflow:Saving 'checkpoint_path' summary for global step 40000: models/autompg-dnnregressor/model.ckpt-40000\n",
      "average_loss    19.54804039001465\n",
      "label/mean      23.611392974853516\n",
      "loss            19.48917007446289\n",
      "prediction/mean 21.746429443359375\n",
      "global_step     40000\n",
      "Average-Loss 19.5480\n"
     ]
    }
   ],
   "source": [
    "def eval_input_fn(df_test, batch_size=8):\n",
    "    df = df_test.copy()\n",
    "    test_x, test_y = df, df.pop('MPG')\n",
    "    dataset = tf.data.Dataset.from_tensor_slices((dict(test_x), test_y))\n",
    "\n",
    "    return dataset.batch(batch_size)\n",
    "\n",
    "eval_results = reloaded_regressor.evaluate(\n",
    "    input_fn=lambda:eval_input_fn(df_test_norm, batch_size=8))\n",
    "\n",
    "for key in eval_results:\n",
    "    print('{:15s} {}'.format(key, eval_results[key]))\n",
    "    \n",
    "print('Average-Loss {:.4f}'.format(eval_results['average_loss']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling model_fn.\n",
      "WARNING:tensorflow:Layer dnn is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.\n",
      "\n",
      "If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.\n",
      "\n",
      "To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.\n",
      "\n",
      "INFO:tensorflow:Done calling model_fn.\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Restoring parameters from models/autompg-dnnregressor/model.ckpt-40000\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n",
      "{'predictions': array([23.719353], dtype=float32)}\n"
     ]
    }
   ],
   "source": [
    "pred_res = regressor.predict(input_fn=lambda: eval_input_fn(df_test_norm, batch_size=8))\n",
    "\n",
    "print(next(iter(pred_res)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Boosted Tree Regressor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Using default config.\n",
      "WARNING:tensorflow:Using temporary folder as model directory: /tmp/tmpbzo1p2wi\n",
      "INFO:tensorflow:Using config: {'_model_dir': '/tmp/tmpbzo1p2wi', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true\n",
      "graph_options {\n",
      "  rewrite_options {\n",
      "    meta_optimizer_iterations: ONE\n",
      "  }\n",
      "}\n",
      ", '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': 100, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7f47bc30b7d0>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}\n",
      "INFO:tensorflow:Calling model_fn.\n",
      "WARNING:tensorflow:From /home/vahid/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow_estimator/python/estimator/canned/boosted_trees.py:214: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use `tf.cast` instead.\n",
      "INFO:tensorflow:Done calling model_fn.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "WARNING:tensorflow:Issue encountered when serializing resources.\n",
      "Type is unsupported, or the types of the items don't match field type in CollectionDef. Note this is a warning and probably safe to ignore.\n",
      "'_Resource' object has no attribute 'name'\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n",
      "WARNING:tensorflow:Issue encountered when serializing resources.\n",
      "Type is unsupported, or the types of the items don't match field type in CollectionDef. Note this is a warning and probably safe to ignore.\n",
      "'_Resource' object has no attribute 'name'\n",
      "INFO:tensorflow:Saving checkpoints for 0 into /tmp/tmpbzo1p2wi/model.ckpt.\n",
      "WARNING:tensorflow:Issue encountered when serializing resources.\n",
      "Type is unsupported, or the types of the items don't match field type in CollectionDef. Note this is a warning and probably safe to ignore.\n",
      "'_Resource' object has no attribute 'name'\n",
      "INFO:tensorflow:loss = 402.19623, step = 0\n",
      "WARNING:tensorflow:It seems that global step (tf.train.get_global_step) has not been increased. Current value (could be stable): 0 vs previous value: 0. You could increase the global step by passing tf.train.get_global_step() to Optimizer.apply_gradients or Optimizer.minimize.\n",
      "WARNING:tensorflow:It seems that global step (tf.train.get_global_step) has not been increased. Current value (could be stable): 0 vs previous value: 0. You could increase the global step by passing tf.train.get_global_step() to Optimizer.apply_gradients or Optimizer.minimize.\n",
      "WARNING:tensorflow:It seems that global step (tf.train.get_global_step) has not been increased. Current value (could be stable): 0 vs previous value: 0. You could increase the global step by passing tf.train.get_global_step() to Optimizer.apply_gradients or Optimizer.minimize.\n",
