Train-Copy1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\nIn this notebook we train our model. The thickness and number of hidden layers were tuned manually, \\nbut otherwise nothing else has. Once could e.g. change the activation function from tanh to relu.\\n\\nThe FunctionScalar package (https://github.com/weissercn/FunctionScaler/) is used to preprocess the data. \\nInstalling this package can be done using pip (pip install FunctionScaler)\\n'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''\n",
    "In this notebook we train our model. The thickness and number of hidden layers were tuned manually, \n",
    "but otherwise nothing else has. Once could e.g. change the activation function from tanh to relu.\n",
    "\n",
    "The FunctionScalar package (https://github.com/weissercn/FunctionScaler/) is used to preprocess the data. \n",
    "Installing this package can be done using pip (pip install FunctionScaler)\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Importing necessary libraries\n",
    "import math\n",
    "import pickle\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from tqdm import tqdm\n",
    "# from __future__ import print_function\n",
    "# from sklearn import metrics\n",
    "from sklearn.preprocessing import StandardScaler, MaxAbsScaler\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "\n",
    "plt.rc('text', usetex=False)\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/anaconda3/lib/python3.6/site-packages/h5py/__init__.py:34: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "\n",
    "config = tf.ConfigProto()\n",
    "config.gpu_options.per_process_gpu_memory_fraction = 0.2\n",
    "config.gpu_options.visible_device_list = '1'\n",
    "\n",
    "import keras\n",
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Input, Dense, Dropout\n",
    "from keras.callbacks import EarlyStopping, TensorBoard\n",
    "from keras import backend as K\n",
    "from keras.backend.tensorflow_backend import set_session\n",
    "\n",
    "tf.set_random_seed(1)\n",
    "sess = tf.Session(config=config)\n",
    "set_session(sess)\n",
    "\n",
    "from Autoencoder_Utils import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Считаем подготовленные для обучения автоэнкодера данные"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Loading the data\n",
    "TYPE = \"\" # 100% of the data. To determine the real performance of the cross check\n",
    "# TYPE = \"_10perc\" # 10% of the data. Used for everything except to determine the real performance of the cross check\n",
    "# TYPE = \"_1perc\" # 1% of the data. For small tests.\n",
    "# TYPE = \"_100\" # 100 samples of the data. For very rudimentary tests\n",
    "\n",
    "df = pd.read_csv('Data/Df_example{}.csv'.format(TYPE))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "f    454948\n",
      "e    450775\n",
      "a    438012\n",
      "c    408195\n",
      "d    377725\n",
      "b    358002\n",
      "Name: pid, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# How many samples of each particle do we have\n",
    "print(df['pid'].value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pids_no = df['pid'].value_counts().to_dict()\n",
    "n_samples = np.shape(df.as_matrix())[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "18 variables to compress\n"
     ]
    }
   ],
   "source": [
    "# Dividing the variables into compressable, auxiliary and simulation truth variables not seen in real datasets\n",
    "vars_list = list(df)\n",
    "vars_list_input, vars_list_aux, vars_list_MC = [], [], []\n",
    "\n",
    "for v in vars_list:\n",
    "    if v == 'pid': \n",
    "        vars_list_MC.append(v)\n",
    "    elif 'aux' in v: \n",
    "        vars_list_aux.append(v)\n",
    "    else: \n",
    "        vars_list_input.append(v)\n",
    "\n",
    "print(\"{} variables to compress\".format(len(vars_list_input)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 11.8 s, sys: 4.88 s, total: 16.7 s\n",
      "Wall time: 10.7 s\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjIAAAVzCAYAAADQb8ceAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/gcmkSML1vxvNgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f60906b1710>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%time\n",
    "# Display histograms of our features\n",
    "print_features_histograms(df[vars_list_input + vars_list_aux], None, save_filename=\"./img/original_distribs.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Разобъем индексы на train и test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# train test split\n",
    "train_size = 0.7\n",
    "test_size = 0.3\n",
    "idx_train, idx_test = train_test_split(np.arange(df.shape[0], dtype='int32'), random_state=5,\n",
    "                                       stratify=df['pid'], train_size=train_size, test_size=test_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((2487657, 27), (1741359,), (746298,))"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape, idx_train.shape, idx_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Переведем значения признаков в отрезок $[-1, 1]$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ранее помимо масштабирования признаков меняли и их распределение на равномерное, но это не имеет смысла, так как энкодер и декодер восстанавливают распределения $P(Z|X)$ и $Q(X|Z)$ соответственно. А по ним уже восстанавливается распределение $P(X)$ каким бы оно не было."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Так то оно правильно, но, может быть, автоэнкодеру проще восстановить равномерное распределение, чем произвольное?!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df_scaled = df.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 7.71 s, sys: 504 ms, total: 8.22 s\n",
