kfold_models.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import cv2 as cv\n",
    "import h5py\n",
    "import matplotlib.pyplot as plt\n",
    "import keras\n",
    "from keras.layers import Conv2D, MaxPool2D, CuDNNGRU, GlobalMaxPool2D, Reshape, \\\n",
    "concatenate, Input, TimeDistributed, Dense, BatchNormalization, SpatialDropout2D, SpatialDropout1D, Dropout, GlobalAvgPool2D, Flatten\n",
    "from keras import Model\n",
    "from keras.applications import Xception\n",
    "import keras.backend as k\n",
    "from sklearn.model_selection import train_test_split, KFold\n",
    "from keras.callbacks import EarlyStopping, ModelCheckpoint\n",
    "from sklearn.metrics import mean_squared_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mean intensity of optical flow: -0.03514303\n",
      "mean intensity of frame: 61.88129\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "time_history = 1 # how many time steps to look back\n",
    "hdf5_path = './train_data/train.hdf5'\n",
    "with h5py.File(hdf5_path, \"r\") as f:\n",
    "    print(\"mean intensity of optical flow:\" , f[\"op_flow\"][1].mean())\n",
    "    print(\"mean intensity of frame:\", f[\"frame\"][5].mean())\n",
    "    plt.imshow(f[\"frame\"][5]/255)\n",
    "    train_size = len(f[\"speed\"])\n",
    "    speed_data = list(f[\"speed\"])\n",
    "#     for i in range(train_size,train_size - 100):\n",
    "        \n",
    "#         frame, axarr = plt.subplots(1,2)\n",
    "#         axarr[0].imshow(f[\"frame\"][i])\n",
    "#         axarr[1].imshow(f[\"op_flow\"][i])\n",
    "#         print(f['speed'][i])\n",
    "#         plt.show()\n",
    "#     plt.plot(speed_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mean intensity of optical flow: -0.005691097\n",
      "20401\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "hdf5_path_comma = './train_data/comma_train.hdf5'\n",
    "with h5py.File(hdf5_path_comma, \"r\") as f:\n",
    "    print(\"mean intensity of optical flow:\" ,f[\"op_flow\"][50].mean())\n",
    "    plt.imshow(f[\"frame\"][6500]/255)\n",
    "    print(len(f[\"speed\"]))\n",
    "    comma_train_size = len(f[\"speed\"])\n",
    "    comma_speed_data = list(f[\"speed\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mean intensity of optical flow: -0.09879465\n",
      "10798\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "hdf5_path_comma_test = './test_data/comma_test.hdf5'\n",
    "with h5py.File(hdf5_path_comma_test, \"r\") as f:\n",
    "    print(\"mean intensity of optical flow:\" ,f[\"op_flow\"][2000].mean())\n",
    "    plt.imshow(f[\"frame\"][2000]/255)\n",
    "    print(len(f[\"speed\"]))\n",
    "    comma_test_size = len(f[\"speed\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x25de756eda0>]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(comma_speed_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "151833 20401\n"
     ]
    }
   ],
   "source": [
    "print(train_size, comma_train_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "class DataGenerator(keras.utils.Sequence):\n",
    "    def __init__(self, batch_size, history_size, hdf5_path, indexes = None, validation_mode = False):\n",
    "        self.hdf5_path = hdf5_path\n",
    "        if indexes is None:\n",
    "            with h5py.File(self.hdf5_path, \"r\") as f:\n",
    "                self.indexes = np.arange(len(f[\"speed\"]))\n",
    "        else:\n",
    "            self.indexes = indexes\n",
    "        self.batch_size = batch_size\n",
    "        self.history_size = history_size\n",
    "        self.validation_mode = validation_mode\n",
    "        if self.validation_mode == False:\n",
    "            self.on_epoch_end()\n",
    "    def __len__(self):\n",
    "        'Denotes the number of batches per epoch'\n",
    "        return int(np.ceil(len(self.indexes) / self.batch_size))\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        'Generate one batch of data'\n",
    "        # Generate indexes of the batch\n",
    "        indexes = self.indexes[index*self.batch_size:(index+1)*self.batch_size]\n",
    "\n",
    "        # Generate data\n",
    "        return self.__data_generation(indexes)\n",
    "\n",
    "    def on_epoch_end(self):\n",
    "        'Updates indexes after each epoch'\n",
    "        if self.validation_mode == False:\n",
    "            np.random.shuffle(self.indexes)\n",
    "\n",
    "    def __data_generation(self, indexes):\n",
    "        'Generates data containing batch_size samples'\n",
    "        indexes = list(indexes)\n",
    "        indexes.sort()\n",
    "        frame = np.zeros((self.batch_size, 224, 224, 3))\n",
    "        op_flow = np.zeros((self.batch_size, 50, 160, 2))\n",
    "        speed = np.zeros((self.batch_size, 1))\n",
    "        with h5py.File(self.hdf5_path, \"r\") as f:\n",
    "            #self.frame = np.array(f[\"frame\"][indexes])[batch_shuffles]\n",
    "            op_flow = np.array(f[\"op_flow\"][indexes])\n",
    "            speed = np.array(f[\"speed\"][indexes])\n",
    "                \n",
    "        return [\n",
    "            #frame,\n",
    "            op_flow], speed"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "index_array = np.arange(int(train_size))\n",
    "#correcting for extra element at beginning of comma training data\n",
    "comma_index_array = np.arange(int(comma_train_size) - 1)\n",
    "comma_test_index_array = np.arange(int(comma_test_size) - 1)\n",
    "# comma_train_indexes_no_shuf = np.copy(comma_train_indexes)\n",
    "# comma_val_indexes_no_shuf = np.copy(comma_val_indexes)\n",
    "# comma_train_indexes, comma_val_indexes = train_test_split(comma_index_array, shuffle = False, test_size = .2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model_flat(history_size):\n",