      "WARNING:tensorflow:It seems that global step (tf.train.get_global_step) has not been increased. Current value (could be stable): 0 vs previous value: 0. You could increase the global step by passing tf.train.get_global_step() to Optimizer.apply_gradients or Optimizer.minimize.\n",
      "WARNING:tensorflow:It seems that global step (tf.train.get_global_step) has not been increased. Current value (could be stable): 0 vs previous value: 0. You could increase the global step by passing tf.train.get_global_step() to Optimizer.apply_gradients or Optimizer.minimize.\n",
      "INFO:tensorflow:loss = 289.26328, step = 80 (0.462 sec)\n",
      "INFO:tensorflow:global_step/sec: 157.704\n",
      "INFO:tensorflow:loss = 93.58242, step = 180 (0.363 sec)\n",
      "INFO:tensorflow:global_step/sec: 422.808\n",
      "INFO:tensorflow:loss = 45.606873, step = 280 (0.243 sec)\n",
      "INFO:tensorflow:global_step/sec: 416.715\n",
      "INFO:tensorflow:loss = 19.545433, step = 380 (0.240 sec)\n",
      "INFO:tensorflow:global_step/sec: 416.626\n",
      "INFO:tensorflow:loss = 6.4179554, step = 480 (0.245 sec)\n",
      "INFO:tensorflow:global_step/sec: 407.822\n",
      "INFO:tensorflow:loss = 4.7701707, step = 580 (0.231 sec)\n",
      "INFO:tensorflow:global_step/sec: 408.05\n",
      "INFO:tensorflow:loss = 4.569898, step = 680 (0.244 sec)\n",
      "INFO:tensorflow:global_step/sec: 420.57\n",
      "INFO:tensorflow:loss = 2.5075686, step = 780 (0.249 sec)\n",
      "INFO:tensorflow:global_step/sec: 410.68\n",
      "INFO:tensorflow:loss = 2.6939745, step = 880 (0.244 sec)\n",
      "INFO:tensorflow:global_step/sec: 411.964\n",
      "INFO:tensorflow:loss = 1.5966964, step = 980 (0.248 sec)\n",
      "INFO:tensorflow:global_step/sec: 403.965\n",
      "INFO:tensorflow:loss = 3.3678646, step = 1080 (0.250 sec)\n",
      "INFO:tensorflow:global_step/sec: 398.728\n",
      "INFO:tensorflow:loss = 2.3181179, step = 1180 (0.238 sec)\n",
      "INFO:tensorflow:global_step/sec: 396.897\n",
      "INFO:tensorflow:loss = 1.8086417, step = 1280 (0.250 sec)\n",
      "INFO:tensorflow:global_step/sec: 414.237\n",
      "INFO:tensorflow:loss = 0.6904925, step = 1380 (0.246 sec)\n",
      "INFO:tensorflow:global_step/sec: 411.693\n",
      "INFO:tensorflow:loss = 1.8734654, step = 1480 (0.250 sec)\n",
      "INFO:tensorflow:global_step/sec: 401.569\n",
      "INFO:tensorflow:loss = 2.5979433, step = 1580 (0.254 sec)\n",
      "INFO:tensorflow:global_step/sec: 395.667\n",
      "INFO:tensorflow:loss = 2.0128171, step = 1680 (0.256 sec)\n",
      "INFO:tensorflow:global_step/sec: 392.234\n",
      "INFO:tensorflow:loss = 2.469627, step = 1780 (0.244 sec)\n",
      "INFO:tensorflow:global_step/sec: 386.751\n",
      "INFO:tensorflow:loss = 0.87159, step = 1880 (0.253 sec)\n",
      "INFO:tensorflow:global_step/sec: 404.765\n",
      "INFO:tensorflow:loss = 0.80283445, step = 1980 (0.254 sec)\n",
      "INFO:tensorflow:global_step/sec: 401.5\n",
      "INFO:tensorflow:loss = 1.524719, step = 2080 (0.261 sec)\n",
      "INFO:tensorflow:global_step/sec: 385.878\n",
      "INFO:tensorflow:loss = 1.0228136, step = 2180 (0.261 sec)\n",
      "INFO:tensorflow:global_step/sec: 382.386\n",
      "INFO:tensorflow:loss = 1.0036705, step = 2280 (0.263 sec)\n",
      "INFO:tensorflow:global_step/sec: 382.23\n",
      "INFO:tensorflow:loss = 1.0771171, step = 2380 (0.245 sec)\n",
      "INFO:tensorflow:global_step/sec: 388.433\n",
      "INFO:tensorflow:loss = 0.9643565, step = 2480 (0.251 sec)\n",
      "INFO:tensorflow:global_step/sec: 409.442\n",
      "INFO:tensorflow:loss = 1.4598124, step = 2580 (0.264 sec)\n",
      "INFO:tensorflow:global_step/sec: 382.398\n",
      "INFO:tensorflow:loss = 0.7518444, step = 2680 (0.260 sec)\n",
      "INFO:tensorflow:global_step/sec: 387.657\n",
      "INFO:tensorflow:loss = 0.71297884, step = 2780 (0.260 sec)\n",
      "INFO:tensorflow:global_step/sec: 387.516\n",
      "INFO:tensorflow:loss = 0.21006158, step = 2880 (0.261 sec)\n",
      "INFO:tensorflow:global_step/sec: 380.228\n",
      "INFO:tensorflow:loss = 0.64975756, step = 2980 (0.252 sec)\n",
      "INFO:tensorflow:global_step/sec: 375.953\n",
      "INFO:tensorflow:loss = 0.3568688, step = 3080 (0.262 sec)\n",
      "INFO:tensorflow:global_step/sec: 394.311\n",
      "INFO:tensorflow:loss = 1.0947809, step = 3180 (0.260 sec)\n",
      "INFO:tensorflow:global_step/sec: 389.576\n",
      "INFO:tensorflow:loss = 0.38473517, step = 3280 (0.262 sec)\n",
      "INFO:tensorflow:global_step/sec: 383.038\n",
      "INFO:tensorflow:loss = 0.37087482, step = 3380 (0.266 sec)\n",