      "Wall time: 8.15 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "scalers = {}\n",
    "\n",
    "for var in (vars_list_input + vars_list_aux):\n",
    "    # normalize with StandardScaler\n",
    "    std_scaler = StandardScaler()\n",
    "    std_scaler.fit(df[var].iloc[idx_train].values.reshape(-1, 1))\n",
    "    df_scaled[var] = std_scaler.transform(df[var].values.reshape(-1, 1))\n",
    "    \n",
    "    max_scaler = MaxAbsScaler()\n",
    "    max_scaler.fit(df_scaled[var].iloc[idx_train].values.reshape(-1, 1))\n",
    "    df_scaled[var] = max_scaler.transform(df_scaled[var].values.reshape(-1, 1))\n",
    "    scalers[var] = {\n",
    "        'std': std_scaler,\n",
    "        'max': max_scaler\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 12 s, sys: 4.83 s, total: 16.9 s\n",
      "Wall time: 10.7 s\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAh0AAAVzCAYAAABuOZ9DAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nDkQjSQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f614c6e5ba8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%time\n",
    "# Display histograms of our features\n",
    "print_features_histograms(df_scaled[vars_list_input + vars_list_aux], None, save_filename=\"./img/scaled_distribs.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Зададим архитектуру автоэнкодера"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def create_autoencoder_aux(n_features, encoding_dim, n_aux_features=5, p_drop=0.5, n_layers=3, thickness=2):\n",
    "    # Энкодер\n",
    "    inputs = Input(shape=(n_features, ), name='main_input')\n",
    "    aux_inputs = Input(shape=(n_aux_features, ), name='aux_inputs')\n",
    "    \n",
    "    \n",
    "    x = keras.layers.concatenate([inputs, aux_inputs])\n",
    "\n",
    "    for i in range(n_layers - 1):\n",
    "        x = Dense(thickness * n_features, activation='tanh')(x)\n",
    "        x = keras.layers.concatenate([x, aux_inputs])\n",
    "\n",
    "#     x = Dropout(p_drop)(x)\n",
    "    \n",
    "    x = Dense(thickness * encoding_dim, activation='tanh')(x)\n",
    "    x = keras.layers.concatenate([x, aux_inputs])\n",
    "#     x = Dropout(p_drop)(x)\n",
    "    encoded = Dense(encoding_dim, activation='tanh', name='encoded')(x)\n",
    "\n",
    "    \n",
    "    # Декодер\n",
    "    input_encoded = Input(shape=(encoding_dim, ))\n",
    "    \n",
    "    x = keras.layers.concatenate([input_encoded, aux_inputs])\n",
    "    x = Dense(thickness * encoding_dim, activation='tanh')(x)\n",
    "#     x = Dropout(p_drop)(x)\n",
    "    \n",
    "    for i in range(n_layers - 1):\n",
    "        x = keras.layers.concatenate([x, aux_inputs])\n",
    "        x = Dense(thickness * n_features, activation='tanh')(x)\n",
    "        \n",
    "#     x = Dropout(p_drop)(x)\n",
    "    decoded = Dense(n_features, activation='tanh')(x)\n",
    "    \n",
    "    \n",
    "    # Модели\n",
    "    encoder = Model([inputs, aux_inputs], encoded, name=\"encoder\")\n",
    "    decoder = Model([input_encoded, aux_inputs], decoded, name=\"decoder\")\n",
    "    autoencoder = Model(\n",
    "        [inputs, aux_inputs], decoder([encoder([inputs, aux_inputs]), aux_inputs]), name=\"autoencoder\"\n",
    "    )\n",
    "\n",
    "    optimizer_adam = optimizers.Adam(lr=0.001)\n",
    "    autoencoder.compile(loss='mse', optimizer=optimizer_adam)\n",
    "    \n",
    "    return autoencoder, encoder, decoder"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Обучим автоэнкодер"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "l_encoding_dim = [1, 3, 6, 9, 12, 15, 18]\n",
    "l_score = []\n",
    "val_size = 0.1\n",
    "\n",
    "train = df_scaled.iloc[idx_train]\n",
    "test = df_scaled.iloc[idx_test]\n",
    "\n",
    "train_input = train[vars_list_input].values\n",
    "train_aux   = train[vars_list_aux].values\n",
    "train_MC    = train[vars_list_MC].values\n",
    "\n",
    "test_input = test[vars_list_input].values\n",
    "test_aux   = test[vars_list_aux].values\n",
    "test_MC    = test[vars_list_MC].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  0%|          | 0/7 [00:00<?, ?it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "encoding_dim :  1\n",
      "Train on 1567223 samples, validate on 174136 samples\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "1567223/1567223 [==============================] - 129s 82us/step - loss: 0.0221 - val_loss: 0.0169\n",
      "Epoch 2/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 0.0158 - val_loss: 0.0143\n",
      "Epoch 3/30\n",
      "1567223/1567223 [==============================] - 128s 82us/step - loss: 0.0136 - val_loss: 0.0132\n",
      "Epoch 4/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 0.0124 - val_loss: 0.0115\n",
      "Epoch 5/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 0.0134 - val_loss: 0.0136\n",
      "Epoch 6/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 0.0132 - val_loss: 0.0127\n",
      "Epoch 7/30\n",
      "1567223/1567223 [==============================] - 123s 79us/step - loss: 0.0117 - val_loss: 0.0116\n",
      "Epoch 00007: early stopping\n",