    "    k.clear_session()\n",
    "    #frame_inp = Input(shape=(224, 224, 3))\n",
    "    op_flow_inp = Input(shape=(50, 160, 2))\n",
    "    filters = [3, 5]\n",
    "    op_flows = []\n",
    "    #op_flow = BatchNormalization()(op_flow_inp)\n",
    "    op_flow = (op_flow_inp)\n",
    "    for i, filter_size in enumerate(filters):\n",
    "        int_layer = Dropout(.0)(op_flow)\n",
    "        int_layer = Conv2D(8, (filter_size,filter_size), activation = \"relu\", data_format = \"channels_last\")(int_layer)\n",
    "        int_layer = MaxPool2D(pool_size = (1,2))(int_layer)\n",
    "        int_layer = Conv2D(16, (filter_size,filter_size), activation = \"relu\", data_format = \"channels_last\")(int_layer)\n",
    "        int_layer = MaxPool2D(pool_size = (1,2))(int_layer)\n",
    "        int_layer = Conv2D(32, (filter_size,filter_size), activation = \"relu\", data_format = \"channels_last\")(int_layer)\n",
    "        int_layer = Conv2D(64, (filter_size,filter_size), activation = \"relu\", data_format = \"channels_last\")(int_layer)\n",
    "        int_layer = Dropout(.0)(int_layer)\n",
    "        int_layer = Conv2D(128, (filter_size,filter_size), activation = \"relu\", data_format = \"channels_last\")(int_layer)\n",
    "        int_layer = MaxPool2D()(int_layer)\n",
    "        int_layer = Conv2D(256, (filter_size,filter_size), activation = \"relu\", data_format = \"channels_last\")(int_layer)\n",
    "        int_layer = MaxPool2D()(int_layer)\n",
    "        int_layer = Conv2D(512, (filter_size,filter_size), activation = \"relu\",\n",
    "                           data_format = \"channels_last\", padding = \"same\")(int_layer)\n",
    "        int_layer = MaxPool2D()(int_layer)\n",
    "        int_layer_max = GlobalMaxPool2D()(int_layer)\n",
    "        int_layer_avg = GlobalAvgPool2D()(int_layer)\n",
    "        conc = concatenate([int_layer_max, int_layer_avg])\n",
    "        op_flows.append(conc)\n",
    "    conc = concatenate(op_flows)\n",
    "    #conc = BatchNormalization()(conc)\n",
    "    #conc = SpatialDropout1D(.2)(conc)\n",
    "    #conc = CuDNNGRU(256)(conc)\n",
    "    conc = Dropout(.2)(conc)\n",
    "    conc = Dense(500, activation = \"relu\")(conc)\n",
    "#     conc = Dropout(.2)(conc)\n",
    "    conc = Dense(250, activation = \"relu\")(conc)\n",
    "#     conc = Dropout(.1)(conc)\n",
    "    result = Dense(1, activation='linear')(conc)\n",
    "    \n",
    "    model = Model(inputs=[\n",
    "        #frame_inp,\n",
    "        op_flow_inp], outputs=[result])\n",
    "    model.compile(loss=\"mse\", optimizer='adam')\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def runningMeanFast(x, N):\n",
    "    return np.convolve(x, np.ones((N,))/N)[(N-1):]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate_metrics(y_true, y_pred, title = None, verbose = False):\n",
    "    #print(pd.DataFrame(y_pred).describe())\n",
    "#     plt.figure(figsize=(12,4))\n",
    "#     plt.subplot(1, 2, 1)\n",
    "#     plt.plot(y_true)\n",
    "#     plt.plot(y_pred)\n",
    "#     plt.title(\"raw \"+ title +  \" predictions\")\n",
    "    for j in range(1,21,5):\n",
    "        rolling_preds = runningMeanFast(y_pred, j)\n",
    "        rolling_mse = mean_squared_error(y_true, rolling_preds)\n",
    "    print(\"unsmoothed mse\" , mean_squared_error(y_true, y_pred) , \"\\t\" + str(j), \"smoothed mse\", str(rolling_mse))\n",
    "#     plt.subplot(1, 2, 2)\n",
    "#     plt.plot(y_true)\n",
    "#     plt.plot(rolling_preds)\n",
    "#     plt.title(\"smoothed \"+ title +  \" predictions\")\n",
    "#     plt.show()\n",
    "    return rolling_mse, rolling_preds\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training on fold 0\n",
      "================================================================================\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 454s 120ms/step - loss: 23.2180\n",
      "custom valid metrics\n",
      "unsmoothed mse 17.845682 \t16 smoothed mse 11.72658360490481\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 5.9751\n",
      "comma val metrics\n",
      "unsmoothed mse 9.990998 \t16 smoothed mse 7.211370667427236\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 456s 120ms/step - loss: 13.5521\n",
      "custom valid metrics\n",
      "unsmoothed mse 20.339148 \t16 smoothed mse 13.216662858436182\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 4.9847\n",
      "comma val metrics\n",
      "unsmoothed mse 9.51533 \t16 smoothed mse 7.981123576114683\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 451s 119ms/step - loss: 11.3718\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.908618 \t16 smoothed mse 7.1700355614965625\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 4.4443\n",
      "comma val metrics\n",
      "unsmoothed mse 8.672356 \t16 smoothed mse 6.7944680330079965\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 449s 118ms/step - loss: 10.0475\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.744772 \t16 smoothed mse 7.000713166746928\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 4.0468\n",
      "comma val metrics\n",
      "unsmoothed mse 9.310321 \t16 smoothed mse 7.6517305377299465\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 455s 120ms/step - loss: 9.1322\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.819598 \t16 smoothed mse 8.135047859210276\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 3.7563\n",
      "comma val metrics\n",
      "unsmoothed mse 10.202352 \t16 smoothed mse 8.331724667753969\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 456s 120ms/step - loss: 7.9932\n",
      "custom valid metrics\n",