      "INFO:tensorflow:global_step/sec: 377.258\n",
      "INFO:tensorflow:loss = 0.37313935, step = 3480 (0.268 sec)\n",
      "INFO:tensorflow:global_step/sec: 375.779\n",
      "INFO:tensorflow:loss = 0.6371509, step = 3580 (0.253 sec)\n",
      "INFO:tensorflow:global_step/sec: 376.039\n",
      "INFO:tensorflow:loss = 0.6737277, step = 3680 (0.258 sec)\n",
      "INFO:tensorflow:global_step/sec: 397.449\n",
      "INFO:tensorflow:loss = 0.22763562, step = 3780 (0.264 sec)\n",
      "INFO:tensorflow:global_step/sec: 379.907\n",
      "INFO:tensorflow:loss = 0.70576984, step = 3880 (0.270 sec)\n",
      "INFO:tensorflow:global_step/sec: 375.692\n",
      "INFO:tensorflow:loss = 0.32033288, step = 3980 (0.266 sec)\n",
      "INFO:tensorflow:global_step/sec: 376.935\n",
      "INFO:tensorflow:loss = 0.5732076, step = 4080 (0.271 sec)\n",
      "INFO:tensorflow:global_step/sec: 369.125\n",
      "INFO:tensorflow:loss = 0.22866802, step = 4180 (0.257 sec)\n",
      "INFO:tensorflow:global_step/sec: 370.509\n",
      "INFO:tensorflow:loss = 0.27701426, step = 4280 (0.262 sec)\n",
      "INFO:tensorflow:global_step/sec: 388.812\n",
      "INFO:tensorflow:loss = 0.2290253, step = 4380 (0.273 sec)\n",
      "INFO:tensorflow:global_step/sec: 373.834\n",
      "INFO:tensorflow:loss = 0.24748756, step = 4480 (0.270 sec)\n",
      "INFO:tensorflow:global_step/sec: 373.023\n",
      "INFO:tensorflow:loss = 0.2879139, step = 4580 (0.275 sec)\n",
      "INFO:tensorflow:global_step/sec: 364.77\n",
      "INFO:tensorflow:loss = 0.28078204, step = 4680 (0.272 sec)\n",
      "INFO:tensorflow:global_step/sec: 368.265\n",
      "INFO:tensorflow:loss = 0.1984863, step = 4780 (0.261 sec)\n",
      "INFO:tensorflow:global_step/sec: 363.698\n",
      "INFO:tensorflow:loss = 0.31559613, step = 4880 (0.271 sec)\n",
      "INFO:tensorflow:global_step/sec: 377.83\n",
      "INFO:tensorflow:loss = 0.2904449, step = 4980 (0.277 sec)\n",
      "INFO:tensorflow:global_step/sec: 363.8\n",
      "INFO:tensorflow:loss = 0.28680754, step = 5080 (0.275 sec)\n",
      "INFO:tensorflow:global_step/sec: 367.857\n",
      "INFO:tensorflow:loss = 0.374867, step = 5180 (0.274 sec)\n",
      "INFO:tensorflow:global_step/sec: 366.626\n",
      "INFO:tensorflow:loss = 0.3683201, step = 5280 (0.280 sec)\n",
      "INFO:tensorflow:global_step/sec: 357.256\n",
      "INFO:tensorflow:loss = 0.2899915, step = 5380 (0.265 sec)\n",
      "INFO:tensorflow:global_step/sec: 359.244\n",
      "INFO:tensorflow:loss = 0.1280297, step = 5480 (0.268 sec)\n",
      "INFO:tensorflow:global_step/sec: 381.8\n",
      "INFO:tensorflow:loss = 0.7579371, step = 5580 (0.274 sec)\n",
      "INFO:tensorflow:global_step/sec: 366.796\n",
      "INFO:tensorflow:loss = 0.20086025, step = 5680 (0.278 sec)\n",
      "INFO:tensorflow:global_step/sec: 363.209\n",
      "INFO:tensorflow:loss = 0.20468965, step = 5780 (0.282 sec)\n",
      "INFO:tensorflow:global_step/sec: 355.186\n",
      "INFO:tensorflow:loss = 0.084839374, step = 5880 (0.284 sec)\n",
      "INFO:tensorflow:global_step/sec: 353.547\n",
      "INFO:tensorflow:loss = 0.7841339, step = 5980 (0.268 sec)\n",
      "INFO:tensorflow:global_step/sec: 355.69\n",
      "INFO:tensorflow:loss = 0.4825125, step = 6080 (0.270 sec)\n",
      "INFO:tensorflow:global_step/sec: 378.825\n",
      "INFO:tensorflow:loss = 0.15031722, step = 6180 (0.278 sec)\n",
      "INFO:tensorflow:global_step/sec: 363.354\n",
      "INFO:tensorflow:loss = 0.09604564, step = 6280 (0.281 sec)\n",
      "INFO:tensorflow:global_step/sec: 359.431\n",
      "INFO:tensorflow:loss = 0.22453651, step = 6380 (0.281 sec)\n",
      "INFO:tensorflow:global_step/sec: 355.953\n",
      "INFO:tensorflow:loss = 0.066752866, step = 6480 (0.287 sec)\n",
      "INFO:tensorflow:global_step/sec: 348.331\n",
      "INFO:tensorflow:loss = 0.13314456, step = 6580 (0.275 sec)\n",
      "INFO:tensorflow:global_step/sec: 346.788\n",
      "INFO:tensorflow:loss = 0.11664696, step = 6680 (0.281 sec)\n",
      "INFO:tensorflow:global_step/sec: 365.916\n",
      "INFO:tensorflow:loss = 0.24780986, step = 6780 (0.280 sec)\n",
      "INFO:tensorflow:global_step/sec: 360.304\n",
      "INFO:tensorflow:loss = 0.16076241, step = 6880 (0.289 sec)\n",
      "INFO:tensorflow:global_step/sec: 347.319\n",
      "INFO:tensorflow:loss = 0.15068403, step = 6980 (0.290 sec)\n",
      "INFO:tensorflow:global_step/sec: 347.161\n",
      "INFO:tensorflow:loss = 0.063347995, step = 7080 (0.288 sec)\n",
      "INFO:tensorflow:global_step/sec: 346.469\n",
      "INFO:tensorflow:loss = 0.17705172, step = 7180 (0.279 sec)\n",
      "INFO:tensorflow:global_step/sec: 342.538\n",