      "746298/746298 [==============================] - 13s 17us/step\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 14%|█▍        | 1/7 [15:43<1:34:19, 943.25s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\n",
      "encoding_dim :  3\n",
      "Train on 1567223 samples, validate on 174136 samples\n",
      "Epoch 1/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 0.0070 - val_loss: 0.0048\n",
      "Epoch 2/30\n",
      "1567223/1567223 [==============================] - 117s 75us/step - loss: 0.0044 - val_loss: 0.0041\n",
      "Epoch 3/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 0.0039 - val_loss: 0.0038\n",
      "Epoch 4/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 0.0036 - val_loss: 0.0034\n",
      "Epoch 5/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 0.0033 - val_loss: 0.0033\n",
      "Epoch 6/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 0.0032 - val_loss: 0.0031\n",
      "Epoch 7/30\n",
      "1567223/1567223 [==============================] - 120s 76us/step - loss: 0.0031 - val_loss: 0.0031\n",
      "Epoch 8/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 0.0030 - val_loss: 0.0030\n",
      "Epoch 9/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 0.0030 - val_loss: 0.0029\n",
      "Epoch 10/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 0.0029 - val_loss: 0.0029\n",
      "Epoch 11/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 0.0029 - val_loss: 0.0028\n",
      "Epoch 12/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 0.0028 - val_loss: 0.0028\n",
      "Epoch 13/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 0.0028 - val_loss: 0.0028\n",
      "Epoch 14/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 0.0029 - val_loss: 0.0028\n",
      "Epoch 00014: early stopping\n",
      "746298/746298 [==============================] - 14s 18us/step\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 29%|██▊       | 2/7 [45:07<1:39:07, 1189.56s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\n",
      "encoding_dim :  6\n",
      "Train on 1567223 samples, validate on 174136 samples\n",
      "Epoch 1/30\n",
      "1567223/1567223 [==============================] - 123s 78us/step - loss: 0.0018 - val_loss: 9.6268e-04\n",
      "Epoch 2/30\n",
      "1567223/1567223 [==============================] - 116s 74us/step - loss: 8.4446e-04 - val_loss: 7.6640e-04\n",
      "Epoch 3/30\n",
      "1567223/1567223 [==============================] - 123s 78us/step - loss: 7.2706e-04 - val_loss: 6.6164e-04\n",
      "Epoch 4/30\n",
      "1567223/1567223 [==============================] - 117s 75us/step - loss: 6.6541e-04 - val_loss: 6.3577e-04\n",
      "Epoch 5/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 6.2535e-04 - val_loss: 6.0046e-04\n",
      "Epoch 6/30\n",
      "1567223/1567223 [==============================] - 115s 74us/step - loss: 5.9583e-04 - val_loss: 5.7348e-04\n",
      "Epoch 7/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 5.7216e-04 - val_loss: 5.5951e-04\n",
      "Epoch 8/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 5.5115e-04 - val_loss: 5.5247e-04\n",
      "Epoch 9/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 5.3508e-04 - val_loss: 5.3040e-04\n",
      "Epoch 10/30\n",
      "1567223/1567223 [==============================] - 118s 75us/step - loss: 5.2016e-04 - val_loss: 5.3187e-04\n",
      "Epoch 11/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 5.0825e-04 - val_loss: 4.9066e-04\n",
      "Epoch 12/30\n",
      "1567223/1567223 [==============================] - 118s 75us/step - loss: 4.9699e-04 - val_loss: 4.8037e-04\n",
      "Epoch 13/30\n",
      "1567223/1567223 [==============================] - 123s 78us/step - loss: 4.8806e-04 - val_loss: 4.7633e-04\n",
      "Epoch 14/30\n",
      "1567223/1567223 [==============================] - 118s 75us/step - loss: 4.8029e-04 - val_loss: 4.7212e-04\n",
      "Epoch 15/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 4.7362e-04 - val_loss: 4.8861e-04\n",
      "Epoch 16/30\n",
      "1567223/1567223 [==============================] - 118s 75us/step - loss: 4.6748e-04 - val_loss: 4.8383e-04\n",
      "Epoch 17/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 4.6165e-04 - val_loss: 4.4053e-04\n",
      "Epoch 18/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 4.5689e-04 - val_loss: 4.5002e-04\n",
      "Epoch 19/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 4.5161e-04 - val_loss: 4.3322e-04\n",
      "Epoch 20/30\n",
      "1567223/1567223 [==============================] - 117s 75us/step - loss: 4.4768e-04 - val_loss: 4.4994e-04\n",
      "Epoch 21/30\n",
      "1567223/1567223 [==============================] - 120s 76us/step - loss: 4.4334e-04 - val_loss: 4.5359e-04\n",
      "Epoch 22/30\n",
      "1567223/1567223 [==============================] - 117s 75us/step - loss: 4.3989e-04 - val_loss: 4.5235e-04\n",
      "Epoch 00022: early stopping\n",
      "746298/746298 [==============================] - 14s 18us/step\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 43%|████▎     | 3/7 [1:30:16<1:49:41, 1645.27s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\n",
      "encoding_dim :  9\n",
      "Train on 1567223 samples, validate on 174136 samples\n",
      "Epoch 1/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 8.7790e-04 - val_loss: 4.0873e-04\n",
      "Epoch 2/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 3.3274e-04 - val_loss: 2.6347e-04\n",
      "Epoch 3/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 2.4896e-04 - val_loss: 2.2469e-04\n",