      "unsmoothed mse 9.940981 \t16 smoothed mse 5.134545480564958\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 3.6417\n",
      "comma val metrics\n",
      "unsmoothed mse 9.956329 \t16 smoothed mse 8.24851220322694\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 457s 120ms/step - loss: 7.4127\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.104558 \t16 smoothed mse 7.532053440535753\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 3.1574\n",
      "comma val metrics\n",
      "unsmoothed mse 9.476342 \t16 smoothed mse 7.081441729706362\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 454s 120ms/step - loss: 6.7968\n",
      "custom valid metrics\n",
      "unsmoothed mse 10.259793 \t16 smoothed mse 4.955664337566183\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 3.1159\n",
      "comma val metrics\n",
      "unsmoothed mse 9.6241 \t16 smoothed mse 8.23230637031835\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 453s 119ms/step - loss: 6.3536\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.463489 \t16 smoothed mse 7.8503487531950835\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 3.4150\n",
      "comma val metrics\n",
      "unsmoothed mse 10.753209 \t16 smoothed mse 8.781264852189876\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 452s 119ms/step - loss: 5.9969\n",
      "custom valid metrics\n",
      "unsmoothed mse 10.327354 \t16 smoothed mse 5.310194126596548\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 2.8828\n",
      "comma val metrics\n",
      "unsmoothed mse 10.798157 \t16 smoothed mse 8.752072118880172\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fold mean histograms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training on fold 1\n",
      "================================================================================\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 460s 121ms/step - loss: 20.4481\n",
      "custom valid metrics\n",
      "unsmoothed mse 26.747513 \t16 smoothed mse 18.128417452622127\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 5.7771\n",
      "comma val metrics\n",
      "unsmoothed mse 15.151678 \t16 smoothed mse 12.269524962613536\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 460s 121ms/step - loss: 12.1852\n",
      "custom valid metrics\n",
      "unsmoothed mse 18.923029 \t16 smoothed mse 13.558099035766505\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 4.5805\n",
      "comma val metrics\n",
      "unsmoothed mse 14.0099535 \t16 smoothed mse 9.82368662883609\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 458s 121ms/step - loss: 10.3624\n",
      "custom valid metrics\n",
      "unsmoothed mse 20.315228 \t16 smoothed mse 13.600535296426566\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 53s 103ms/step - loss: 4.3625\n",
      "comma val metrics\n",
      "unsmoothed mse 13.34155 \t16 smoothed mse 8.034885175397083\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 442s 116ms/step - loss: 8.5187\n",
      "custom valid metrics\n",
      "unsmoothed mse 18.320545 \t16 smoothed mse 13.426095716574526\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 3.8800\n",
      "comma val metrics\n",
      "unsmoothed mse 12.955739 \t16 smoothed mse 10.124128354774781\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 455s 120ms/step - loss: 7.7386\n",
      "custom valid metrics\n",
      "unsmoothed mse 19.77938 \t16 smoothed mse 13.373748157343815\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 3.7201\n",
      "comma val metrics\n",
      "unsmoothed mse 14.194487 \t16 smoothed mse 10.58621532186525\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 455s 120ms/step - loss: 7.1480\n",
      "custom valid metrics\n",
      "unsmoothed mse 19.994188 \t16 smoothed mse 14.475169750656427\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 3.3685\n",
      "comma val metrics\n",
      "unsmoothed mse 12.305954 \t16 smoothed mse 8.965952505611186\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 455s 120ms/step - loss: 6.4076\n",
      "custom valid metrics\n",
      "unsmoothed mse 17.271408 \t16 smoothed mse 10.318172694844664\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 3.3045\n",
      "comma val metrics\n",
      "unsmoothed mse 13.853578 \t16 smoothed mse 10.180401010888257\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 457s 120ms/step - loss: 5.7863\n",
      "custom valid metrics\n",
      "unsmoothed mse 17.315653 \t16 smoothed mse 10.451808676552389\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 3.2412\n",
      "comma val metrics\n",
      "unsmoothed mse 19.202044 \t16 smoothed mse 14.223487613931594\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 458s 121ms/step - loss: 5.3954\n",
      "custom valid metrics\n",
      "unsmoothed mse 15.71907 \t16 smoothed mse 9.955128200221285\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 2.8327\n",
      "comma val metrics\n",
      "unsmoothed mse 13.61799 \t16 smoothed mse 9.612096919148936\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 456s 120ms/step - loss: 4.9509\n",
      "custom valid metrics\n",
      "unsmoothed mse 18.774666 \t16 smoothed mse 13.533393821255741\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 2.5834\n",
      "comma val metrics\n",