      "INFO:tensorflow:loss = 0.1235522, step = 7280 (0.283 sec)\n",
      "INFO:tensorflow:global_step/sec: 362.863\n",
      "INFO:tensorflow:loss = 0.19375022, step = 7380 (0.283 sec)\n",
      "INFO:tensorflow:global_step/sec: 356.934\n",
      "INFO:tensorflow:loss = 0.09878422, step = 7480 (0.285 sec)\n",
      "INFO:tensorflow:global_step/sec: 350.67\n",
      "INFO:tensorflow:loss = 0.044014893, step = 7580 (0.288 sec)\n",
      "INFO:tensorflow:global_step/sec: 348.951\n",
      "INFO:tensorflow:loss = 0.090123355, step = 7680 (0.292 sec)\n",
      "INFO:tensorflow:global_step/sec: 343.379\n",
      "INFO:tensorflow:loss = 0.1411789, step = 7780 (0.277 sec)\n",
      "INFO:tensorflow:global_step/sec: 343.451\n",
      "INFO:tensorflow:loss = 0.0606481, step = 7880 (0.286 sec)\n",
      "INFO:tensorflow:global_step/sec: 358.731\n",
      "INFO:tensorflow:loss = 0.11701955, step = 7980 (0.293 sec)\n",
      "INFO:tensorflow:global_step/sec: 342.975\n",
      "INFO:tensorflow:loss = 0.2144481, step = 8080 (0.299 sec)\n",
      "INFO:tensorflow:global_step/sec: 337.35\n",
      "INFO:tensorflow:loss = 0.13061918, step = 8180 (0.301 sec)\n",
      "INFO:tensorflow:global_step/sec: 331.849\n",
      "INFO:tensorflow:loss = 0.013081398, step = 8280 (0.298 sec)\n",
      "INFO:tensorflow:global_step/sec: 336.503\n",
      "INFO:tensorflow:loss = 0.027076408, step = 8380 (0.286 sec)\n",
      "INFO:tensorflow:global_step/sec: 332.512\n",
      "INFO:tensorflow:loss = 0.010121934, step = 8480 (0.293 sec)\n",
      "INFO:tensorflow:global_step/sec: 352.257\n",
      "INFO:tensorflow:loss = 0.023727953, step = 8580 (0.294 sec)\n",
      "INFO:tensorflow:global_step/sec: 343.327\n",
      "INFO:tensorflow:loss = 0.13345344, step = 8680 (0.296 sec)\n",
      "INFO:tensorflow:global_step/sec: 339.463\n",
      "INFO:tensorflow:loss = 0.06767905, step = 8780 (0.298 sec)\n",
      "INFO:tensorflow:global_step/sec: 336.43\n",
      "INFO:tensorflow:loss = 0.03239054, step = 8880 (0.299 sec)\n",
      "INFO:tensorflow:global_step/sec: 337.212\n",
      "INFO:tensorflow:loss = 0.03417517, step = 8980 (0.288 sec)\n",
      "INFO:tensorflow:global_step/sec: 329.745\n",
      "INFO:tensorflow:loss = 0.04349177, step = 9080 (0.295 sec)\n",
      "INFO:tensorflow:global_step/sec: 346.215\n",
      "INFO:tensorflow:loss = 0.10747677, step = 9180 (0.297 sec)\n",
      "INFO:tensorflow:global_step/sec: 341.222\n",
      "INFO:tensorflow:loss = 0.08463769, step = 9280 (0.302 sec)\n",
      "INFO:tensorflow:global_step/sec: 333.558\n",
      "INFO:tensorflow:loss = 0.022979608, step = 9380 (0.303 sec)\n",
      "INFO:tensorflow:global_step/sec: 329.165\n",
      "INFO:tensorflow:loss = 0.07760788, step = 9480 (0.310 sec)\n",
      "INFO:tensorflow:global_step/sec: 322.089\n",
      "INFO:tensorflow:loss = 0.038779423, step = 9580 (0.292 sec)\n",
      "INFO:tensorflow:global_step/sec: 329.556\n",
      "INFO:tensorflow:loss = 0.014404967, step = 9680 (0.297 sec)\n",
      "INFO:tensorflow:global_step/sec: 343.326\n",
      "INFO:tensorflow:loss = 0.06990504, step = 9780 (0.305 sec)\n",
      "INFO:tensorflow:global_step/sec: 333.686\n",
      "INFO:tensorflow:loss = 0.036858298, step = 9880 (0.305 sec)\n",
      "INFO:tensorflow:global_step/sec: 326.461\n",
      "INFO:tensorflow:loss = 0.047570646, step = 9980 (0.312 sec)\n",
      "INFO:tensorflow:global_step/sec: 321.895\n",
      "INFO:tensorflow:loss = 0.059428196, step = 10080 (0.309 sec)\n",
      "INFO:tensorflow:global_step/sec: 325.738\n",
      "INFO:tensorflow:loss = 0.05054853, step = 10180 (0.293 sec)\n",
      "INFO:tensorflow:global_step/sec: 327.29\n",
      "INFO:tensorflow:loss = 0.04085783, step = 10280 (0.300 sec)\n",
      "INFO:tensorflow:global_step/sec: 337.825\n",
      "INFO:tensorflow:loss = 0.06833278, step = 10380 (0.309 sec)\n",
      "INFO:tensorflow:global_step/sec: 328.799\n",
      "INFO:tensorflow:loss = 0.03984513, step = 10480 (0.309 sec)\n",
      "INFO:tensorflow:global_step/sec: 325.714\n",
      "INFO:tensorflow:loss = 0.029430978, step = 10580 (0.313 sec)\n",
      "INFO:tensorflow:global_step/sec: 320.448\n",
      "INFO:tensorflow:loss = 0.015103683, step = 10680 (0.310 sec)\n",
      "INFO:tensorflow:global_step/sec: 321.814\n",
      "INFO:tensorflow:loss = 0.055365227, step = 10780 (0.303 sec)\n",
      "INFO:tensorflow:global_step/sec: 315.217\n",
      "INFO:tensorflow:loss = 0.016110064, step = 10880 (0.316 sec)\n",
      "INFO:tensorflow:global_step/sec: 323.304\n",
      "INFO:tensorflow:loss = 0.006240257, step = 10980 (0.315 sec)\n",
      "INFO:tensorflow:global_step/sec: 321.096\n",
      "INFO:tensorflow:loss = 0.007149349, step = 11080 (0.321 sec)\n",