      "Epoch 4/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 2.1355e-04 - val_loss: 2.0325e-04\n",
      "Epoch 5/30\n",
      "1567223/1567223 [==============================] - 125s 80us/step - loss: 1.8948e-04 - val_loss: 1.8257e-04\n",
      "Epoch 6/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 1.7383e-04 - val_loss: 1.5989e-04\n",
      "Epoch 7/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 1.6244e-04 - val_loss: 1.5645e-04\n",
      "Epoch 8/30\n",
      "1567223/1567223 [==============================] - 120s 76us/step - loss: 1.5267e-04 - val_loss: 1.6016e-04\n",
      "Epoch 9/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 1.4518e-04 - val_loss: 1.4503e-04\n",
      "Epoch 10/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 1.3880e-04 - val_loss: 1.4860e-04\n",
      "Epoch 11/30\n",
      "1567223/1567223 [==============================] - 124s 79us/step - loss: 1.3339e-04 - val_loss: 1.2690e-04\n",
      "Epoch 12/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 1.2872e-04 - val_loss: 1.1685e-04\n",
      "Epoch 13/30\n",
      "1567223/1567223 [==============================] - 123s 79us/step - loss: 1.2500e-04 - val_loss: 1.2857e-04\n",
      "Epoch 14/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 1.2161e-04 - val_loss: 1.2137e-04\n",
      "Epoch 15/30\n",
      "1567223/1567223 [==============================] - 123s 79us/step - loss: 1.1886e-04 - val_loss: 1.1768e-04\n",
      "Epoch 00015: early stopping\n",
      "746298/746298 [==============================] - 14s 18us/step\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 57%|█████▋    | 4/7 [2:02:05<1:26:13, 1724.49s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\n",
      "encoding_dim :  12\n",
      "Train on 1567223 samples, validate on 174136 samples\n",
      "Epoch 1/30\n",
      "1567223/1567223 [==============================] - 123s 78us/step - loss: 5.9001e-04 - val_loss: 2.9048e-04\n",
      "Epoch 2/30\n",
      "1567223/1567223 [==============================] - 120s 76us/step - loss: 2.3322e-04 - val_loss: 1.9539e-04\n",
      "Epoch 3/30\n",
      "1567223/1567223 [==============================] - 125s 80us/step - loss: 1.6600e-04 - val_loss: 1.4524e-04\n",
      "Epoch 4/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 1.3220e-04 - val_loss: 1.4754e-04\n",
      "Epoch 5/30\n",
      "1567223/1567223 [==============================] - 125s 79us/step - loss: 1.1155e-04 - val_loss: 1.0321e-04\n",
      "Epoch 6/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 9.0810e-05 - val_loss: 7.8256e-05\n",
      "Epoch 7/30\n",
      "1567223/1567223 [==============================] - 125s 80us/step - loss: 8.1059e-05 - val_loss: 7.6180e-05\n",
      "Epoch 8/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 7.3667e-05 - val_loss: 6.5856e-05\n",
      "Epoch 9/30\n",
      "1567223/1567223 [==============================] - 123s 78us/step - loss: 6.8686e-05 - val_loss: 5.8677e-05\n",
      "Epoch 10/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 6.4521e-05 - val_loss: 6.9818e-05\n",
      "Epoch 11/30\n",
      "1567223/1567223 [==============================] - 124s 79us/step - loss: 6.1803e-05 - val_loss: 5.2915e-05\n",
      "Epoch 12/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 5.9262e-05 - val_loss: 4.8251e-05\n",
      "Epoch 13/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 5.7397e-05 - val_loss: 6.0130e-05\n",
      "Epoch 14/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 5.4708e-05 - val_loss: 4.4761e-05\n",
      "Epoch 15/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 5.0286e-05 - val_loss: 4.3384e-05\n",
      "Epoch 16/30\n",
      "1567223/1567223 [==============================] - 120s 76us/step - loss: 4.7404e-05 - val_loss: 4.6801e-05\n",
      "Epoch 17/30\n",
      "1567223/1567223 [==============================] - 124s 79us/step - loss: 4.5569e-05 - val_loss: 3.8993e-05\n",
      "Epoch 18/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 4.4107e-05 - val_loss: 4.8819e-05\n",
      "Epoch 19/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 4.2361e-05 - val_loss: 3.7761e-05\n",
      "Epoch 20/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 4.1271e-05 - val_loss: 4.3970e-05\n",
      "Epoch 21/30\n",
      "1567223/1567223 [==============================] - 120s 76us/step - loss: 4.0390e-05 - val_loss: 3.6329e-05\n",
      "Epoch 22/30\n",
      "1567223/1567223 [==============================] - 121s 78us/step - loss: 3.9658e-05 - val_loss: 4.2308e-05\n",
      "Epoch 23/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 3.9036e-05 - val_loss: 4.1332e-05\n",
      "Epoch 24/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 3.8271e-05 - val_loss: 3.3523e-05\n",
      "Epoch 25/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 3.6794e-05 - val_loss: 3.4041e-05\n",
      "Epoch 26/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 3.6759e-05 - val_loss: 3.6905e-05\n",
      "Epoch 27/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 3.5928e-05 - val_loss: 3.3340e-05\n",
      "Epoch 28/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 3.5262e-05 - val_loss: 3.2556e-05\n",
      "Epoch 29/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 3.5016e-05 - val_loss: 3.3641e-05\n",
      "Epoch 30/30\n",
      "1567223/1567223 [==============================] - 123s 78us/step - loss: 3.4802e-05 - val_loss: 4.0294e-05\n",