      "unsmoothed mse 15.080271 \t16 smoothed mse 11.365928950366998\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsMAAAEICAYAAAC6S/moAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzs3Xd4XGeV+PHvGY16taSRLFmy5Sbbsi0ptux0O8UOTgJJCAScBJYsgRB6DwRY6gIBloXdLBAChPALaQTSSCGxE2KnW3LvVbJVbPXepXl/f8zITGTJajNzp5zP88yjmblX9x7dGd05897zvq8YY1BKKaWUUioc2awOQCmllFJKKatoMqyUUkoppcKWJsNKKaWUUipsaTKslFJKKaXClibDSimllFIqbGkyrJRSSimlwpYmwyFIRL4rIn/20rYuEZEqb2wrlAw/LiKyV0QuGc+6k9jXPSLyH5P9faVUYNNztn+IyCsi8jH3/ZtF5MXxrDuJ/cwUkQ4RiZhsrMq/NBkOcqF24vPmh4I/GWMWG2Nemep2ROQWEXlt2LZvN8b8YKrbVkpZL9TO2cHKGPOgMeYKb2xLRCpEZI3Htk8YYxKMMYPe2L7yPU2GlVJKKaVU2NJkeIpE5GsiUi0i7SJyUEQudz//XRF5TET+7F62W0TyReROEakTkUoRucJjO9ki8rSINInIERH5uMeyaBH5pYjUuG+/dD8XDzwPZLsvyXSISLb716JE5P+5971XREqG7etvIlIvIuUi8jmPZbEicr+INIvIPmDFGH//YhHZ4I67VkS+4X7+fhH5T4/1hpcVnHHcRGQd8A3gg+6/Zec4js2EjvOw2L8uIn8d9tz/iMj/uu//u4jsd2/3mIh84izH4XTLwFjH0L3fo+7t7hOR97qfXwTcA5zv/vtbRjmWH3cfhyb3ccn2WGZE5HYROeze/69EREaJeaLv0Vvcx6Hd/b652WPZR93HqllEXhCRWaMdK6WsNNK5x/18uJ+zR4zZvewSEakSkTvcx+KkiFwnIleJyCH3tr7hsY+JHstxnWvdMbaIyBKP5xwi0i0iGSIyTUSecR+nZvf9nFG29Y6rcCKyVkQOiEiriPwfIB7L5orIyyLSKCINIvKgiKS4lz0AzAT+7n497xCRPHGdi+0er9/ZPsP+MtprP0LcRkQ+Ja5zfLuI/MAd35si0ubeVpR73XT3MWhx7/tVEbF5xDTieyosGWP0NskbsACoBLLdj/OAue773wV6gHcBduD/AeXAN4FI4ONAuce2NgG/BmKAYqAeuNy97PvAW0AG4ADeAH7gXnYJUDUsrqF9XwVEAD8G3nIvswFbgW8DUcAc4BjwLvfyu4BXgVQgF9gzfPse+0kETgJfdsedCJzrXnY/8J8e656OcxzH7c/D9nO2YzOh4zxsu7OALiDJ/TjC/fec5358NTAX10lxtXvdZSMdd6ACWDOeYwjcAGS7X4sPAp1AlnvZLcBrw+I8fSyBy4AGYBkQDdwNbPZY1wDPACm4TtD1wLpR/v5xHzsgHmgDFrgfZwGL3fevA44Ai9zb+RbwhtX/n3rT2/DbOM494XzOHivmAXcMQ8eiHnjIvY3F7vjnTPJYjnquHeFvuA/4ocfjTwP/cN9PA94HxLnjegx40mPdV4CPue/fgvtcC6TjOr+93x3fF91/79C684C1uM65DmAz8EuP7VbgPv97vK8MYB/He2XU136Uv98ATwNJ7uPeC7zkfl8kA/uAj7jX/TGuBpZI9+1i9zE+63sqHG+WBxDMN/c/SB2wBogctuy7wAaPx+8BOoAI9+NE95s6BdcJbBBI9Fj/x8D97vtHgas8lr0LqHDfv4SRT6wbPR4XAN3u++cCJ4atfyfwR/f9Y3gkT8Btw7fvsexGYPsoy+5n9GR4rOP2Z4/HYx2bcR/nUeJ8Dfg39/21wNGzvN5PAp8f6bjzzmR43MfQvXwHcK37/i2cPRn+A/BTj2UJQD+Q535sgIs8lv8F+Poo+x33scOVDLfg+qCJHbad54FbPR7bcH2YzbLy/1Nveht+G8e5J5zP2WPF3D3CsTjXY/2twHUTPZajxHL6XDvCsjXAMY/Hr+M+h4+wbjHQ7PH4FUZOhv8NjwQUV8JYNbTuCNu9zvM4cpZkeBzvlVFf+1H2bYALhx33r3k8/jnuRB3XF5yngHnDtnHW91Q43rRMYgqMMUeAL+B6M9eJyCMel7wAaj3udwMN5l8F9d3unwm4WgmbjDHtHusfB2a472e7H3su89zPSE553O8CYtyXbGbhukTXMnTDVZqQ6bGvymH7Gk0urhPohIzjuHka69jA+I/zSB7C9QEBcJP7MQAicqWIvOW+vNSC65t7+ln/uH/FPOoxFJF/E5EdHsd/yTi3O7Tt09szxnQAjbzzeAx/7Uf722Gcx84Y04mrFft24KSIPCsiC93LZwH/4/H3NOH6MPGMSSnL6Tn7rOfssWJuHOFYDD9enueacZ+XJ3iufRmIFZFzxVWOVQw84d5OnIj8VkSOi0gbrhbcFBl7VId3HEPjyg5PP3aXYDwirvKaNuDPZ4lvpG2P9Rk22ms/muHHdrTX4We4rtq96C4/+br7+bHeU2FHk+EpMsY8ZIy5CNebywA/mcRmaoBUEUn0eG4mUO2xfNawZTVDIUxwX5W4Lk+leNwSjTFXuZefxHXC9NzX2bY1d5RlnbguVQ2Z7rnwLMdt+N8z1rGZqseAS9x1Ze/FnQyLq1bub8B/AZnGmBTgOTzqyM5i1GPoPnn/DvgMkObe7h6P7Y71er7jvSCuGsQ0vHc8RmWMecEYsxZXicQBXH8HuN4Hnxj2noo1xrzh65iUmig9Z496zj5bzD4z0XOtMcaJ64rXjbgaMJ7xSDS/jKsU5lxjTBKwamg3Y4TxjmMoIsI7j+mPcb1uhe7tfmjYNs/2mvr6M2xUxph2Y8yXjTFzcLXOf0lcNfJjvafCjibDUyAiC0TkMvc/cw+ub2QTHkrFGFOJqz7rxyISIyKFwK3Ag+5VHga+5e4okI6rzmdo+LFaIE1Ekse5uy1Am7g6kcSKSISILBGRoU4XfwHuFFdHhBzgs2fZ1jPAdBH5grg6NiSKyLnuZTuAq0QkVUSm42qNAcY8brVA3lCR/ziOzZQYY+pxXTr7I66Tw373oihc9WH1wICIXAmMdxiesx3DeFwnznpwdRzB1TI8pBbIGeoAMYKHgH8XkWL38fsR8LYxpmKcsU2KiGSKyDXu5LsX1yXPodfsHlx/72L3uskicoMv41FqMvScfdZz9tli9qXJnGsfwnWl6mY8rubhKr/oBlpEJBX4zjhjeBZYLCLXu1tkP8c7G3AScZ3zWkRkBvDVYb9fi6vu9gy+/gw7GxF5t4jMcyf3bbje64OM/Z4KO5oMT000rs4LDbguc2TgutQwGTfiqjOqwXXJ5zvGmA3uZf8JlAG7gN3ANvdzGGMO4DqJHXNf7jjrpTj3Zar34Lq0VO6O/fe4Cu8BvofrEk458CLwwFm21Y6rzvY9uP7+w8Cl7sUPADtx1VK9CDzq8atnO26PuX82isi2cRwbb3gIVx3a6ZOq+2/7HK4PmmZcLRBPj3N7ox5DY8w+XDVdb+I6gS7FVfM25GVgL3BKRBqGb9gY8xLwH7haUk7iauVZP864psKGq9WlBlcZxGrgU+6YnsDVuvaI+xLiHuBKP8Sk1ETpOXv0c/aoMfvSZM61xpi3cV19zMbVZ2HIL4FYXMfoLeAf44yhAVfH5rtwlZ3N553n5e/h6rTciitxfnzYJn6M64tEi4h8ZYRd+PozbDTzgY24Evk3gV8bY14Zx3sq7IirNEYppZRSSqnwoy3DSimllFIqbGkyrJRSSimlwpYmw0oppZRSKmxpMqyUUkoppcLW2QZ19rr09HSTl5fnz10qpZRXbN26tcEY47A6Dn/Sc7ZSKpiN97zt12Q4Ly+PsrIyf+5SKaW8QkTONrNXSNJztlIqmI33vK1lEkoppZRSKmxpMqyUUkoppcKWJsNKKaWUUipsaTKslFJKKaXClibDSimllFIqbGkyrJRSSimlwpYmw0oppZRSKmwFdDLc0z/IPZuO8vqRBqtDUUoppZSyTHVLN//Yc8rqMELSmMmwiNwnInUissfjuWIReUtEdohImYis9EVwkRE2frvpKH/bWuWLzSullFJKBYW7XzrMJx/cSntPv9WhhJzxtAzfD6wb9txPge8ZY4qBb7sfe12ETbh4voPNhxtwOo0vdqGUUkopFfC2VDRhDOyubrU6lJAzZjJsjNkMNA1/Gkhy308Garwc12mr8h00dPSy/1Sbr3ahlFJKKRWwGjp6OVbfCcDOSk2Gvc0+yd/7AvCCiPwXroT6gtFWFJHbgNsAZs6cOeEdrZqfDsCmQ/Uszk6eTKxKKaWUUkGrrMLVJmm3CbuqWiyOJvRMtgPdJ4EvGmNygS8CfxhtRWPMvcaYEmNMicPhmPCOMpJiWDg9kc2H6icZqlJKhZeR+np4LPuKiBgRSbciNqXUxG0pbyYm0saaRZnsqtKWYW+bbDL8EeBx9/3HAJ90oBuyOt/B1uPNdPYO+HI3SikVKu7nzL4eiEgusBY44e+AlFKTV1rRRHFuCiV506hu6aa+vdfqkELKZJPhGmC1+/5lwGHvhDOy1fkO+gcNbx5t9OVulFIqJIzS1wPgF8AduPp9KKWCQEfvAHtrWlmZl0pRbgqAlkp42XiGVnsYeBNYICJVInIr8HHg5yKyE/gR7ppgX1meN43YyAg2H9ZSCaWUmgwRuQaoNsbsHGO929xDZpbV1+s5VymrbTvejNPAitmpLM5Owiaws1KTYW8aswOdMebGURYt93Iso4q2R3D+3DQ2ad2wUkpNmIjEAd8ErhhrXWPMvcC9ACUlJdqCrJTFSiuaiLAJ58ycRlyUnfzMRHZq3bBXBfQMdJ5WzU/neGMXxxs7rQ5FKaWCzVxgNrBTRCqAHGCbiEy3NCql1Ji2lDdRkJVEQrSr/bIoJ4WdVS0Yo99VvSV4kuF810gUOqqEUkpNjDFmtzEmwxiTZ4zJA6qAZcYYndtVqQDWOzDIjsoWVuSlnn6uMDeZlq5+Kpu6LYwstARNMjw7PZ7c1Fg2HWqwOhSllApoo/T1UEoFmT3VrfQOOFk5e9rp54pyXJ3odmonOq8JmmRYRFg138GbRxvoG3BaHY5SSgUsY8yNxpgsY0ykMSbHGPOHYcvzjDHasqBUgNtS3gxAiUfL8ILpiUTbbdqJzouCJhkGV6lEZ98gW483Wx2KUkoppZRPlVU0MccRT3pC9OnnIiNsLM5O0sk3vCiokuEL5qZht4kOsaaUUkqpkOZ0GsqON7PSo1V4SGFOCrurWxkY1Cvl3hBUyXBiTCTLZk3TTnRKKaWUCmmH6tpp7e5/R+e5IcW5KXT3D3KkvsOCyEJPUCXD4JqNbm9Nm05FqJRSSqmQVVrumkRy5eyRWoaTAZ18w1uCLhleNd81xNprR7R1WCmllFKhaUtFM9OTYsiZFnvGsry0eBJj7Dr5hpcEXTK8ODuJtPgoNh3UZFgppZRSoccYQ2l5EytmpyIiZyy32YSinBR26fBqXhF0ybDNJlw0P51XDzfgdOrsK0oppZQKLVXN3Zxq62Fl3rRR1ynMSebAyXZ6+gf9GFloCrpkGFx1w42dfew72WZ1KEoppZRSXrXFXS+8YoR64SFFuSkMOI3mQl4QlMnwxe664U06qoRSSimlQkxpRRPJsZHkZySOus7pmei0E92UBWUy7EiMpiArSZNhpZRSSoWcLRVNlMyahs12Zr3wkOnJMWQmRevkG14QlMkwuGaj23a8mfaefqtDUUoppZTyioaOXo7Vd561RGJIYU6Ktgx7wZjJsIjcJyJ1IrLH47lHRWSH+1YhIjt8G+aZVuWnM+A0vHm00d+7VkoppZTyibIKd73wCJNtDFecm8Kxhk5au7VhcCrG0zJ8P7DO8wljzAeNMcXGmGLgb8DjPojtrEpmpRIXFaFTMyullFIqZGwpbybabmPpjOQx1x2afGNPtZZKTMWYybAxZjPQNNIycQ1+9wHgYS/HNaYou40L5qax6VA9xugQa0oppZQKfqUVTRTnphBlH7u9snCGqxPdDi2VmJKp1gxfDNQaYw57I5iJWpXvoLKpm4rGLit2r5RSSinlNR29A+ytaR1xCuaRJMdFMjs9XiffmKKpJsM3MkarsIjcJiJlIlJWX+/dkoahqZk366gSSimllApy24434zTjqxceUpiTzM5KLZOYikknwyJiB64HHj3besaYe40xJcaYEofDMdndjSgvPZ6ZqXGaDCullFIq6JVWNGETWDZr9JnnhivKSeFUWw+1bT0+jCy0TaVleA1wwBhT5a1gJmN1voM3jzXSN+C0MgyllFJKqSnZUt7E4uxkEqLt4/6dolxXJzodYm3yxjO02sPAm8ACEakSkVvdi9ZjQce54VblO+jqG6Ts+Ih9/JRSSimlAl7vwCA7KlsmVCIBsDg7mQib6OQbUzDmVw9jzI2jPH+L16OZhPPnpmG3CZsO1XPB3HSrw1FKKaWUmrA91W30DjhZOXv8JRIAMZERLMhMZKd2opu0oJ2BbkhCtJ3ls6ax+VCD1aEopZRSSk1KqXuyjZIJtgyDq1RiV1WrDjU7SUGfDIOrVGL/yTbq2rV4XCmllFLBp7S8iTmOeNIToif8u0U5KbR293Nch5qdlJBIhlfnu0apeFVbh5VSSikVZJxOQ9nxZlZOolUYoDDHNfmGlkpMTkgkwwVZSaQnROnUzEopBYjIfSJSJyJ7PJ77gYjsEpEdIvKiiGRbGaNS6l8O1bXT2t0/4c5zQ/IzE4iJtOl4w5MUEsmwzSZcPN/Bq4cbcDq1XkYpFfbuB9YNe+5nxphCY0wx8Azwbb9HpZQaUWm5q154vDPPDWePsLEkO1lbhicpJJJhgFX56TR19rGnRr8VKaXCmzFmM9A07Lk2j4fxgLYcKBUgtlQ0Mz0phpxpsZPeRlFuCntrWukf1HkXJipkkuGLdWpmpZQ6KxH5oYhUAjczSsuwiNwmImUiUlZfr+dTpXzNGENpeRMrZqciIpPeTmFOMj39Tg7VtnsxuvAQMslwekI0S2Yk6RBrSik1CmPMN40xucCDwGdGWedeY0yJMabE4XD4N0ClwlBVczen2npYmTex8YWHK3J3otPJNyYuZJJhgFXzHWw70UxbT7/VoSilVCB7CHif1UEopVxTMAOsmGS98JBZaXEkx0ayS+uGJyy0kuF8BwNOwxtHGq0ORSmlAoqIzPd4eA1wwKpYlFL/UlrRRFKMnfyMxCltR0QozElmh44oMWEhlQwvmzmN+KgIHWJNKRXWRORh4E1ggYhUicitwF0iskdEdgFXAJ+3NEilFABbKpooyUvFZpt8vfCQ4twUDtW209036IXIwofd6gC8Kcpu4/y56Ww+VI8xZkqF6EopFayMMTeO8PQf/B6IUuqsGjp6OVbfyQ3Lc72yvcKcFAadhr01rZOa1jlchVTLMMDqBQ6qmrspb+i0OhSllFJKqVGVVQyNLzy1znNDinKSAdipnegmJPSSYfcQa5t0iDWllFJKBbAt5c1E220snZHile1lJMWQlRzDzkrtRDcRIZcMz0yLIy8tTscbVkoppVRAK61oojg3hSi799KxopwUHVFigkIuGQbXqBJvHWuid0ALyJVSSikVeDp6B9hb0zrpKZhHU5ibTEVjFy1dfV7dbigbMxkWkftEpE5E9gx7/rMiclBE9orIT30X4sStmu+gu3+Qsopmq0NRSimllDrDtuPNOA2s8HJHN518Y+LG0zJ8P7DO8wkRuRS4Fig0xiwG/sv7oU3e+XPTiIwQLZVQSimlVEAqrWjCJrBslnc6zw1Z6u5Ep6US4zdmMmyM2Qw0DXv6k8Bdxphe9zp1Poht0uKj7ZTMStVOdEoppZQKSKUVTSzOTiYh2ruj3CbFRDLHEa+Tb0zAZGuG84GLReRtEdkkIitGW1FEbhORMhEpq6/3X3K6Kt/BgVPt1Lb1+G2fSimllFJj6Rtwsv1Ei9dLJIYU56Sws6oFY4xPth9qJpsM24FpwHnAV4G/yCgzXBhj7jXGlBhjShwOxyR3N3Gr8tMBtFRCKaWUUgFld3UrvQNOr40vPFxhTjL17b2c0gbBcZlsMlwFPG5ctgBOIN17YU3doulJpCdEs/lwg9WhKKWUUkqdVuqebMNXs8QV5bo60e3UUolxmWwy/CRwGYCI5ANRQEBlnTabsCo/ndcO1zPo1MsESimllAoMpeVNzHHEk54Q7ZPtL8pKwm4T7UQ3TuMZWu1h4E1ggYhUicitwH3AHPdwa48AHzEBWJiyOt9Bc1c/u6v1m5FSSimlrOd0GsqON7PSR63CADGRESzMSmSnJsPjMmYXRmPMjaMs+pCXY/G6i+alI+KqGy7O9c5Uh0oppZRSk3Worp3W7n6fdZ4bUpSTwtM7a3A6DTbbiN26lFtIzkA3JC0hmiXZydqJTimllFIBobTcVS/s7ZnnhivKSaG9Z4Dyxk6f7icUhHQyDK5RJbZXttDW0291KEoppZQKc1sqmpmeFEPOtFif7meoE53WDY8t5JPh1fkZDDoNbxwJqP59SimllAozxhhKy5tYMTuVUUak9Zp5GQnERUXoiBLjEPLJ8DkzU0iItutsdEop5Qe9A4NUNnVZHYZSAamquZtTbT2syPPN+MKeImzCkhnJ2oluHEI+GY6MsHHB3DQ2H2rQmViUUsrHbvrd23zx0R1Wh6FUQNrirhf2dee5IUU5yeytaaN/0OmX/QWrkE+GwTU1c3VLN0frtYhcKaV86YqCTMqON3O4tt3qUJQKOKUVTSTF2FmQmeiX/RXmpNA34OTgKf1/PJuwSIZX57umgdZRJZRSyreuX5aD3SY8UlppdShKBZwtFU2U5KX6baizoWFltVTi7MIiGc5NjWNOerzWDSullI85EqNZW5DJ49uq6B0YtDocpQJGQ0cvx+o7/VYiAZAzLZZpcZHsrNRk+GzCIhkGV6nE2+WN9PTryVkppXxp/cqZNHf18+LeWqtDUSpglFUMjS/s+85zQ0SEotwUdlXpiBJnE0bJcDo9/U5K3W9GpZRSvnHxvHRmpMTySOkJq0NRKmBsKW8m2m5j6Qz/zohbmJPCodp2uvoG/LrfYBI2yfB5c9KIirBp3bBSKuSJyH0iUiciezye+5mIHBCRXSLyhIj47BPZZhM+UJLL60caOa6zXykFuDrPFeemEGX3b+pVnJuM08Ce6ja/7jeYhE0yHBdlZ8XsaWw+pJNvKKVC3v3AumHPbQCWGGMKgUPAnb4M4AMrcrAJ/KVMO9Ip1dE7wN6aVp9PwTySwhydiW4sYZMMA6ya7+BgbTsnW7utDkUppXzGGLMZaBr23IvGmKHrpG8BOb6MISs5lksWZPBYWRUDOsapCnPbTzTjNP4bX9hTekI0M1Ji2aGd6EYVXsmwe4i1V7V1WCkV3j4KPD/SAhG5TUTKRKSsvn5qZWXrV+RS197LywfqprQdpYJdaXkTNoFls/zXec5TUW6ydqI7i7BKhhdOTyQjMZpNh7VuWCkVnkTkm8AA8OBIy40x9xpjSowxJQ6HY0r7unRhBo7EaB7VMYdVmNtS0cTi7GQSou2W7L8wJ4UTTV00dfZZsv9AN2YyPEpHjO+KSLWI7HDfrvJtmN4hIlw838FrhxsYdOrUzEqp8CIiHwHeDdxs/DA/fWSEjRuW5/DPg3VanqbCVt+Ak+0nWiwpkRhSpHXDZzWeluH7ObMjBsAvjDHF7ttz3g3Ld1YvcNDa3a9vCKVUWBGRdcDXgGuMMV3+2u8HV+TiNPBYWZW/dqlUQNld3UrvgNOv4wsPtzQnGRHYWamlEiMZMxkeqSNGMLt4Xjoi6Gx0SqmQJSIPA28CC0SkSkRuBf4PSAQ2uK/o3eOPWGalxXPhvDQeLa3EqVfkVBgamt+gxMKW4YRoO/McCdoQOIqp1Ax/xj1e5X0iMurXHW92xvCGafFRFM5I1vGGlVIhyxhzozEmyxgTaYzJMcb8wRgzzxiT63FF73Z/xfPBFTOpbunmtSPaeVmFn9LyJuY44klPiLY0jsKcFHZWteKHCqmgM9lk+DfAXKAYOAn8fLQVvdkZw1tW5TvYUdlCa1e/1aEopVTIe9fiTKbFReqMdCrsOJ2GsuPNrLSwVXhIcW4yDR291LT2WB1KwJlUMmyMqTXGDBpjnMDvgJXeDcu3VuU7cBp4/ai2UiillK9F2yO4flkOG/bV0tDRa3U4SvnNobp2Wrv7LS2RGHJ68g0db/gMk0qGRSTL4+F7gT2jrRuIzslNITHGrqUSSinlJ+tX5NI/aPjbVu1Ip8JHabmrXjgQWoYXZiUSGSHs0LrhM4xnaLWROmL8VER2i8gu4FLgiz6O06vsETYunJvOpkP1WjujlFJ+MD8zkeWzpvFoaaWed1XY2FLRTGZSNLmpsVaHQrQ9goKsJHbpiBJnGM9oEiN1xPiwMWapMabQGHONMeakP4L1plX5Dk629nCkrsPqUJRSKiysX5HLsYZOtpSHzABFSo3KGENpeRMr8lIREavDAVylErurW3Vkl2HCagY6T6vy0wEdYk0ppfzl6sIsEqPtPKIz0qkwUNXczam2HlbOtr5EYkhRbgodvQMca9CGQE9hmwznTItjjiOezYe1E51SSvlDXJSda8/J5rndJ3U0HxXyhq6AWDnz3HBFOcmATr4xXNgmwwCr8x28fayRnv5Bq0NRSqmwsH7FTHoHnDy5o9rqUJTyqdKKJpJi7CzITLQ6lNPmOBJIiLazUzvRvUNYJ8Or8h30Djh5W+vXlFLKL5bMSGbJjCQe3nJCO9KpkLaloomSvFRstsCoFwaIsAlLZiSxs0pbhj2FdTJ83uw0ouw2HWJNKaX8aP2KmRw41a4fyCpkNXT0cqy+M6BKJIYU5aSwv6aNvgGn1aEEjLBOhmOjIliZl6rJsKKrb8DqEJQKG9cWZxMbGcGjOiOdClFlFe7xhWdPsziSMxXlptA36OTAqTarQwkYYZ0Mg2tUicN1HdS0dFsdirKA02m46/kDLP3ui2w93mx1OEqFhcSYSK4uzOLpHTV09uoXURVo/n84AAAgAElEQVR6tpQ3E223sXRGitWhnKHwdCc6rRseEvbJ8Or8DABtHQ5DPf2DfPaR7dyz6SiDTsMzu2qsDkmpsHHjylw6+wb5+079v1Ohp+x4E8W5KUTZAy/NmpESS3pClJYpeQi8V8nP8jMTmJ4Uw+bDmgyHk6bOPj70+7d5dtdJ7rxyIZctzGDDvlrt0KOUnyybOY35GQk8rGMOqxDT2TvA3pq2gBpf2JOIUJiTwq4gGVHi1cP1Ph/1K+yTYRHh4vnpvHa4gYFBLSYPB+UNnVz/69fZVd3Kr25axidWz2VtQSZVzd0crG23OjylwoKIsH7lTHZWtrD/pNYuqtCx7UQzg04TkJ3nhhTlpHC4roOOAC9TOlrfwS1/LOV/Xzrs0/2EfTIMriHW2noG9JJBGCiraOL6X79OW88AD3/8XK4uzALg8oWucpmN+2qtDE+psPLec2YQFWHjUW0dViGktLwJm8CyWYHXeW5IYW4yxsCe6sDOe37y/AFi7DY+etFsn+5Hk2HgonnpiGjdcKh7ZlcNN/3+bVLionj8kxewfNa/vrVnJMVQnJvCBk2GlfKb1Pgo3rVkOo9vq9LJj1TI2FLRxOLsZBKi7VaHMqqiHFfHvkDuRFda0cSL+2q5ffVc0hOifbovTYaBafFRFOWksEmT4ZBkjOE3rxzlMw9tpygnmcc/eQF56fFnrLe2IJOdVa3UtvVYEKVS4enGFbm09Qzw/J6TVoei1JT1DTjZfqIloEskwPVFNDc1ll0BekXcGMOPnttPZlI0H7t4js/3p8mw26p8B7uqWmjp6rM6FOVFA4NOvvHEHn7yjwO8pyibB249l2nxUSOuu2ZRJgAv7a/zZ4hKhbXz5qQxKy2OR7ZoqYQKfrurW+kdcLIiL3BLJIYU5qSwI0Bbhp/fc4rtJ1r40tp8YqMifL4/TYbdVuen4zTw2pEGq0NRXtLRO8Ctfyrj4S0n+NQlc/mfDxYTEzn6P1V+ZgIzU+PYuF9LJZTyF5tN+EBJLm+XN3GsvsPqcJSaklL3ZBslAd4yDFCck0J1SzcNHb1Wh/IOfQNOfvKPA+RnJvD+5bl+2eeYybCI3CcidSKyZ4RlXxERIyLpvgnPf4pyUkiMsWvdcIg42drNDfe8yWtHGrjr+qXcsW7hmPPDiwhrFmXy2pEGnQhAKT+6YXkOETbRjnQq6JWWNzEnPR5Hom9rXL1haPKNQBti7aG3j3O8sYs7r1xExBif294ynpbh+4F1w58UkVxgLRAS82naI2xcNC+dzYcadKzZILe3ppXrfvU6lU1d3HfLCtavnDnu311TkEHfgJNXD+sVAqX8JSMphssXZvDXrVX0DegQlyo4OZ2GsuPNAV8vPGTJjGRsAjsrA6duuK2nn/956TAXzE3jkgUOv+13zGTYGLMZaBph0S+AO4CQyRxX5zs41dbDoVq9VBes/nmwjg/c8yY2ER67/XxW50/sn2lFXipJMXYtlVDKz25cOZPGzj5e0v89FaQO1bXT2t3PigCdbGO4+Gg78zMSA6pl+J5XjtLc1c+dVy5CxD+twjDJmmERuQaoNsbsHMe6t4lImYiU1dcHdgnCKnfipKUSwemht0/wsT+VMSstnic+dSGLspImvI3ICBuXLczg5QN1DDpD5nueCjMjlbeJyA0isldEnCJSYmV8I1mV7yArOUZnpFNBq7Tc1W64MkhahgGKcpPZWdUaEFfEa1q6+cNr5VxXnM1SdwmHv0w4GRaROOCbwLfHs74x5l5jTIkxpsTh8F+T92Rkp8QyLyNBp2YOMk6n4a7nD/CNJ3azan46f7n9fKYnx0x6e2sKMmnq7GPbiWYvRqmUX93PmeVte4Drgc1+j2YcImzCDSW5vHq4nsqmLqvDUWrCtlQ0k5kUTW5qrNWhjFthTgpNnX1UNXdbHQr/veEQxsCXr1jg931PpmV4LjAb2CkiFUAOsE1EpnszMKusmu/g7fImuvt0APhg0NM/yGcf2c49m45y87kz+d2/lUx5oPPV+Q4iI0Rno1NBa6TyNmPMfmPMQYtCGpcPlOQA8NjWKosjUWpijDGUljexIi/Vr5f3p+r05BsWl0rsq2njb9uquOXCPHJT4/y+/wknw8aY3caYDGNMnjEmD6gClhljTnk9Ogusyk+nb8DJ2+WNVoeixtDU2cfNv3+bZ3ed5M4rF/Kf1y3BHjH10QITYyI5b04aG7R2UYUhK0vbcqbFsWq+g8fKKrVMSQWVquZuTrX1sDJI6oWHLJieSJTdZvnkG3f94wBJMZF8+pJ5lux/PEOrPQy8CSwQkSoRudX3YVnnvDlpRNttOhtdgCtv6OT6X7/O7upWfnXTMj6xeq5Xv42vLcjkWH0nR3XcUxVmrC5tW78il5OtPWw6pJPfqOCxxV0vHCwjSQyJstsoyEqydPKNVw/Xs/lQPZ+9bB7JcZGWxDCe0SRuNMZkGWMijTE5xpg/DFueZ4wJmXGoYiIjWDk7VTvRBbCyiiau//XrtPUM8PDHz+Xqwiyv7+Ny92x0WiqhlH9dviiT9IQoHtYZ6VQQKa1oIinGzoLMRKtDmbDi3BT2VLdacjXG6TT86LkD5EyL5cPnz/L7/ofoDHQjWJ3v4Gh9J9Ut1heUq3f6+84abvr926TERfHEpy5g+SzffAufkRJLQVaSDrGmlJ9F2W28b3kOLx+oo66tx+pwlBqXLRVNlOSljjm5UyAqzEmmq2/QkiuhT2yvZv/JNr76rgVE230/7fJoNBkegQ6xFniMMfzmlaN89uHtFOUk8/gnL2BWWrxP97m2IJOtx5tpDLCpKpUay0jlbSLyXhGpAs4HnhWRF6yNcnTrV8xk0Gm0I50KCg0dvRyr7wy6EokhRbmuTnT+LpXo6R/k5y8eZOmMZN5TmO3XfQ+nyfAI5mckMD0phk0HNRkOBP2DTr7xxG5+8o8DvKcomwduPZdp8VE+3+/agkycBl4+oLWLKriMVN5mjHnCfT/aGJNpjHmX1XGOZnZ6POfOTuXR0kqc2pFOBbiyCtcwnCtnT7M4ksmZnRZPYrTd75Nv/PH1Cmpae7jzqoWWt6hPbQyqECUirM538Nyek/QNOBlwOuntd9IzMEhPv5PeoZ/9g/QMjPyzd8BJj+dP9+//aztDy1zb6/Xc7sAgIkLutFhmpcUzMzWOWWmu28zUeHJTYy29nOBP7T39fPqh7Ww+VM+nL53Ll9cu8Ns/zeLsJLKSY9i4v5YbSnL9sk+llMuNK2fyhUd38OaxRi6cl251OEqNqrSiiWi7jaUzUqwOZVJsNmFpTrJfp2Vu6uzj1/88wmULM7hgrvX/35oMj2JVvoNHyyrJ/9bzU9pOTKSNmMgIou1n/kyItpMWbyN6hOUDg04qm7qpaOzkrWONdHmMeywC2cmxp5PkmWlxzEqNP30/Kcaa3pjedrK1m3//YymH6zq46/qlrF8506/7FxHWLMrkr1ur6OkfJCYyPL6AhLKO3gGe2FbFe4qySYnz/dUFNXnrlkwn+elIHimt1GRYBbTSiiaKc1OIsgfvxfai3BR+/+oxv33W3f3yYTr7Bvj6lQt9vq/x0GR4FGsKMvjquxbQP+gcMZmNibQRbT/zZ7TH46gIm1eG+zLG0NDRx4mmTo43dnG8sYsTTV1UNHayYV8tjZ1971g/NT7qX63JqXHMTIs/fd+RGB0UA4LvrWnlo/eX0tk7yB9vWXG6jtvf1hRk8sBbx3njaAOXLcy0JAblPS/uPcV/PLWXRVlJlARpfV+4iImM4L3nzOCht0/Q1NlHqh9Ko5SaqM7eAfbWtPGpS+ZaHcqUFOUk0z9o2H+yjXNm+rbc43hjJ39+6zgfKMklP0BG39BkeBTR9gg+fak1gz8PJyI4EqNxJEaPOHpCe08/J5q6ONHYxfGmLnfC3ElZRTN/31mDZ8ldXFQEM1PjPFqV45nlvj8jJdYrk1ZM1T8P1vGZB7eRFBvJY7efz6KsJMtiOW9OKvFREWzYV6fJcAh4ckcNOdNiWT4rOGv7ws36lbnc/0YFj2+r4mMXz7E6HKXOsO1EM4NOE/Rfroc60e2qavV5MvzTFw5it9n44tp8n+5nIjQZDgGJMZEszk5mcXbyGcv6BpxUNbuS5BON/0qUjzV08sqhevoGnKfXtduEGdNiyUiMJiYygrioCGIjI4iNshM79Nj9nOf92CjXY9fv2D3uRxA5weT6wbeP8+2n9rIgM5E//vsKMpNipnx8piLaHsHqBQ5e2l+L07nE8iJ/NXn17b28drieT17i3QlalO8snJ5EcW4Kj5RWcutFs/V1UwGntLwJm8CymcFZLzxkelIMjsRon0/LvP1EM8/uOsnnLptn+ee7J02GQ1yU3cYcRwJzHAlnLHM6DbXtPa6yi8YujrvLMJo6+2jvGaC+vZeuvkG6+wfp7hukq2+AiXbsttvkHQn0UJIcF2UflnBH0NzVx1M7arh0gYO7b1pGQnRgvD3XFmTy3O5T7K5uPf3tWQWfZ3a5rpJcVzzD6lDUBNy4Mpev/W032040+2xccaUma0tFEwXZSSQGeV8dEaEoJ4WdPhxezRjDj587QHpCFLetDqyyksDINpQlbDYhKzmWrORYzpuTNub6xhj6Bp10uxPkrr7B0/ddybJrlIx/JdADZ6znuU5TZ987Eu2+QScfOX8W//HugoAo1xhy6YIMImzChn21mgwHsSd31FCQlcT8AKlRU+Pz7sJsvv/3fTy8pVKT4QBzpK6D2KgIZqTEWh2KJfoGnGw/0cJN5/q3c7evFOUk89KBWtp6+n3SEX/j/jq2VDTxg+uWBExj15DAikYFNBFxdRK0RxBOKWFKXBQls6axcX8tX3nXAqvDUZNQ3tDJzsoWvnFVYPRcVuMXH23nmuJsnthezbffUxAyo+UEs+0nmvnVP4+wcX8dM1Ji2fil1cRGhd9oO7urW+kdcLIyyOuFhxTlpmAM7Klq5QIvj+AyMOjkruf3M8cRz/oVgTdUaeA0vykVwNYWZHLgVDuVTV1Wh6Im4akd1YjANUVaIhGM1q+YSU+/k6d21FgdStgyxvDG0QZu/v1bvPfXb1B2vJmbz51JdUs3v3nliNXhWaK0ogkg6DvPDSnMcfU72lnl/fGGHy2r5Gh9J19bt3DCfYn8IfAiUioArS1wjSSxYV+txZGoiTLG8NSOGs6bncb05MDpsKHGrzAnmUVZSTxaesLqUMKOMYaXD9Tyvt+8wU2/e5tDtR1886pFvP61y/jhe5dyXXE292w+xvHGTqtD9bvS8ibmpMfjSIy2OhSvSImLYlZanNfrhjt6B/jFhsOsyJvGFQWBOSqTJsNKjcOstHjmZySwcb8mw8FmV1Ur5Q2dXHdOttWhqEkSEW5cmcue6jb2VPtvlqxwNug0PLOrhqv+9zU+en8ZtW29/OC6Jbx6x6V8fNUc4t01n3detYhIm/CDZ/ZZHLF/OZ2GsuPNrAiRVuEhRTkpXp+W+Xebj9HQ0cudVy0K2BFhNBlWapzWFGTydnkTrV39VoeiJuDJHdVERdhYtyTL6lDUFFxbNINou42Ht2jrsC/1Dzp5rKyStb/YxGce2k7vwCD/dUMRr3z1Ej583qwzZifLTIrh82vms3F/HS8fCJ/GgkN17bR297Nidmglw4U5ydS09lDX3uOV7dW19XDv5mNcvTSLZT4ev3gqxkyGReQ+EakTkT0ez/1ARHaJyA4ReVFEtMlFhby1BZkMOg2vHKqzOhQ1TgODTv6+8ySXLnSQHKsdr4JZclwkVy/N4ukdNXT1DVgdTsjp6R/kgTcruORnr/DVv+4i2h7Br25axoYvrub9y3POWud5ywWzmeuI53t/30dP/6D/grZQabmrXjhUOs8NKR6afKPSO1dgfrHxMANOJ18N8M7n42kZvh9YN+y5nxljCo0xxcAzwLe9HZhSgaY4J4X0hCitGw4ibxxtpKGjV8cWDhHrV86kvXeAZ3edtDqUkNHZO8C9m49y8U//yX88tZfMpGjuu6WE5z53EVcXZhExjomGouw2vnvNYo43dvGH18r9ELX1tlQ0k5kUTW5qaA0rtzg7mQibeKVU4nBtO4+WnuDmc2eRlx7vheh8Z8yh1Ywxm0Ukb9hzbR4P44EJTsWgVPCx2YTLF2by3O6T9A04ibJrlVGge3JHNYkxdi5dmGF1KMoLVuRNY44jnkdKK7mhJPCGZwomrV393P9GBX98o5yWrn4unJfG/6wv5vw5aZOq67x4voMrl0zn7pcPc905M0J67GFjDKXlTazISw3YGtjJio2KID8zkR1eGFHiJ/84QHyUnc9dPt8LkfnWpD/NReSHIlIJ3MxZWoZF5DYRKRORsvr6+snuTqmAsKYgk/beAba4L5GpwNXdN8gLe05x5ZLpZ9Q5quAkIqxfkcvW480cqm23OpygVN/ey13PH+DCn7zMLzYeomRWKk986gIe/Nh5XDA3fUrJ3TevXgTAj57d761wA1JVczen2npYGWL1wkOKcpLZVdWCMZNv53zrWCMb99fxyUvnkhof5cXofGPSybAx5pvGmFzgQeAzZ1nvXmNMiTGmxOFwTHZ3SgWEi+alExNpY8O+U1aHosawcX8tnX2DWiIRYt63LIfICOHR0kqrQwkq1S3dfOepPVz0k5e5d/NRLl2YwfOfv5jff6SEc7zUsSlnWhyfvmQez+4+yetHGryyzUA01BgSaiNJDCnMSaGlq58TkxxX3+k0/Oi5/WQlx/DRC2d7OTrf8MZ13oeA93lhO0oFvNioCC6a52Dj/ropfWtWvvfUjhoyk6I5dxxTjavgkZYQzRUF03l