      "INFO:tensorflow:global_step/sec: 314.465\n",
      "INFO:tensorflow:loss = 0.0066786045, step = 11180 (0.312 sec)\n",
      "INFO:tensorflow:global_step/sec: 320.341\n",
      "INFO:tensorflow:loss = 0.025937172, step = 11280 (0.312 sec)\n",
      "INFO:tensorflow:global_step/sec: 321.417\n",
      "INFO:tensorflow:loss = 0.016570274, step = 11380 (0.303 sec)\n",
      "INFO:tensorflow:global_step/sec: 317.392\n",
      "INFO:tensorflow:loss = 0.0033354259, step = 11480 (0.308 sec)\n",
      "INFO:tensorflow:global_step/sec: 330.218\n",
      "INFO:tensorflow:loss = 0.017488046, step = 11580 (0.314 sec)\n",
      "INFO:tensorflow:global_step/sec: 320.864\n",
      "INFO:tensorflow:loss = 0.02159322, step = 11680 (0.322 sec)\n",
      "INFO:tensorflow:global_step/sec: 310.693\n",
      "INFO:tensorflow:loss = 0.020893702, step = 11780 (0.323 sec)\n",
      "INFO:tensorflow:global_step/sec: 310.939\n",
      "INFO:tensorflow:loss = 0.017859623, step = 11880 (0.326 sec)\n",
      "INFO:tensorflow:global_step/sec: 304.814\n",
      "INFO:tensorflow:loss = 0.014102906, step = 11980 (0.310 sec)\n",
      "INFO:tensorflow:global_step/sec: 311.383\n",
      "INFO:tensorflow:loss = 0.014420295, step = 12080 (0.316 sec)\n",
      "INFO:tensorflow:global_step/sec: 323.922\n",
      "INFO:tensorflow:loss = 0.012980898, step = 12180 (0.323 sec)\n",
      "INFO:tensorflow:global_step/sec: 312.002\n",
      "INFO:tensorflow:loss = 0.008047884, step = 12280 (0.324 sec)\n",
      "INFO:tensorflow:global_step/sec: 309.195\n",
      "INFO:tensorflow:loss = 0.005332183, step = 12380 (0.328 sec)\n",
      "INFO:tensorflow:global_step/sec: 307.363\n",
      "INFO:tensorflow:loss = 0.009909308, step = 12480 (0.331 sec)\n",
      "INFO:tensorflow:global_step/sec: 303.166\n",
      "INFO:tensorflow:loss = 0.018593434, step = 12580 (0.310 sec)\n",
      "INFO:tensorflow:global_step/sec: 307.677\n",
      "INFO:tensorflow:loss = 0.009453268, step = 12680 (0.318 sec)\n",
      "INFO:tensorflow:global_step/sec: 323.497\n",
      "INFO:tensorflow:loss = 0.0074377223, step = 12780 (0.317 sec)\n",
      "INFO:tensorflow:global_step/sec: 318.278\n",
      "INFO:tensorflow:loss = 0.0067944657, step = 12880 (0.326 sec)\n",
      "INFO:tensorflow:global_step/sec: 307.95\n",
      "INFO:tensorflow:loss = 0.009621896, step = 12980 (0.332 sec)\n",
      "INFO:tensorflow:global_step/sec: 303.108\n",
      "INFO:tensorflow:loss = 0.007392729, step = 13080 (0.329 sec)\n",
      "INFO:tensorflow:global_step/sec: 303.111\n",
      "INFO:tensorflow:loss = 0.0070271464, step = 13180 (0.317 sec)\n",
      "INFO:tensorflow:global_step/sec: 302.852\n",
      "INFO:tensorflow:loss = 0.01419846, step = 13280 (0.325 sec)\n",
      "INFO:tensorflow:global_step/sec: 311.988\n",
      "INFO:tensorflow:loss = 0.00879844, step = 13380 (0.330 sec)\n",
      "INFO:tensorflow:global_step/sec: 307.168\n",
      "INFO:tensorflow:loss = 0.0035331238, step = 13480 (0.333 sec)\n",
      "INFO:tensorflow:global_step/sec: 301.33\n",
      "INFO:tensorflow:loss = 0.004036055, step = 13580 (0.334 sec)\n",
      "INFO:tensorflow:global_step/sec: 300.952\n",
      "INFO:tensorflow:loss = 0.0021674812, step = 13680 (0.335 sec)\n",
      "INFO:tensorflow:global_step/sec: 298.644\n",
      "INFO:tensorflow:loss = 0.0044945157, step = 13780 (0.318 sec)\n",
      "INFO:tensorflow:global_step/sec: 302.261\n",
      "INFO:tensorflow:loss = 0.008261169, step = 13880 (0.328 sec)\n",
      "INFO:tensorflow:global_step/sec: 310.556\n",
      "INFO:tensorflow:loss = 0.007413184, step = 13980 (0.337 sec)\n",
      "INFO:tensorflow:global_step/sec: 300.33\n",
      "INFO:tensorflow:loss = 0.01038721, step = 14080 (0.331 sec)\n",
      "INFO:tensorflow:global_step/sec: 304.304\n",
      "INFO:tensorflow:loss = 0.0020925598, step = 14180 (0.329 sec)\n",
      "INFO:tensorflow:global_step/sec: 303.857\n",
      "INFO:tensorflow:loss = 0.0072769765, step = 14280 (0.337 sec)\n",
      "INFO:tensorflow:global_step/sec: 297.895\n",
      "INFO:tensorflow:loss = 0.0018916101, step = 14380 (0.326 sec)\n",
      "INFO:tensorflow:global_step/sec: 294.092\n",
      "INFO:tensorflow:loss = 0.0027799625, step = 14480 (0.327 sec)\n",
      "INFO:tensorflow:global_step/sec: 312.093\n",
      "INFO:tensorflow:loss = 0.0037557913, step = 14580 (0.334 sec)\n",
      "INFO:tensorflow:global_step/sec: 301.829\n",
      "INFO:tensorflow:loss = 0.0015468008, step = 14680 (0.334 sec)\n",
      "INFO:tensorflow:global_step/sec: 302.182\n",
      "INFO:tensorflow:loss = 0.0018402252, step = 14780 (0.332 sec)\n",
      "INFO:tensorflow:global_step/sec: 301.447\n",
      "INFO:tensorflow:loss = 0.0063510793, step = 14880 (0.339 sec)\n",
      "INFO:tensorflow:global_step/sec: 294.111\n",
      "INFO:tensorflow:loss = 0.003960237, step = 14980 (0.327 sec)\n",
      "INFO:tensorflow:global_step/sec: 295.082\n",
      "INFO:tensorflow:loss = 0.0021010689, step = 15080 (0.333 sec)\n",
      "INFO:tensorflow:global_step/sec: 306.512\n",
      "INFO:tensorflow:loss = 0.0011556938, step = 15180 (0.338 sec)\n",
      "INFO:tensorflow:global_step/sec: 298.883\n",
      "INFO:tensorflow:loss = 0.0009854774, step = 15280 (0.337 sec)\n",
      "INFO:tensorflow:global_step/sec: 299.258\n",
      "INFO:tensorflow:loss = 0.0059409747, step = 15380 (0.333 sec)\n",
      "INFO:tensorflow:global_step/sec: 299.457\n",
      "INFO:tensorflow:loss = 0.0022082897, step = 15480 (0.338 sec)\n",
      "INFO:tensorflow:global_step/sec: 298.035\n",
      "INFO:tensorflow:loss = 0.0036195924, step = 15580 (0.323 sec)\n",
      "INFO:tensorflow:global_step/sec: 297.116\n",
      "INFO:tensorflow:loss = 0.005268056, step = 15680 (0.332 sec)\n",
      "INFO:tensorflow:global_step/sec: 304.785\n",
      "INFO:tensorflow:loss = 0.0021239321, step = 15780 (0.342 sec)\n",
      "INFO:tensorflow:global_step/sec: 298.12\n",
      "INFO:tensorflow:loss = 0.0127066765, step = 15880 (0.339 sec)\n",
      "INFO:tensorflow:global_step/sec: 295.993\n",
      "INFO:tensorflow:loss = 0.0021492667, step = 15980 (0.341 sec)\n",
      "INFO:tensorflow:global_step/sec: 293.45\n",
      "INFO:tensorflow:loss = 0.003911408, step = 16080 (0.343 sec)\n",
      "INFO:tensorflow:global_step/sec: 291.821\n",
      "INFO:tensorflow:loss = 0.004051245, step = 16180 (0.334 sec)\n",
      "INFO:tensorflow:global_step/sec: 287.44\n",
      "INFO:tensorflow:loss = 0.0049018306, step = 16280 (0.342 sec)\n",
      "INFO:tensorflow:global_step/sec: 297.459\n",
      "INFO:tensorflow:loss = 0.0026472202, step = 16380 (0.345 sec)\n",
      "INFO:tensorflow:global_step/sec: 293.164\n",
      "INFO:tensorflow:loss = 0.0038542324, step = 16480 (0.348 sec)\n",
      "INFO:tensorflow:global_step/sec: 288.779\n",
      "INFO:tensorflow:loss = 0.003773787, step = 16580 (0.346 sec)\n",
      "INFO:tensorflow:global_step/sec: 289.185\n",
      "INFO:tensorflow:loss = 0.0026647656, step = 16680 (0.343 sec)\n",
      "INFO:tensorflow:global_step/sec: 291.876\n",
      "INFO:tensorflow:loss = 0.0024704284, step = 16780 (0.334 sec)\n",
      "INFO:tensorflow:global_step/sec: 288.324\n",
      "INFO:tensorflow:loss = 0.0034512142, step = 16880 (0.347 sec)\n",
      "INFO:tensorflow:global_step/sec: 292.507\n",
      "INFO:tensorflow:loss = 0.0062024607, step = 16980 (0.346 sec)\n",
      "INFO:tensorflow:global_step/sec: 291.147\n",
      "INFO:tensorflow:loss = 0.0022722099, step = 17080 (0.351 sec)\n",
      "INFO:tensorflow:global_step/sec: 287.208\n",
      "INFO:tensorflow:loss = 0.0014444834, step = 17180 (0.352 sec)\n",
      "INFO:tensorflow:global_step/sec: 283.574\n",
      "INFO:tensorflow:loss = 0.0074605285, step = 17280 (0.357 sec)\n",
      "INFO:tensorflow:global_step/sec: 281.604\n",
      "INFO:tensorflow:loss = 0.003752734, step = 17380 (0.339 sec)\n",
      "INFO:tensorflow:global_step/sec: 283.366\n",
      "INFO:tensorflow:loss = 0.0012563546, step = 17480 (0.342 sec)\n",
      "INFO:tensorflow:global_step/sec: 297.925\n",
      "INFO:tensorflow:loss = 0.003298856, step = 17580 (0.347 sec)\n",
      "INFO:tensorflow:global_step/sec: 292.149\n",
      "INFO:tensorflow:loss = 0.0021164892, step = 17680 (0.346 sec)\n",
      "INFO:tensorflow:global_step/sec: 289.272\n",
      "INFO:tensorflow:loss = 0.0027668625, step = 17780 (0.350 sec)\n",
      "INFO:tensorflow:global_step/sec: 286.518\n",
      "INFO:tensorflow:loss = 0.0038928108, step = 17880 (0.356 sec)\n",
      "INFO:tensorflow:global_step/sec: 280.948\n",
      "INFO:tensorflow:loss = 0.00068626396, step = 17980 (0.340 sec)\n",
      "INFO:tensorflow:global_step/sec: 281.988\n",
      "INFO:tensorflow:loss = 0.0011843208, step = 18080 (0.349 sec)\n",
      "INFO:tensorflow:global_step/sec: 292.284\n",
      "INFO:tensorflow:loss = 0.0018866074, step = 18180 (0.351 sec)\n",
      "INFO:tensorflow:global_step/sec: 288.176\n",
      "INFO:tensorflow:loss = 0.0005333081, step = 18280 (0.352 sec)\n",
      "INFO:tensorflow:global_step/sec: 285.166\n",
      "INFO:tensorflow:loss = 0.0005375584, step = 18380 (0.360 sec)\n",
      "INFO:tensorflow:global_step/sec: 279.2\n",
      "INFO:tensorflow:loss = 0.0067465273, step = 18480 (0.355 sec)\n",
      "INFO:tensorflow:global_step/sec: 280.193\n",
      "INFO:tensorflow:loss = 0.0013988668, step = 18580 (0.344 sec)\n",
      "INFO:tensorflow:global_step/sec: 281.697\n",