      "746298/746298 [==============================] - 14s 19us/step\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 71%|███████▏  | 5/7 [3:04:18<1:17:34, 2327.16s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\n",
      "encoding_dim :  15\n",
      "Train on 1567223 samples, validate on 174136 samples\n",
      "Epoch 1/30\n",
      "1567223/1567223 [==============================] - 126s 80us/step - loss: 4.2642e-04 - val_loss: 1.4054e-04\n",
      "Epoch 2/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 1.1775e-04 - val_loss: 8.1189e-05\n",
      "Epoch 3/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 8.3561e-05 - val_loss: 9.2177e-05\n",
      "Epoch 4/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 6.5774e-05 - val_loss: 6.3916e-05\n",
      "Epoch 5/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 5.5614e-05 - val_loss: 4.8949e-05\n",
      "Epoch 6/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 4.8269e-05 - val_loss: 3.7174e-05\n",
      "Epoch 7/30\n",
      "1567223/1567223 [==============================] - 123s 78us/step - loss: 4.3484e-05 - val_loss: 3.6961e-05\n",
      "Epoch 8/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 4.0126e-05 - val_loss: 4.5578e-05\n",
      "Epoch 9/30\n",
      "1567223/1567223 [==============================] - 122s 78us/step - loss: 3.7807e-05 - val_loss: 3.2545e-05\n",
      "Epoch 10/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 3.5604e-05 - val_loss: 3.7725e-05\n",
      "Epoch 11/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 3.4178e-05 - val_loss: 3.6192e-05\n",
      "Epoch 12/30\n",
      "1567223/1567223 [==============================] - 120s 76us/step - loss: 3.2941e-05 - val_loss: 3.2897e-05\n",
      "Epoch 00012: early stopping\n",
      "746298/746298 [==============================] - 14s 19us/step\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 86%|████████▌ | 6/7 [3:30:10<34:54, 2094.51s/it]  "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\n",
      "encoding_dim :  18\n",
      "Train on 1567223 samples, validate on 174136 samples\n",
      "Epoch 1/30\n",
      "1567223/1567223 [==============================] - 120s 77us/step - loss: 4.1082e-04 - val_loss: 1.7079e-04\n",
      "Epoch 2/30\n",
      "1567223/1567223 [==============================] - 123s 78us/step - loss: 1.1284e-04 - val_loss: 8.2098e-05\n",
      "Epoch 3/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 7.1011e-05 - val_loss: 5.4903e-05\n",
      "Epoch 4/30\n",
      "1567223/1567223 [==============================] - 124s 79us/step - loss: 5.6552e-05 - val_loss: 5.0483e-05\n",
      "Epoch 5/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 4.8488e-05 - val_loss: 4.9606e-05\n",
      "Epoch 6/30\n",
      "1567223/1567223 [==============================] - 124s 79us/step - loss: 4.2988e-05 - val_loss: 5.0854e-05\n",
      "Epoch 7/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 3.9087e-05 - val_loss: 6.0549e-05\n",
      "Epoch 8/30\n",
      "1567223/1567223 [==============================] - 123s 79us/step - loss: 3.6064e-05 - val_loss: 4.8365e-05\n",
      "Epoch 9/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 3.3008e-05 - val_loss: 2.9134e-05\n",
      "Epoch 10/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 3.0969e-05 - val_loss: 2.9388e-05\n",
      "Epoch 11/30\n",
      "1567223/1567223 [==============================] - 119s 76us/step - loss: 2.9689e-05 - val_loss: 2.4983e-05\n",
      "Epoch 12/30\n",
      "1567223/1567223 [==============================] - 124s 79us/step - loss: 2.8709e-05 - val_loss: 5.0731e-05\n",
      "Epoch 13/30\n",
      "1567223/1567223 [==============================] - 121s 77us/step - loss: 2.6891e-05 - val_loss: 2.6051e-05\n",
      "Epoch 14/30\n",
      "1567223/1567223 [==============================] - 123s 79us/step - loss: 2.5768e-05 - val_loss: 2.5433e-05\n",
      "Epoch 00014: early stopping\n",
      "746298/746298 [==============================] - 14s 18us/step\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      "100%|██████████| 7/7 [4:00:06<00:00, 2004.92s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\n",
      "CPU times: user 4h 48min 8s, sys: 1h 3min 31s, total: 5h 51min 40s\n",
      "Wall time: 4h 6s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAD8CAYAAAB3u9PLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XuYHVWd7vHvu/fuTtKdhCQSQ25yMwIBlUsLEQZBBO2A\nEBQPQ44KMjiRQRx1PHPM8ZEDyKiIoDMoEwYcNDAMHLyMREBijAjCEEwHAQkQaAKYhNwgkKRz6cve\nv/NHVXd2Oh16d7o7fdnv53nqqapVa+1aqyup365VVWsrIjAzM8v0dQXMzKx/cEAwMzPAAcHMzFIO\nCGZmBjggmJlZygHBzMyAEgOCpFpJyyTVS5rdwXZJuj7d/pSko9ttz0r6k6R7itLGSFog6YV0Prr7\nzTEzsz3VaUCQlAVuAKYDU4GZkqa2yzYdmJJOs4A57bZ/EXi2XdpsYGFETAEWputmZtZHSrlCOBao\nj4jlEdEE3AnMaJdnBnBrJBYBoySNB5A0CTgD+FEHZeamy3OBs/ewDWZm1gNyJeSZCKwoWl8JHFdC\nnonAauCfgf8NjGhXZlxErE6X1wDjOtq5pFkkVx1UV1cfc+ihh5ZQZTMza7VkyZLXImJsZ/lKCQh7\nTNJHgXURsUTSybvLFxEhqcMxNCLiJuAmgJqamqirq+uVupqZDVaSXiklXyldRquAyUXrk9K0UvKc\nAJwl6WWSrqZTJP1HmmdtUbfSeGBdKRU2M7PeUUpAWAxMkXSgpErgPGBeuzzzgPPTp42mARsjYnVE\n/J+ImBQRB6TlfhcRnyoqc0G6fAFwd3cbY2Zme67TLqOIaJF0KTAfyAK3RMRSSRen228E7gNOB+qB\nrcCFJez7auAuSRcBrwDn7lkTzMysJ2ggDX/tewhmZl0naUlE1HSWz28qm5kZ4IBgZmYpBwQzMwMc\nEMzMLOWAYGZmgAOCmZmlHBDMzAxwQDAzs5QDgpmZAQ4IZmaWckAwMzPAAcHMzFIOCGZmBjggmJlZ\nygHBzMwABwQzM0s5IJiZGeCAYGZmqZICgqRaScsk1Uua3cF2Sbo+3f6UpKPT9KGS/ijpSUlLJV1Z\nVOYKSaskPZFOp/dcs8zMrKtynWWQlAVuAE4DVgKLJc2LiGeKsk0HpqTTccCcdN4InBIRDZIqgIcl\n/ToiFqXlvh8R1/Zcc8zMbE+VcoVwLFAfEcsjogm4E5jRLs8M