8WxW9A+HRWWsqyhs6ueOvO7nkZ//kwbdPcG1xNi99+RLuvvEcFmUleX1/H181h5mpcXzn6b30Dzq9vv1AUFrRRFKMnQUhOrV7Ue7UJt94ZvdJdlW18uUrFgTNVblJJcMi4lkAcg1wwDvhKBX41hZkUN3Szf6Tepk2ULV09bHpUB3XFGWPqwOQCi7rV+bS3NXPC3u1M+toDpxq47MPb+fyn7/CUztquGnlTDbdcSk/fX8Rs33YmSkmMoLvvKeAI3Ud/OmNCp/tx0qlFU2U5KViC9FzS35mItF2G7smMflG78AgP3vhAAunJ/Lec4LnqtyYHehE5GHgEiBdRKqA7wBXicgCwAkcB273ZZBKBZLLFmYispuN+2spyPZ+y4qaumd3n6R/0HCtlkiEpAvnppMzLZZHtpzgmiId2dPTjsoW/u/lI2zcX0t8VAQfXzWHj100x6+zpF2+KJPLFmbwy42HuaYom4yk0Jn5sbGjl6P1nbx/eeh24IyMsLFkRjI7JzGixANvHqeyqZv/99GVQdUQMWbLsDHmRmNMljEm0hiTY4z5gzHmfcaYJe7h1d5jjKn2R7BKBQJHYjTn5KboEGsB7KntNczLSGCxflkJSTab8MGSXN442hiW0wAPZ4zhzaONfOj3b3Pdr16n7HgTX1yTz+tfv4w7r1xkyXTB3353AX0DTu56PrQuHN/98hEALpwX2uVXhTnJ7KluY2ACpS6tXf3c/fIRLp6fzqr84OojpmNDKTUJawoy2V3dysnWbqtDUcNUNXexpaKJ64qzQ27YI/UvN5TkYhPCviPd0foO3n/Pm9z4u7c4WNvON65ayGtfu4zPr5lPSpx1vfjz0uO5bdUcHt9eTWlFaIy+88T2Ku5/o4JbL5pNYU6K1eH4VHFuCt39gxyu6xj37/z6lSO09fTz9SsX+jAy39BkWKlJWLsoE4CX9utsdIHm6Z01AFoiEeKmJ8dw2cIMHttaFbIdtcbju0/v5XBtOz+4djGv3nEpt62aS0L0mBWQfvGpS+eSnRzDfzy5Z0ItjIFoT3UrX//bbs6dnRqUyd5EDSX74518o6q5iz++UcF7z5nB4uxkX4bmE5oMKzUJ8zISyEuL01KJAPTU9hqWz5pGbmqc1aFYRkTuE5E6Ednj8VyqiGwQkcPun94ZT8tCH1wxk/r2Xl4+EJ5fSvfWtPLq4QZuv2QuHz4/L+B67sdF2fnWuws4cKqdh7acsDqcSWvu7OP2P28lNT6KX9287KxTU4eKvLQ4kmLs7BjntMw/f/EQAF+5IrCnXR5N6L+iSvmAiLBmUSZvHm2ko3fA6nCU2/6TbRysbee64rDvVHU/sG7Yc18HXjLGzAdecj8OapcucJCZFB22pRL3bj5GfFQEN587y+pQRnXlkulcOC+N/3rhII0dvVaHM2GDTsPnHtlOXVsvv/nQctIT/F9/bQURoSg3ZVwtw3uqW3liezUfvXA22UE686Amw0pN0pqCTPoGnbx6SGdWDBRP7qjGbhOuLgzvZNgYsxkYXqh5LfAn9/0/Adf5NSgfsEfYuGF5Lq8crAu7+v2q5i6e2XWSm86dSXJspNXhjEpE+N41i+nqG+RnLxy0OpwJ+9kLB3n1cAM/uG4xxbmhXSc8XGFOMgdPtdPTP/p43sYYfvz8fqbFRfKpS+f6MTrv0mRYqUkqmTWNlLhINuzXUolA4HQa/r6jhlX5jqCY/tMCmcaYkwDunxkjrSQit4lImYiU1dcH/he9D67IxWngL6VVVofiV394rRwBPnpR4M/wNS8jkY9eNJtHyyrZMYmxa63y/O6T3LPpKDedO5MPrphpdTh+V5STwoDTsLembdR1Nh2q5/UjjXz2svkkxQTul7KxaDKs1CTZI2xctiCDlw/UBX3nkFCwpaKJmtYertUSiSkxxtxrjCkxxpQ4HIE/PFJuahwXzUvnL2WVDDrDY1bI5s4+HtlSybXFM8hKDo7L0p+9bB6OhGi+89QenEHwOh2ubecrj+3knJkpfOc9BVaHY4mi3LN3oht0Gn783AFmpsbxofMCt1RnPDQZVmoK1hRk0tLVz9bjzVaHEvae2lFNXFQEawsyrQ4lUNWKSBaA+2fI9DpbvzKX6pZuXjvSYHUofvHnt47T3T/IbavmWB3KuCXGRPKNqxaxs6qVx7YGdo13W08/tz2wldgoO/d8aDnR9sDqmOgvmUkxTE+KYecorfl/21bFwdp27li3gCh7cKeTwR29UhZble8gKsLGRi2VsFTvwCDP7jrJFQWZxEUFxrBSAehp4CPu+x8BnrIwFq9aW5BJanwUjwTxiAXj1dM/yP1vVHDpAgcLpidaHc6EXFuczYq8afzkHwdp7eq3OpwROZ2GLz26g8qmLn598zIyQ2j2vMkozElmV9WZI0p09w3y8xcPUpSbwtVLsyyIzLs0GVZqChKi7Zw/N40N+2oxJvAv/YWqVw7W09YzwLXn6NjCACLyMPAmsEBEqkTkVuAuYK2IHAbWuh+HhGh7BNefM4MN+2qpbw++EQsm4q9bq2js7OMTq4Ovs5KrM90SWrr6+O8NgdmZ7u6Xj7Bxfx3funoRK2enWh2O5YpyUzjW0Elr9zu/vNz3ejm1bb1886pFITG5kSbDSk3RmoJMKhq7OFo//pl6lHc9taOatPgoLp6XbnUoAcEYc6MxJssYE2mMyTHG/MEY02iMudwYM9/9MzSmBXNbvzKXAafh0dLQbR0edBp+9+oxinJTODdIE7WC7CQ+fN4sHnjrOPvO0jHLCi8fqOWXLx3i+nNm8JEL8qwOJyAUuSff2O3ROtzQ0ctvXjnK2oLMkPnCoMmwUlO0ZpGrU/6GfSFTghlU2nr62bi/jncXZmEPg8Hw1cjmZSRy2cIM7t18jJauPqvD8YkX9p7ieGMXt6+aE9StcV9au4CUuCi+8/SegLmiVt7Qyecf2UFBVhI/un5pUB9fb1qa45pNbqdHJ7q7XzpMd/8gX1sXOjPx6SeHUlOUlRzLkhlJWjdskRf2nKJvwKklEoo71i2gvXeA37xy1OpQvM4Yw283HSUvLY4rFk+3OpwpSY6L5GvrFlBa0cxTO2qsDofO3gE+8UAZdptwz4eWB9xMflZKjo1kTnr86U50x+o7ePDtE6xfkcu8jASLo/MeTYaV8oK1i6az7URzyNcrBqKndtQwKy2Oc8JsQHx1poXTk7j+nBz++EYFNS2hNQnHW8ea2FnVysdXzSHCFvytljcsz6UoJ5kfPref9h7rOtMZY7jjb7s4UtfB3TcuC+tp3Efj2YnuZy8cJMpu4/Nr5lsclXdpMqyUF6wpyMAY+OcBLZXwp7q2Ht442sC1Rdl6WVMB8KUr8gH45cZDFkfiXb/dfJT0hCjetyzH6lC8wmYTvn/tEho6ern75SOWxfG7V4/x7K6T3LFuIRfN1z4HIynKTeFUWw/P7z7J83tO8YlVc8lIDK1RNsZMhkXkPhGpE5E9Hs/9TEQOiMguEXlCRLRJRoW1gqwkZqTE6mx0fvb0zhqcBi2RUKfNSInl386bxV+3VnGott3qcLxi/8k2XjlYzy0X5IXUJfyi3BQ+WJLLfa+Vc9iC1+r1Iw3c9fwBrl6axSeCaMxmfyt0d6K746+7cCRG87GLA3/Ww4kaT8vw/cC6Yc9tAJYYYwqBQ8CdXo5LqaAiIqxZlMGrh+vp7ht9HnflXU/tqGHpjGTmOkKndk1N3acvnUd8lJ2f/iMwh++aqN9tPkZcVETQz/I1kq++awFxURF89+97/dqZrqq5i888tI25jgR++v5CvbJ0Fouzk7DbhPbeAb60Np/46NAby33MZNgYsxloGvbci8aYAffDt4DQuG6j1BSsKcikp9/J62EyC5bVjtZ3sLu6VadfVmeYFh/F7ZfMZeP+WkorgnsEueqWbp7eWcP6FTNJiYuyOhyvS0uI5ivvWsDrRxp5fs8pv+yzp3+Q2/+8lYFBw28/vDwkkztviomMYMmMZOZnJHDD8tBM97xRM/xR4PnRForIbSJSJiJl9fX1XtidUoHp3NlpJETbdVQJP3lqezU2gWuKNBlWZ/rohbPJSIzmJ88fCJjhuybjvtfKMcCtIXhpeshNK2eyKCuJ/3xmH119A2P/whQYY/jmE3vYU93GL9cXM0evKo3LvR9ezr/EiI8AABhxSURBVMO3nReyw1dO6a8SkW8CA8CDo61jjLnXGFNijClxOBxT2Z1SAS3KbmP1Agcb99fhdAbvh28wMMbw5I4aLpibTkaYT5eqRhYbFcEX1uRTdryZjfuDs2Nra1c/D285wTVF2cxIibU6HJ+xR9j4/rWLqWnt4df/9O2weA+8dZy/bavi85fP5/JFmT7dVyjJSIohPSHa6jB8ZtLJsIh8BHg3cLMJ5q/dSnnRFQWZNHT0ssNjgHLlfdsrWzjR1KUlEuqsPlCSw5z0eH76jwMMDDqtDmfC/vz2cbr6BrktDDp3rchL5b3nzODezceoaOj0yT5KK5r4/t/3cfnCDD5/eWgNDaamZlLJsIisA74GXGOM6fJuSEoFr0vyM4iwCRv3aamELz21vZpou411S4J78gHlW/YIG3esW8Dhug4e31ZtdTgT0tM/yB9fr2B1voNFWUlWh+MXd165kCi7je8/s8/r265t6+FTD24jNzWO//5gMbYQGKtZec94hlZ7GHgTWCAiVSJyK/B/QCKwQUR2iMg9Po5TqaCQHBfJyrxUrRv2of5BJ8/sOsmaRZkkxkRaHY4KcO9aPJ3i3BT+e8MhevqDZ6SXJ7ZX09DRG1ZDfmUkxfCFNfN5+UAdL3nxHNo34OSTf95KZ+8Av/3wcpJj9byh3mk8o0ncaIzJMsZEGmNyjDF/MMbMM8bkGmOK3bfb/RGsUsFgbUEmh2o7ON7om0t94e61Iw00dvZpiYQaFxHh61cu5FRbD396o8LqcMZl0Gn43eZjLJ2RzPlz06wOx68+ckEe8zIS+N7f93nty8v3n9nLthMt/Oz9ReRnJnplmyq0hGa3QKUstMbdKWODlkr4xFPbq0mOjeSSBRlWh6KCxHlz0rh0gYNf/fMIrV3WTf07Xhv21XKsoZNPrJ4TduPfRkbY+N41iznR1MXvNh+b8vb+UlbJn986wSdWz+HqwiwvRKhCkSbDSnnZzLQ4FmQmaqmED3T1DfDivlquWppFlF1PX2r87li3kPbeAX69ybqpf8fDGMM9m44yMzWOdYvDsyb+wnnpXL00i1+9coSq5sl3S9pV1cK3ntzDhfPS+OoVC7wYoQo1+mmilA+sKcigtKKZlq4+q0MJKRv21dLVN8h1WiKhJmhRVhLvPWcG979ewcnWbqvDGVVpRTM7Klv4+MWzQ3ZM1/H4xtWLEIQfPrt/Ur/f0NHL7Q9sxZEQzd03LgvrY6nGpu8OpXxgbcF0Bp2Gfx4MzvFNA9WT26vJTo5hRV6q1aGoIPSltfkYA7/ccNjqUEb1201HSY2P4v3Lc60OxVIzUmL5zGXzeH7PKV49PLEJuwYGnXzmoW00dvbx2w8vJzU+9GbuU96lybBSPlA4IxlHYjQb92ky7C2NHb1sPtzAe4qzdVgk9f/bu/PoKqt7jePfX+YBEhJCAiQhhBADYVIJYFRAGRRnvXWgVy2r2kJba7Wtq7a37bXtutdWb+tt7+0krUO11qFqQauiiApiZSbIPIUxCUkQTIKQkGHfP5JyUWQInHPec/I+n7VYnJyc877PXsnZ+Z199rv3aclJS+LW0jz+unwXm6sbvI5zjE3VDczbUMO00v4kxkV7HcdzXxqbT/+eSfzopbUcbjn1daJ/9toGFpXv4/