      "INFO:tensorflow:loss = 0.0014645823, step = 18680 (0.351 sec)\n",
      "INFO:tensorflow:global_step/sec: 288.122\n",
      "INFO:tensorflow:loss = 0.0014383025, step = 18780 (0.360 sec)\n",
      "INFO:tensorflow:global_step/sec: 282.176\n",
      "INFO:tensorflow:loss = 0.0014143193, step = 18880 (0.361 sec)\n",
      "INFO:tensorflow:global_step/sec: 277.737\n",
      "INFO:tensorflow:loss = 0.0013943117, step = 18980 (0.357 sec)\n",
      "INFO:tensorflow:global_step/sec: 281.123\n",
      "INFO:tensorflow:loss = 0.0006448065, step = 19080 (0.357 sec)\n",
      "INFO:tensorflow:global_step/sec: 281.612\n",
      "INFO:tensorflow:loss = 0.0014809513, step = 19180 (0.348 sec)\n",
      "INFO:tensorflow:global_step/sec: 274.105\n",
      "INFO:tensorflow:loss = 0.0008602524, step = 19280 (0.358 sec)\n",
      "INFO:tensorflow:global_step/sec: 285.319\n",
      "INFO:tensorflow:loss = 0.0006964795, step = 19380 (0.363 sec)\n",
      "INFO:tensorflow:global_step/sec: 277.315\n",
      "INFO:tensorflow:loss = 0.00035264163, step = 19480 (0.366 sec)\n",
      "INFO:tensorflow:global_step/sec: 274.599\n",
      "INFO:tensorflow:loss = 0.0010025422, step = 19580 (0.367 sec)\n",
      "INFO:tensorflow:global_step/sec: 273.115\n",
      "INFO:tensorflow:loss = 0.0007096651, step = 19680 (0.362 sec)\n",
      "INFO:tensorflow:global_step/sec: 276.323\n",
      "INFO:tensorflow:loss = 0.0013329595, step = 19780 (0.351 sec)\n",
      "INFO:tensorflow:global_step/sec: 274.789\n",
      "INFO:tensorflow:loss = 0.0008460893, step = 19880 (0.357 sec)\n",
      "INFO:tensorflow:global_step/sec: 283.638\n",
      "INFO:tensorflow:loss = 0.0011283578, step = 19980 (0.368 sec)\n",
      "INFO:tensorflow:global_step/sec: 275.094\n",
      "INFO:tensorflow:loss = 0.00089822686, step = 20080 (0.365 sec)\n",
      "INFO:tensorflow:global_step/sec: 275.392\n",
      "INFO:tensorflow:loss = 0.0014473142, step = 20180 (0.364 sec)\n",
      "INFO:tensorflow:global_step/sec: 276.458\n",
      "INFO:tensorflow:loss = 0.0008915104, step = 20280 (0.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 268.018\n",
      "INFO:tensorflow:loss = 0.0004781757, step = 20380 (0.353 sec)\n",
      "INFO:tensorflow:global_step/sec: 272.515\n",
      "INFO:tensorflow:loss = 0.0004186085, step = 20480 (0.363 sec)\n",
      "INFO:tensorflow:global_step/sec: 280.449\n",
      "INFO:tensorflow:loss = 0.0008953349, step = 20580 (0.364 sec)\n",
      "INFO:tensorflow:global_step/sec: 278.265\n",
      "INFO:tensorflow:loss = 0.0015090622, step = 20680 (0.371 sec)\n",
      "INFO:tensorflow:global_step/sec: 270.082\n",
      "INFO:tensorflow:loss = 0.0010438098, step = 20780 (0.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 267.97\n",
      "INFO:tensorflow:loss = 0.00050447625, step = 20880 (0.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 267.517\n",
      "INFO:tensorflow:loss = 0.00037436924, step = 20980 (0.364 sec)\n",
      "INFO:tensorflow:global_step/sec: 262.304\n",
      "INFO:tensorflow:loss = 0.0005487846, step = 21080 (0.371 sec)\n",
      "INFO:tensorflow:global_step/sec: 276.63\n",
      "INFO:tensorflow:loss = 0.0012135495, step = 21180 (0.372 sec)\n",
      "INFO:tensorflow:global_step/sec: 271.146\n",
      "INFO:tensorflow:loss = 0.00050225714, step = 21280 (0.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 266.848\n",
      "INFO:tensorflow:loss = 0.0005835245, step = 21380 (0.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 266.179\n",
      "INFO:tensorflow:loss = 0.0004619556, step = 21480 (0.375 sec)\n",
      "INFO:tensorflow:global_step/sec: 266.419\n",
      "INFO:tensorflow:loss = 0.00033856914, step = 21580 (0.363 sec)\n",
      "INFO:tensorflow:global_step/sec: 265.468\n",
      "INFO:tensorflow:loss = 0.0008394742, step = 21680 (0.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 272.316\n",
      "INFO:tensorflow:loss = 0.00030781276, step = 21780 (0.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 270.614\n",
      "INFO:tensorflow:loss = 0.00032267775, step = 21880 (0.375 sec)\n",
      "INFO:tensorflow:global_step/sec: 266.912\n",
      "INFO:tensorflow:loss = 0.00024132222, step = 21980 (0.378 sec)\n",
      "INFO:tensorflow:global_step/sec: 265.246\n",
      "INFO:tensorflow:loss = 0.00028675678, step = 22080 (0.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 266.509\n",
      "INFO:tensorflow:loss = 0.0009781871, step = 22180 (0.365 sec)\n",
      "INFO:tensorflow:global_step/sec: 263.828\n",
      "INFO:tensorflow:loss = 0.0010109144, step = 22280 (0.370 sec)\n",
      "INFO:tensorflow:global_step/sec: 274.649\n",