4NZILAJGSRqfrjekeSrSKXqq8mZm\n1nNKCQgTgRVF6yvTtJLySMpKegJYByyIiMeK8n0h7WK6RdLoLtfezMx6TK/fVI6IfEQcCUwCjpV0\nRLppDnAQcCSwGriuo/KSZkmqk1S3fv363q6umVnZKiUgrAImF61PStO6lCci3gQeAGrT9bVpsCgA\nN5N0Te0iIm6KiJqIqBk7dmwJ1TUzsz1RSkBYDEyRdKCkSuA8YF67PPOA89OnjaYBGyNitaSxkkYB\nSBpGcmP6uXR9fFH5jwFPd7MtZmbWDZ0+ZRQRLZIuBeYDWeCWiFgq6eJ0+43AfcDpQD2wFbgwLT4e\nmJs+qZQB7oqIe9Jt10g6kuQm88vA53qsVWZm1mWKGDgP/dTU1ERdXV1fV8PMbECRtCQiajrL5zeV\nzcwMcEAwM7OUA4KZmQEOCGZmlnJAMDMzwAHBzMxSDghmZgY4IJiZWcoBwczMAAcEMzNLOSCYmRng\ngGBmZikHBDMzAxwQzMws5YBgZmaAA4KZmaUcEMzMDHBAMDOzlAOCmZkBJQYESbWSlkmqlzS7g+2S\ndH26/SlJR6fpQyX9UdKTkpZKurKozBhJCyS9kM5H91yzzMysqzoNCJKywA3AdGAqMFPS1HbZpgNT\n0mkWMCdNbwROiYj3AkcCtZKmpdtmAwsjYgqwMF03M7M+UsoVwrFAfUQsj4gm4E5gRrs8M4BbI7EI\nGCVpfLrekOapSKcoKjM3XZ4LnN2dhpiZWfeUEhAmAiuK1lemaSXlkZSV9ASwDlgQEY+lecZFxOp0\neQ0wrqOdS5olqU5S3fr160uorpmZ7Ylev6kcEfmIOBKYBBwr6YgO8gQ7rhzab7spImoiombs2LG9\nXFszs/JVSkBYBUwuWp+UpnUpT0S8CTwA1KZJayWNB0jn60qvtpmZ9bRSAsJiYIqkAyVVAucB89rl\nmQecnz5tNA3YGBGrJY2VNApA0jDgNOC5ojIXpMsXAHd3sy1mZtYNuc4yRESLpEuB+UAWuCUilkq6\nON1+I3AfcDpQD2wFLkyLjwfmpk8qZYC7IuKedNvVwF2SLgJeAc7tuWaZmVlXKem+Hxhqamqirq6u\nr6thZjagSFoSETWd5fObymZmBjggmJlZygHBzMwABwQzM0s5IJiZGeCAYGZmKQcEMzMDHBDMzCzl\ngGBmZoADgpmZpRwQzMwMcEAwM7OUA4KZmQEOCGZmlnJAMDMzwAHBzMxSDghmZgY4IJiZWaqkgCCp\nVtIySfWSZnewXZKuT7c/JenoNH2ypAckPSNpqaQvFpW5QtIqSU+k0+k91ywzM+uqXGcZJGWBG4DT\ngJXAYknzIuKZomzTgSnpdBwwJ523AF+JiMcljQCWSFpQVPb7EXFtzzXHzMz2VClXCMcC9RGxPCKa\ngDuBGe3yzABujcQiYJSk8RGxOiIeB4iIzcCzwMQerL+ZmfWQUgLCRGBF0fpKdj2pd5pH0gHAUcBj\nRclfSLuYbpE0uqOdS5olqU5S3fr160uorpmZ7Ym9clNZ0nDg58CXImJTmjwHOAg4ElgNXNdR2Yi4\nKSJqIqJm7Nixe6O6ZmZlqZSAsAqYXLQ+KU0rKY+kCpJgcHtE/KI1Q0SsjYh8RBSAm0m6pszMrI+U\nEhAWA1MkHSipEjgPmNcuzzzg/PRpo2nAxohYLUnAvwPPRsT3igtIGl+0+jHg6T1uhZmZdVunTxlF\nRIukS4H5QBa4JSKWSro43X4jcB9wOlAPbAUuTIufAHwa+LOkJ9K0r0XEfcA1ko4EAngZ+FyPtcrM\nzLpMEdHXdShZTU1N1NXV9XU1zMwGFElLIqKms3x+U9nMzAAHBDMzSzkgmJkZ4IBgZmYpBwQzMwMc\nEMzMLOWAYGZmgAOCmZmlyiIgvPAC/PKXfV0LM7P+rSwCwrXXwqc/DY2NfV0TM7P+qywCwllnQUMD\nPPhgX9fEzKz/KouAcMopMGwYzGs/RquZmbUpi4AwbBicdhr86lcwgMbyMzPbq8oiIEDSbfSXv8Cf\n/9zXNTEz65/KJiCccUYyd7eRmVnHyiYg7LcfHHts0m1kZma7KpuAAEm30R//CGvW9HVNzMz6n7IK\nCGeemczvuadv62Fm1h+VFBAk1UpaJqle0uwOtkvS9en2pyQdnaZPlvSApGckLZX0xaIyYyQtkPRC\nOh/dc83q2LvfDe94h7uNzMw60mlAkJQFbgCmA1OBmZKmtss2HZiSTrOAOWl6C/CViJgKTAM+X1R2\nNrAwIqYAC9P1XiUl3UYLFsC2bb29NzOzgaWUK4RjgfqIWB4RTcCdwIx2eWYAt0ZiETBK0viIWB0R\njwNExGbgWWBiUZm56fJc4OxutqUkZ56ZBIOFC/fG3szMBo5SAsJEYEXR+kp2nNRLziPpAOAo4LE0\naVxErE6X1wDjOtq5pFmS6iTVrV+/voTqvrWTToLhw91tZGbW3l65qSxpOPBz4EsRsan99ogIoMN3\niCPipoioiYiasWPHdrsuQ4ZAbW1yY7lQ6PbHmZkNGqUEhFXA5KL1SWlaSXkkVZAEg9sj4hdFedZK\nGp/mGQ+s61rV99yZZ8Krr8Ljj++tPZqZ9X+lBITFwBRJB0qqBM4D2r/vOw84P33aaBqwMSJWSxLw\n78CzEfG9DspckC5fANy9x63ootNPh0zG3UZmZsU6DQgR0QJcCswnuSl8V0QslXSxpIvTbPcBy4F6\n4GbgkjT9BODTwCmSnkin09NtVwOnSXoBODVd3yv23ReOP97DWJiZFVMMoOE/a2pqoq6urkc+65pr\n4KtfTQa8mzy58/xmZgOVpCURUdNZvrJ6U7nYWWclc7+1bGaWKNuAcMgh8M53utvIzKxV2QYEKXna\n6He/S35e08ys3JVtQICk26ipKRnKwsys3JV1QDjhBBg1yt1GZmZQ5gGhogKmT4d774V8vq9rY2bW\nt8o6IEDSbbR+ffLDOWZm5azsA0JtLeRy7jYyMyv7gDBqFJx4ooexMDMr+4AASbfR0qWwfHlf18TM\nrO+UR0C49174h3/Y7ebW31r2VYKZlbPyCAiLFsH3vw8//nGHmw8+GA47zAHBzMpbeQSEK66AU06B\nSy6Bp57qMMtZZ8GDD8LGjXu3amZm/UV5BIRsFv7zP2H0aPjEJ2DTLj/axplnQksL3H9/H9TPzKwf\nKI+AADBuHNx5Z3Ln+LOfhXbDfk+blvxOgruNzKxclU9AAPjAB+Cb34Sf/hR++MOdNmWzcMYZcN99\nyZWCmVm5Ka+AAPCP/wgf/Sh85Svw2GM7bTrzTHjjjeRegplZuSm/gJDJwNy5MGECnHsuvP5626aP\nfCTpWbrwQli5sg/raGbWB0oKCJJqJS2TVC9pdgfbJen6dPtTko4u2naLpHWSnm5X5gpJqzr4reXe\nN2ZM0m20Zg2cfz4UCgAMHw6//nXypNGHP7xTrDAzG/Q6DQiSssANwHRgKjBT0tR22aYDU9JpFjCn\naNtPgNrdfPz3I+LIdLqvi3Xvnve9D773veSmwXe+05Z81FFw993JveczzoAtW/ZqrczM+kwpVwjH\nAvURsTwimoA7gRnt8swAbo3EImCUpPEAEfEQsKEnK91jLrkEzjsPvv51+P3v25JPPjl5IGnxYjjn\nnORHdMzMBrtSAsJEYEXR+so0rat5OvKFtIvpFkmjO8ogaZakOkl169evL+Eju0CCm26CKVOSwLBm\nTdums89ONs2fD5/5TFuvkpnZoNWXN5XnAAcBRwKrges6yhQRN0VETUTUjB07tudrMWIE/Oxnyctq\nM2fu9MzpRRfBt78Nd9wBX/rSLq8umJkNKqUEhFXA5KL1SWlaV/PsJCLWRkQ+IgrAzSRdU33jiCNg\nzpyk2+jyy3fa9NWvJuPi/eAHySsMZmaDVa6EPIuBKZIOJDnJnwf8z3Z55gGXSroTOA7YGBGr3+pD\nJY0vyvMx4Om3yt/rLrgA/vAH+Na3YORI+PKXobISCb77XXjtNbjssuRt5osv7tOampn1ik6vECKi\nBbgUmA88C9wVEUslXSyp9dR4H7AcqCf5tn9Ja3lJdwCPAodIWinponTTNZL+LOkp4IPAl3uqUXvs\nBz9I3k6bPRve+1747W+B5NWFH/0oeerokkuSHiYzs8FGMYA6xmtqaqKurq73d3TPPfDFLybPnp5z\nDlx3Hey/P1u3Ju8nLF6cPK36oQ/1flXMzLpL0pKIqOksX/m9qVyKj340+Qm1q65KzvyHHQb/9E9U\nZbbzq1/BIYckTyHtjdhkZra3OCDsztChyfsJzz2X9BVddhkcfjijH7mH++9P7iV88INJL1M+30P7\nnDMnuSJZsMCPNJnZXueA0Jl3vCMZ5mLBAqishDPPZMKsj/LI3HqOPx7+/u/h+OPhySe7t5vmm38C\nl1xC8933woc/zBuT38PSr9zCc09sZ/PmHmmJmdlbckAo1amnJmf9a6+Fhx5iwmmHc//7r+T22wq8\n9BIcc0zyiOrWrV372G3b4J5Z89Csz/IbTuPA6vVcwE/4y6oMh3/vIkYftT/XjrySg4avY+rU5B7G\n3/wNXHklrFjR+eebmZUsIgbMdMwxx0S/8OqrETNnRkDEuefG66u2xUUXJasHHhhx//2df8TmzRHX\nXBMxY/SDsZWh8cyI98XCuzdHoRCxfXvE8hcL8eT3F8bKo86IgGjKDonf7v83cd4Rf44JEyIymYjR\noyN++cveb66ZDWxAXZRwju3zk3xXpn4TECIiCoWI7343+ROeeGLE66/Hgw9GHHJIkjRzZsSaNbsW\ne/PNiKuuihgzJuI9PBENuZGx5R2HRqxfv/t9PfdcxN/9XcSwYcmHn3ZarLz5vjj6qEJAxBe+kAQR\nM7OOOCDsLXfcEVFZGXHYYREvvxzbt0dcfnmSNHp0xM03R+TzEa+9FvH1r0fss0/yV7/ogy9G45hx\nEZMmRbzySmn7eu21iG99K2L8+AiI/LTj43vnPRYQcdRREc8/36stNbMBygFhb/r97yNGjYrYb7+I\nxx+PiIhnn434wAei7WRdXZ0sn3NOxFO/WR1x0EHJZcIzz3R9f42NSaTZb78IiBUnfTKO2OcvMXx4\nxG239XDbzGzAKzUg+KZyTzjpJHj4YaioSH63ef58Dj0UHnggecN582aYMQOefhp+9qM3efc/1sLa\ntTveceiqykr47Gfh+efha19j0qKf8WTjIVw/5nI+9+ktfOYz0NDQ4600s0HOAaGnHH44LFoEBx+c\nvLfw4x+TySQjpr7wAtx+Oxx+0DY46yx45hn4xS/guOO6t88RI5IR95YtIzPjLC78yzdYPeJdMHcu\n7zum0O1HYc2svDgg9KQJE+Chh+CUU3Y8GxrpC2YtLclvLjz8MNx2W/L8aE/Zf//kF30eeYSRh03i\nJ3yGO146ji+/72H+9V/9jpuZlcYBoaeNHAn33puMnnrFFUnXTlMTzJoF8+Ylrzb/9V/3zr6PPx4e\nfRRuu41377ua3zWfyNjP/w/+rvYltm3rnV2a2eDhwe16S0Ty2wpXXQUHHAAvv5ysX3HF3tn/li0U\nvnst+W9fwzaqGPHmCjRs6N7Zt5n1Kx7crq9J8I1vJL/DuWIFfP7zu/z4Tq+qriZzxeVULH+eET/7\nsYOBmXXKVwh7w8aNSVeS1Nc1MbMyVOoVQim/mGbdtc8+fV0DM7NOucvIzMwABwQzM0uVFBAk1Upa\nJqle0uwOtkvS9en2pyQdXbTtFknrJD3drswYSQskvZDOR3e/OWZmtqc6DQiSssANwHRgKjBT0tR2\n2aYDU9JpFjCnaNtPgNoOPno2sDAipgAL03UzM+sjpVwhHAvUR8TyiGgC7gRmtMszA7g1HUdpETBK\n0niAiHgI2NDB584A5qbLc4Gz96QBZmbWM0oJCBOB4t/mWpmmdTVPe+MiYnW6vAYY11EmSbMk1Umq\nW79+fQnVNTOzPdEvbiqnw7N2+EJERNwUETURUTN27Ni9XDMzs/JRSkBYBUwuWp+UpnU1T3trW7uV\n0vm6EupiZma9pJSAsBiYIulASZXAecC8dnnmAeenTxtNAzYWdQftzjzggnT5AuDuLtTbzMx6WKcB\nISJagEuB+cCzwF0RsVTSxZIuTrPdBywH6oGbgUtay0u6A3gUOETSSkkXpZuuBk6T9AJwarpuZmZ9\nxGMZmZkNch7t1MzMusQBwczMAAcEMzNLOSCYmRnggGBmZikHBDMzAxwQzMws5YBgZmaAA4KZmaXK\nJiDk831dAzOz/q0sAsJ118FHPgItLX1dEzOz/qssAsLYsbBwIVx5ZV/XxMys/yqLgHD++XDhhfDN\nb8JvftPXtTEz65/KIiAA/PCHMHUqfOpT8OqrfV0bM7P+p2wCQlUV/PSnsGULzJzp+wlmZu2VTUAA\nOOwwuPFGeOghuPzyvq6NmVn/UlYBAeDTn4aLLoJvfQvmz+/r2piZ9R9lFxAArr8ejjgiuZ+walVf\n18bMrH8oKSBIqpW0TFK9pNkdbJek69PtT0k6urOykq6QtErSE+l0es80qXOt9xO2bfP9BDOzVp0G\nBElZ4AZgOjAVmClparts04Ep6TQLmFNi2e9HxJHpdF93G9MVhx4K//Zv8Ic/wP/9v3tzz2Zm/VMp\nVwjHAvURsTwimoA7gRnt8swAbo3EImCUpPEllu0zn/wk/O3fwre/Db/+dV/Xxsysb5USECYCK4rW\nV6ZppeTprOwX0i6mWySNLrnWXbV+PTQ0dLjpX/4F3vOe5GbzypW9VgMzs36vL28qzwEOAo4EVgPX\ndZRJ0ixJdZLq1q9fv2d7uuoqGDcuOevPn7/TTYNhw+Cuu6CxEc47D5qb92wXZmYDXSkBYRUwuWh9\nUppWSp7dlo2ItRGRj4gCcDNJ99IuIuKmiKiJiJqxY8eWUN0OfPKTySNF99wDtbUweTL8wz/An/4E\nERxyCNx0EzzyCFx22Z7twsxsoCslICwGpkg6UFIlcB4wr12eecD56dNG04CNEbH6rcqm9xhafQx4\nuptt2b3jjkvuIK9ZAz//OUybloxlcfTR8O53w9VXM/OvVvC5z8F3vpPEj7lz3YVkZuVFEdF5puSR\n0H8GssAtEfFNSRcDRMSNkgT8EKgFtgIXRkTd7sqm6beRdBcF8DLwuTSI7FZNTU3U1dXtSTt3tWFD\n0ld0223w3/8NEvkPnMxthU/xo6enUffGQTQylEMOgQ99CE49FU4+GUb33p0OM7NeIWlJRNR0mq+U\ngNBf9GhAKPbii3D77UlwqK9vS948ahIvZd7Jkk3vZFnLwSzXO6mY+k6mfORgTjx9BMcfn9yDMDPr\nzxwQ9kQEPPkkPPNMEhjq6+HFF4n6erRu3U5Z1/J2nuUwnh89jQ3vmgbvfz8HHDeOww+Hd70Lhgzp\nvWqamXWFA0JP27w5uZKor6fxmRdZ+0g9maefYtzqP1ERyaNJL3EAj/J+HtP7efUd08ge/V4OeXcl\nhx8OBxyQ3MseNw4yZTlgiJn1FQeEvWX7dnj8cVoeXkTDbx+lcsmjVG1IHsLarqHUxTEsYhprGcdW\nqmjMVDFszDCq9q1ixLgqRk2oYvTEKsbuX8Xb9x/Gvu8cxajJI6iuBqmP22Zmg4IDQl9auRIefRQW\nLaLwyKPw+BIyzU0lF9/ISFYwmXUVk3ht2GQ2jphEw+jJbN93Es37TYbJk6keN5x99oGRI9llPnIk\nDB/uKxEzSzgg9CeFQjKS3tatu0yxZSub125lw8qtbFy9lcbVG8iuWcnQ9SupfnMlozavYNT2tbt8\n5EZG0kwFGQpkKJAl37ZcPGUp0EwFG3Nj2Fwxhi2VY9gy9G1sGzaG7VVjaKweQ9OIt5HfZwz5fcag\nEcPRiOFkRlSTHVlNbp9qKvapYugwMWQIDB268zRs2I7lbLYP/rZm1qlSA0Jub1Sm7GUyUF2dTO0I\nGJlOu9XUlPzu54oVsHIl8ZcVDH15FZlteZqaM2xrybC9JUtjc4bGpgzb0/m2pgzbmzLEtkaGbHuD\nqm0bqG7cwKiNrzD59T+xT34D1bGl0+oXEFupYgvVNDCcLVTzOsNoIUczFTRTQQs5WlRBIVNB5HIU\nshVEtoLIVdBcMYymimqaK6pprqymZUgVLZXVtAytpjC0msLQKgrDqonKIWRzkM0kwaV4ymQgl0vn\n2SCXg8qKIJcNKioglwsqc0l6RW5HOoUC27YU2NaQZ9vWAtsb8mzfWqBxa57GbQW2b83TuLVA47Y8\np/2vI3n/J9qPymJWPhwQBoLKyuSu9AEHAEkQGZJO3dbYCG+8Qbz2Oi3rNtD8RgPNG7fQ8mYD+Y1b\nyG/aQmFzA7F5C2zZwpCGBoZu3QLbt0FzSzLWR8sWaGlBLc3JlG8hk28mk28m29RMZX4rQ/NbyEa+\nJ2rca/606D/gE5/s62qY9RkHhHI3ZAjstx/abz8qgIre3FdTU/Kj1lu2JF1mrcut642Nu9xJj0im\nQgHy+R3zfEHk89CcFy0tyXJLXsnUsmMZiaHVWYZWZRhanWVYdTIfMixDpiK99EgvQY46+ODebL1Z\nv+eAYHtPZWUydeF1b6VTBv9jNettfg7FzMyAMgkIjY1JV4OZme1eWQSEb38b3vte+MUvkv5oMzPb\nVVkEhPe8J7mfec45cMwx8KtfOTCYmbVXFgHh4x+HpUuT3zjYuBHOOiv5SYT58x0YzMxalUVAgOSl\npvPPh+eeg5tvTn4rp7YWTjwRHnigr2tnZtb3yiMgvPIKLFsGmzdTUQGf/Sw8/zzccAO89BKcckoy\nPfxwX1fUzKzvlMVYRv960p08+JCo5X5qq/7A+ElZmDABJkyg+e0T+O+XJnD7AxN4dtMEDj5hPFNP\nGM2wcSMZMaaCUaNg1Khk4LjW+T77eNweMxs4PJZRkVcP+SC/f2Ikd236a9gK7133MrVbH6Z22X0c\nv+HfOKlxMye1Zn4knYBtDGUTI9umVxjRtrwtN5L8kCrI5VBFFlXkyOSyZCqyqDJHtiJLpjJHtjKZ\nqyKHKit2zCsryFTmyAzZsZwdWkFmSAWZtGzbvOhzskOS5eyQHJlchmxFJpnn1Lacq0zSszntmIrG\nBGpd9vDaZlasLALCky+NYN2mYQwfXmDc28VrjQfw3VcP4DuFTzF8eHDqqc3UHvMatVNeZP/8cvIb\nNtK4fhNNr20iu2ETI97czPCNm9DmTWQaVpDbuonK7ZuoaNyKtuXJRku/H6engIiiqbndeuuUJ0tB\nWQpk2pZ3pGXJK1kOMskrxKQzgYid1ltneeVoyVTQkqkkrwqaM5XkMxXklaZlk3lBuSRKSYQyhITS\n5eL01mUAMkr3pzR/sleJZD3d1pZfaqugVFQOyFBAkYwSKwpkokAm8m3LiiQ9+Uux02d2tK9QhkI2\nRyFbST5bSSFXSSFbQT5X2ZYWuWQ9lE33m4zVIYIMhbZlEcn+W+ucyRKZDCiT/E3SSN+6HEp6g1vb\n1NouokCmkG9ry452pfsVbfMdx3PnXoT2XyTa1ovSBUy8qJbxJxxU6j9R6wdKCgiSaoF/AbLAjyLi\n6nbblW4/HdgKfCYiHn+rspLGAP8POAB4GTg3It7ofpN2teyRDUAVDQ0ZtjTkqaSJEWxnKNuoatjK\nmnvX8at7X+N+Whi5z0hGjxnNkKoMQ6syVFWJYcOzVI3NUDUiR2V1BZXVFQwZUUllVfJtPTckS7Yi\nQy4H2ZzI5YJcup7LBtlskFOBjApkVYBCgUJznmgpoOYmCk1N0NRCobGZ/PZmCs35ZGpqSfI1tyTL\nLXmiqaVtPQpB5As7pvbraRqFAoVCQCGSPOlE7LweEWQKeSgkgwYpn0eFdIody5l0e7SeKtLxhgKI\nUDqnNYFMtJAtNJMtNJMrNFFRaGJYvoFsoZmKQhPZaCYXzeQKzdB28ou2ky8RO50sMyRvGbadqCJ2\nOXm1zduh+mN2AAAGX0lEQVQq0sE2di7XFgTbDSLePi3Q7vdX9JkZCuRooZImKmkiR//+0tDTHhv7\ncweEAabTgCApC9wAnAasBBZLmhcRzxRlmw5MSafjgDnAcZ2UnQ0sjIirJc1O17/ac03bYeGw0xmx\n7RUaGMGbjGIDY3idt9HA8LapiUoyFKjc2EhuYwvBzv/JtwGNyRku+WZF67dFwU4ngdYTkNKTQyHN\nsetpCWg7tbHTNtrKF8jsVHrH6D5Bpq1E62mq0JaWSdPbvs22lSpeVloDIQrpetv34x0tVYZQBXmG\ntKUFWUS+7amELPm0Lju+3bbtV4W0Rulnpt+og+QbbQChLJGBbFu59DPUukcl32Zb17Mio7QNIql/\nKN0fbcvJXyN2XFmQoSCSK4/Wv5aSo5TLBtlMW7EdM+04Qghi+EgO+9rHk/a25otkWem369ZB+Vor\nFIiIJGArnyfyLURz8uUg8klwz6QFkiBb9K280LqfJOomxy6I1rfvI9L2RvL3SHceIci0Xq0kV1u0\nHvf0GLSGMGXSfykZQaRXYq1pbX+r1n/Z0HbvsW30waJ/o4VkPnpiFS/+11OUJJNp+zu1XmUpo13T\nOujm7LDrs+hqpzhtp/nuPmh3893urMQ6daT1y0zsZoK27W87en+q9xtR4gfvmVKuEI4F6iNiOYCk\nO4EZQHFAmAHcGsm/kkWSRkkaT/Ltf3dlZwAnp+XnAr+nlwLC5NOmwoJVjGrexKTm1yH/PIV8gZZC\ncvrJkKeSlt7Y9eBQHI8GsUd4PyfwaGmZL/yn3q2MWTt1V/2amq/X9uo+SgkIE4EVResrSa4COssz\nsZOy4yJidbq8BhjX0c4lzQJmpasNkpaVUOeO7Au8tnPSgL+E76BNA1oft6fEYNA1Pkb938Bo02XT\n4bKScnbUnv1LKdgvbipHREjq8DtoRNwE3NTdfUiqK+Wxq4FksLVpsLUHBl+bBlt7YPC1qTvtKeXF\ntFXA5KL1SWlaKXnequzatFuJdL6u9GqbmVlPKyUgLAamSDpQUiVwHjCvXZ55wPlKTAM2pt1Bb1V2\nHnBBunwBcHc322JmZt3QaZdRRLRIuhSYT/Lo6C0RsVTSxen2G4H7SB45rSd57PTCtyqbfvTVwF2S\nLgJeAc7t0ZbtqtvdTv3QYGvTYGsPDL42Dbb2wOBr0x63Z0ANXWFmZr2nPAa3MzOzTjkgmJkZUCYB\nQVKtpGWS6tO3ogc0SS9L+rOkJyR1ffjXfkDSLZLWSXq6KG2MpAWSXkjno/uyjl2xm/ZcIWlVepye\nkHR6X9axKyRNlvSApGckLZX0xTR9IB+j3bVpQB4nSUMl/VHSk2l7rkzT9/gYDfp7COnwGc9TNHwG\nMLPd0BsDiqSXgZqI6P8v0+yGpA8ADSRvuB+Rpl0DbCgazmR0RPTK2+s9bTftuQJoiIhr+7JueyJ9\nFHx8RDwuaQSwBDgb+AwD9xjtrk3nMgCPUzqGXHVENEiqAB4Gvgh8nD08RuVwhdA29EZENAGtw2dY\nH4qIh4AN7ZJnkAxjQjo/e69Wqht2054BKyJWtw5QGRGbgWdJRh4YyMdod20akCLRkK5WpFPQjWNU\nDgFhd8NqDGQB/FbSknRoj8GipOFMBpgvSHoq7VIaMN0rxSQdABwFPMYgOUbt2gQD9DhJykp6guTF\n3gUR0a1jVA4BYTD6q4g4kmSU2c+n3RWDSjpQ4kDvz5wDHAQcCawGruvb6nSdpOHAz4EvRcSm4m0D\n9Rh10KYBe5wiIp+eCyYBx0o6ot32Lh2jcggIpQy9MaBExKp0vg74L5JuscFgUA1nEhFr0/+wBeBm\nBthxSvulfw7cHhG/SJMH9DHqqE0D/TgBRMSbwANALd04RuUQEEoZemPAkFSd3hBDUjXwYeDpty41\nYAyq4Uxa/1OmPsYAOk7pDct/B56NiO8VbRqwx2h3bRqox0nSWEmj0uVhJA/OPEc3jtGgf8oIIH2M\n7J/ZMXzGN/u4SntM0kEkVwWQDD3ynwOxPZLuIPk9jH2BtcDlwC+Bu4B3kA5nEhED4kbtbtpzMkk3\nRJD8KuDnivp2+zVJfwX8Afgzrb/jBF8j6XMfqMdod22ayQA8TpLeQ3LTOEvy5f6uiPiGpLexh8eo\nLAKCmZl1rhy6jMzMrAQOCGZmBjggmJlZygHBzMwABwQzM0s5IJiZGeCAYGZmqf8Pl0+mqTEXwowA\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f60f00991d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%time\n",
    "nb_epoch = 30\n",
    "batch_size = 128\n",
    "n_layers = 4\n",
    "thickness = 6\n",
    "\n",
    "early_stopping = EarlyStopping(monitor='val_loss', patience=3, verbose=3, mode='min')\n",
    "tensorboard = TensorBoard(log_dir='./Graph', histogram_freq=0, write_graph=True) # you can try to run this visualization\n",
    "\n",
    "save_decoded_dict = {}\n",
    "save_encoded_dict = {}\n",
    "    \n",
    "for encoding_dim in tqdm(l_encoding_dim):\n",
    "    print(\"encoding_dim : \", encoding_dim)\n",
    "\n",
    "    autoencoder, encoder, decoder = create_autoencoder_aux(\n",
    "        train_input.shape[1], encoding_dim, \n",
    "        n_aux_features=train_aux.shape[1], n_layers=n_layers, thickness=thickness\n",
    "    )\n",
    "\n",
    "    history = autoencoder.fit(\n",
    "        [train_input, train_aux], train_input,\n",
    "        epochs=nb_epoch, batch_size=batch_size,\n",
    "        shuffle=True, verbose=1,\n",
    "        validation_split=val_size, callbacks =[early_stopping, tensorboard]\n",
    "    )\n",
    "\n",
    "    score = autoencoder.evaluate([test_input, test_aux], test_input, batch_size=batch_size)\n",
    "    # print (\"\\n\\nscore : \", score)\n",
    "    l_score.append(score)\n",
    "\n",
    "    encoded_test = encoder.predict([test_input, test_aux])\n",
    "    decoded_test = decoder.predict([encoded_test, test_aux]) # test mode = 0\n",
    "#     print_features_histograms(pd.DataFrame(decoded_test), target=pd.DataFrame(test_input))\n",
    "\n",
    "    save_decoded_dict[encoding_dim] = decoded_test.astype(np.float64)\n",
    "    save_encoded_dict[encoding_dim] = encoded_test.astype(np.float64)\n",
    "\n",
    "#     plt.clf()\n",
    "    plt.plot(history.history['loss'], \"b\")\n",
    "    plt.plot(history.history['val_loss'], \"r\")\n",
    "    plt.ylim(0, 0.04)\n",
    "#     plt.xlim(0, 40)\n",
    "    print(\"\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Save what you want to retain to this dictionary\n",
    "retain_dict = {}\n",
    "retain_dict['save_decoded_dict'] = save_decoded_dict\n",
    "retain_dict['save_encoded_dict'] = save_encoded_dict\n",
    "retain_dict['test_input'] = test_input.astype(np.float64)\n",
    "retain_dict['test_aux'] = test_aux.astype(np.float64)\n",
    "retain_dict['test_MC'] = test_MC\n",
    "retain_dict['l_score'] = l_score\n",
    "retain_dict['vars_list_input'] = vars_list_input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_input_unslaced = test_input.copy()\n",
    "test_aux_unslaced = test_aux.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %%time\n",
    "for i, var in enumerate(vars_list_input):\n",
    "    test_input_unslaced[:, i] = scalers[var]['std'].inverse_transform(\n",
    "        scalers[var]['max'].inverse_transform(test_input[:, i].reshape(-1, 1))\n",
    "    ).reshape(-1)\n",
    "    \n",
    "for i, var in enumerate(vars_list_aux):\n",
    "    test_aux_unslaced[:, i] = scalers[var]['std'].inverse_transform(\n",
    "        scalers[var]['max'].inverse_transform(test_aux[:, i].reshape(-1, 1))\n",
    "    ).reshape(-1)\n",
    "\n",
    "retain_dict['test_input_dec'] = test_input_unslaced\n",
    "retain_dict['test_aux_dec'] = test_aux_unslaced"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "save_decoded_dict_dec = {}\n",
    "for encoding_dim in save_decoded_dict:\n",
    "    tmp_decoded_unscaled = save_decoded_dict[encoding_dim].copy()\n",
    "    for i, var in enumerate(vars_list_input):\n",
    "        tmp_decoded_unscaled[:, i] = scalers[var]['std'].inverse_transform(\n",
    "            scalers[var]['max'].inverse_transform(tmp_decoded_unscaled[:, i].reshape(-1, 1))\n",
    "        ).reshape(-1)\n",
    "    save_decoded_dict_dec[encoding_dim] = tmp_decoded_unscaled\n",
    "\n",
    "retain_dict['save_decoded_dict_dec'] = save_decoded_dict_dec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pickle.dump(retain_dict, open(\"retain_dict_{}.p\".format(TYPE), \"wb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}