      "INFO:tensorflow:loss = 0.00025149249, step = 22380 (0.378 sec)\n",
      "INFO:tensorflow:global_step/sec: 267.999\n",
      "INFO:tensorflow:loss = 0.00020908765, step = 22480 (0.378 sec)\n",
      "INFO:tensorflow:global_step/sec: 265.322\n",
      "INFO:tensorflow:loss = 0.0004320807, step = 22580 (0.384 sec)\n",
      "INFO:tensorflow:global_step/sec: 260.926\n",
      "INFO:tensorflow:loss = 0.0002488165, step = 22680 (0.385 sec)\n",
      "INFO:tensorflow:global_step/sec: 259.177\n",
      "INFO:tensorflow:loss = 0.0004015111, step = 22780 (0.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 256.529\n",
      "INFO:tensorflow:loss = 0.00037404272, step = 22880 (0.385 sec)\n",
      "INFO:tensorflow:global_step/sec: 264.999\n",
      "INFO:tensorflow:loss = 0.00039812157, step = 22980 (0.387 sec)\n",
      "INFO:tensorflow:global_step/sec: 260.906\n",
      "INFO:tensorflow:loss = 0.0005162174, step = 23080 (0.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 259.663\n",
      "INFO:tensorflow:loss = 0.00032000744, step = 23180 (0.387 sec)\n",
      "INFO:tensorflow:global_step/sec: 258.701\n",
      "INFO:tensorflow:loss = 0.00025557584, step = 23280 (0.391 sec)\n",
      "INFO:tensorflow:global_step/sec: 255.811\n",
      "INFO:tensorflow:loss = 0.00018507428, step = 23380 (0.372 sec)\n",
      "INFO:tensorflow:global_step/sec: 260.467\n",
      "INFO:tensorflow:loss = 0.00010121861, step = 23480 (0.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 271.391\n",
      "INFO:tensorflow:loss = 0.00043678225, step = 23580 (0.381 sec)\n",
      "INFO:tensorflow:global_step/sec: 264.417\n",
      "INFO:tensorflow:loss = 0.0002813889, step = 23680 (0.391 sec)\n",
      "INFO:tensorflow:global_step/sec: 257.244\n",
      "INFO:tensorflow:loss = 9.453914e-05, step = 23780 (0.393 sec)\n",
      "INFO:tensorflow:global_step/sec: 254.353\n",
      "INFO:tensorflow:loss = 0.0002390909, step = 23880 (0.390 sec)\n",
      "INFO:tensorflow:global_step/sec: 240.95\n",
      "INFO:tensorflow:loss = 0.0008116873, step = 23980 (0.442 sec)\n",
      "INFO:tensorflow:global_step/sec: 229.283\n",
      "INFO:tensorflow:Saving checkpoints for 24000 into /tmp/tmpbzo1p2wi/model.ckpt.\n",
      "WARNING:tensorflow:Issue encountered when serializing resources.\n",
      "Type is unsupported, or the types of the items don't match field type in CollectionDef. Note this is a warning and probably safe to ignore.\n",
      "'_Resource' object has no attribute 'name'\n",
      "INFO:tensorflow:Loss for final step: 0.00040755837.\n",
      "INFO:tensorflow:Calling model_fn.\n",
      "INFO:tensorflow:Done calling model_fn.\n",
      "INFO:tensorflow:Starting evaluation at 2019-11-03T11:19:05Z\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/tmpbzo1p2wi/model.ckpt-24000\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n",
      "INFO:tensorflow:Finished evaluation at 2019-11-03-11:19:05\n",
      "INFO:tensorflow:Saving dict for global step 24000: average_loss = 12.3817, global_step = 24000, label/mean = 23.611393, loss = 12.283247, prediction/mean = 22.392288\n",
      "WARNING:tensorflow:Issue encountered when serializing resources.\n",
      "Type is unsupported, or the types of the items don't match field type in CollectionDef. Note this is a warning and probably safe to ignore.\n",
      "'_Resource' object has no attribute 'name'\n",
      "INFO:tensorflow:Saving 'checkpoint_path' summary for global step 24000: /tmp/tmpbzo1p2wi/model.ckpt-24000\n",
      "{'average_loss': 12.3817, 'label/mean': 23.611393, 'loss': 12.283247, 'prediction/mean': 22.392288, 'global_step': 24000}\n",
      "Average-Loss 12.3817\n"
     ]
    }
   ],
   "source": [
    "boosted_tree = tf.estimator.BoostedTreesRegressor(\n",
    "    feature_columns=all_feature_columns,\n",
    "    n_batches_per_layer=20,\n",
    "    n_trees=200)\n",
    "\n",
    "boosted_tree.train(\n",
    "    input_fn=lambda:train_input_fn(df_train_norm, batch_size=BATCH_SIZE))\n",
    "\n",
    "eval_results = boosted_tree.evaluate(\n",
    "    input_fn=lambda:eval_input_fn(df_test_norm, batch_size=8))\n",
    "\n",
    "print(eval_results)\n",
    "\n",
    "print('Average-Loss {:.4f}'.format(eval_results['average_loss']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "Readers may ignore the next cell."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[NbConvertApp] Converting notebook ch14_part2.ipynb to script\n",
      "[NbConvertApp] Writing 6364 bytes to ch14_part2.py\n"
     ]
    }
   ],
   "source": [
    "! python ../.convert_notebook_to_script.py --input ch14_part2.ipynb --output ch14_part2.py"
   ]
  }
 ],
 "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}