7rhjE0OzWICaWrUDEsEgRRUcakwZnM31RLU0vkLOcUzl5dXUVrm+Pas7O9jiIR7I6LB5IcF8ODr2/0OsoxZi4oJzE2mi+U5nkdJSzEx0Rz31VD2Fr7MY//Y9spPWd2WQV/XLiNaaV5fG5kTpATSlehYlgkSCYXZ3GgqYXF5fu8jtIlzCqrpCiru282IJDgSE+OY8b4AcxdV82y7eHz2qyqO8TssgpuGpVLmj7WP+LiQZlMGpzJr97cTHV94wkfu76qnntf+IBR/dP4wZXFIUooXYGKYZEgOb8gg8TYaC2xFgC79h1k+Y79XHOOLpyTM3fbhfn06h7PA3M24JzzOg4Aj723nTYHt1+Y73WUsPPDK4tpbnPc/+rxL6b76OBhZjy5nNTEWH5z87nE6oI56QT9togESUJsNGMLM3hzfXXY/MGNVLPL2rfSvXqEimE5c0lxMdw9qZCl2/czb7338/rrDjXzl8U7uWJYH3LTk7yOE3byeibzlXEDmF1WyeLyD4/5fmub465nyqiqO8Rvbx5JZvcED1JKJFMxLBJEk4qzqKprZG1lvddRIpZzjllllYzun05OmgoFCYwbS3IZkJHMA3M20Nrm7ZvVvyzeyYGmFqb7aOvlzvrqRQPJ7pHIfS+tpaX1kxfT/fLNTczfVMt9Vw1hZF6aRwklkqkYFgmiiYMyMdNudGdibWU9W2oOaIpEAJjZXWa2xszWmtndXufxUmx0FPdcWsTmmgO8sGK3ZzmaWlp59L1tjC3M0MoHJ5AYF80PrxzMhj0N/HnRjiP3v752D//71hZuLMnh5jH9PEwokUzFsEgQ9ewWz8h+adqN7gzMLqsgNtq4Ypi2Uj0TZjYU+DIwGhgBXGlmhd6m8tZlQ3szIrcH/z13E43N3qz6MntlJbUNTcwYV+DJ+SPJpUN6M7Ywg1/M3cTeA01sqTnAt59bxYicVH5yzVDfbV0tgaNiWCTIJhVnsbaynsqPwnfXq3DV2uZ4aVUl48/KpEeSrrA/Q4OBRc65g865FmA+cJ3HmTxlZnx3yiCq6hp54v3tIT9/W5vj4QVbKe6TwgUDe4b8/JHGzLjvqiE0Nrfy45fXMePJZcTHRPG7W0aSEKt1meX0qRgWCbLJxVkAzNPocKctLv+Q6vomrtUUiUBYA4wzs55mlgRcDhyzzZmZTTezZWa2rLa2czt/RaLSgp5cVNSL37y9lbqDzSE997wNNWyt/ZgZ4wdoVPMUDczsxm0X5vPyqkq2f3iQX//rufTtkeh1LIlwKoZFgqygVzcGZCTzhuYNd9qssgq6xccwaXCW11EinnNuPfAAMBeYA6wCWj7jcTOdcyXOuZJevXqFOKU3vnPpIOobm/nd/K0hPe/D87eS3SNRU4A66c4JhYzOT+cn1wyhtEAj6nLmTloMm9mjZlZjZmuOuu+Gjgsw2sysJLgRRSLfpOIsFpV/SENjaEeeIlljcyuvrd7DpUN66yPQAHHOPeKcO9c5Nw7YB2z2OlM4KO6bwrVnZ/PYe9vYU3fijR0CZdn2fSzbsZ8vj80nRmvidkq3+Biem1HKzWO0U58Exqm8Ah8HpnzqvjXAvwALAh1IpCuaNDiL5lbHgk17vY4SMd7eUENDU4umSASQmWV2/N+P9j78aW8ThY9vTT4L59qX6QqFhxeU0yMplhtHHTNTRURC7KTFsHNuAe0jCEfft945F34bu4uEqZF5aaQlxTJ33R6vo0SMWWUV9Ooez/kFGV5H6UpeMLN1wMvAHc65/V4HChe56Unccl4ezy3bxZaahqCea0vNAeauq+YLpf1JiosJ6rlE5OT02YxICERHGRMGZfHWhhqaP7VgvByr7mAzb2+o5arhfYmO0oVFgeKcG+ucK3bOjXDOzfM6T7j5+oSBJMXF8OCc4I71/GFBOfExUUwr1cf8IuEg6MWw365MFjmeycWZ1De2sGy7BuNO5rU1VRxubdMUCQmp9OQ4ZowbwBvrqlm+Iziv0+r6Rv62soIbS3Lp2S0+KOcQkc4JejHsxyuTRT7L2MJexMVEaQOOUzCrrIIBGckM045cEmK3j80no1s8D7y2AecCv03zY+9tp6WtjS+NzQ/4sUXk9GiahEiIJMfHcEFBT+auqw7KH9muoqruEIu37ePqs/tq7VUJuaS4GO6eVMiS7ft4a0NNQI/d0NjMU4t2cNmwPuT1TA7osUXk9J3K0mpPA+8DRWa228xuN7PrzGw3UAq8YmavBzuoSFcwqTiLnfsOsrnmgNdRwtbLqypxDq49O9vrKOJTN43KJT8jmQfmbKC1LXBvXJ9espOGphZmjBsQsGOKyJk7ldUkPu+c6+Oci3XO5XSsU/m3jtvxzrks59yloQgrEun+uXnEXG3AcVyzVlYyIrcH/TM0cibeiI2O4p5LithUfYC/rawIyDEPt7TxyMJtnF/Qk+E5PQJyTBEJDE2TEAmhrJQERuSkqhg+js3VDayrqufas3XhnHjr8mG9GZGTykNvbKSxufWMjze7rILq+iZmjC8IQDoRCSQVwyIhNmlwFmW7PqKmITQ7XUWSWWUVREcZVw5XMSzeMjPuvWwQlXWNPPn+jjM6VlubY+aCcgb17s64Qq2bLRJuVAyLhNik4vapEm+tD+zFOZHOOcfsskouGJhBr+5ackq8d35BBuPO6sWv395C3aHT30r9nU01bK45wFfGF+iiUJEwpGJYJMQG9e5Odo9ELbH2Kct37Gf3/kOaIiFh5d4pRdQdaubh+VtP+xi/n19O39QErhjeJ4DJRCRQVAyLhJiZMbk4i3c37+Xg4Rav44SNWWUVJMRGccmQ3l5HETliSN9Urj27L4++t409dZ2f2rRi536WbNvH7WMHEButP7ki4UivTBEPTC7OoqmljYWb93odJSw0t7bxygdVTC7uTbf4GK/jiHzCty8porXN8at5mzr93Jnzy0lNjGXqqNwgJBORQFAxLOKB0fnpdE+I0VSJDgs21bL/YLOmSEhYyk1P4uYxeTy7dBdbOrFGeHntAV5ft4dbz8sjWW/yRMKWXp0iHoiNjuLiokzmra+htc0RHRX4i2qcc9QfamFPfSN76huprmv8xO3qhkZ6pyRwQ0kuEwZlevoR7qyyStKSYhl3lrZsl/B054SBPL98Nz9/fSO/v3XkKT3nD+9uIzY6imnn9w9uOBE5IyqGRTwyqTiLl1ZVUrZrPyPz0jv13MMtbdQ0NFJd38ieuqb2Are+kT0dBW91x7/G5rZjnpueHEdWSgJZKfGsrqjjzfU1ZHSL5/qROUd23gqlA00tzF23h+tH5mhOpYStnt3imT5uAA/N3cSKnfs5t1/aCR9f09DICyt2c/3IHK2OIhLmVAyLeGT8Wb2IiTLmrqs5Ugw756g71Nw+glv3yWK3pv7/C929Bw4fc7y4mCiyUuLpnZLAsOxUJg/OondqAlkpCfROTaB3SgKZKfHEx0QfeU5LaxvzN9XyzNJd/OHdcn4/fytj8tOZOjqXy4b2ISE2+pjzBNoba/fQ2Nym7Zcl7N1+YT5PvL+Dn726gWdnnHfCZdL+9I/tNLe28eWx2npZJNypGBbxSGpiLGMGpPPcsl2s2Ln/yMhuU8vxR3N7p8QzPCe143YCWR1Fbu+UBHokxXZ6DdOY6CgmDs5i4uAsauobeX7Fbp5duotvPruKf5+9luvOyeamUbkM6ZsaqGYfY1ZZJTlpiYzMO/FIm4jXkuNjuGviQH44ey1vb6xhwqCsz3zcgaYWnnx/B1OG9A75Jy0i0nkqhkU8dOt5eTw4ZyM4GJ7Tg0uK4zumMBx/NDdYMlMS+NpFA/nKuAIWb9vHs0t38szSXTzx/g6GZady06hcrj67LykJsQE7Z21DEws31/LVi7QZgUSGqaP78cjCbTw4ZyPjz8r8zPn+zyzZSX1jC9PHaVRYJBKoGBbx0JShfZgyNLwW4o+KMkoLelJa0JMfH2xmVlkFTy/ZyQ9mreE/XlnHFcP6MnV0LiV5aWdcwP79g0raHJoiIREjNjqKey4t4ut/WcmslRV8bmTOJ77f3NrGIwu3MSY/nXNOMq9YRMKDimEROa7UpFimnd+fL5TmsbqijmeW7uKlskpeWLGbgl7JTB3Vj+vOzSaj2+ldIDS7rJLiPikUZnUPcHKR4Ll8aB+GZZfz0NxNXDH8k3PrX15VSVVdI/dfN8zDhCLSGbp0W0ROyswYntOD+68bxpLvT+S/rh9Oj6Q4/vPV9ZT+dB5fe2o572xsXybuVG3f+zFluz7i2nO0trBElqgo47uXDaLio0P8edGOI/c753h4fjlFWd25qEjLBIpECo0Mi0inJMXFcENJLjeU5LKlpoFnl+7ihRUVvLp6D31T29ctvnFULtk9Ek94nNlllZjBVSNUDEvkuWBgBmMLM/j121u4cVQuKQmxvLOplo3VDfzihhGaAy8SQU46Mmxmj5pZjZmtOeq+dDOba2abO/7XxCgRHxqY2Z3vX1HMou9N5Lc3n8vArO78z1ubufCBt5j26BJeW13F4c9YHcM5x+yyCsbkp9Mn9cRFs0i4unfKID462MzD87cC8PD8rfRJTdAbPJEIcyrTJB4Hpnzqvu8C85xzhcC8jq9FxKfiYqK4fFgfnrhtNO9+52LunFDIpuoGvvrUCkp/Oo/7X13/iW1sV1fUUb73Y104JxFtaHYqV4/oyyMLtzF3XTWLyvdx2wX5xMVoBqJIJDnpNAnn3AIz6/+pu68BLuq4/SfgHeDeAOYSkQiVk5bEtyafxV0TC1mwuZZnl+zi0YXbmLmgnFH907hpVD9W7txPXHQUlw0Lr5U0/MDMvgl8CXDAauCLzrlGb1NFrnsuKeK1NVXc8dQKuifEMHV0rteRRKSTTnfOcJZzrgrAOVdlZpnHe6CZTQemA/Tr1+80TycikSY6yri4KJOLizKpbWjixY4NPe756yoALh2SRWpi4NYslpMzs2zgG0Cxc+6QmT0HTKX9E0A5Df16JnHzmDwe/8d2bh+bT/cArsMtIqER9AvonHMzgZkAJSUlp36puYh0Gb26xzNjfAHTxw1g2Y79vPJBFTeU5Jz8iRIMMUCimTUDSUClx3ki3l0TCwGYrq2XRSLS6RbD1WbWp2NUuA9QE8hQItI1mRmj+qczqn+611F8yTlXYWY/B3YCh4A3nHNvHP0YfZrXeWnJcfzo6iFexxCR03S6s/xfAqZ13J4GzA5MHBERCZaOlX+uAfKBvkCymd1y9GOcczOdcyXOuZJevbRWroh0faeytNrTwPtAkZntNrPbgZ8Bk81sMzC542sREQlvk4Btzrla51wz8CJwvseZREQ8dSqrSXz+ON+aGOAsIiISXDuB88wsifZpEhOBZd5GEhHxlhZDFBHxCefcYuB5YAXty6pF0XGBs4iIX2k7ZhERH3HO3Qfc53UOEZFwoZFhEREREfEtFcMiIiIi4lsqhkVERETEt8y50G0KZ2a1wI7TeGoGsDfAccKdH9sM/my3H9sMkdfuPOecrxbeVZ/daX5st9rsH5HY7lPqt0NaDJ8uM1vmnCvxOkco+bHN4M92+7HN4N92+4Fff7Z+bLfa7B9dud2aJiEiIiIivqViWERERER8K1KKYT8uCu/HNoM/2+3HNoN/2+0Hfv3Z+rHdarN/dNl2R8ScYRERERGRYIiUkWERERERkYBTMSwiIiIivhXWxbCZTTGzjWa2xcy+63WeUDCzXDN728zWm9laM7vL60yhYmbRZrbSzP7udZZQMbMeZva8mW3o+JmXep0p2Mzsmx2/22vM7GkzS/A6kwSO3/pt9dnqs73OFGx+6LPDthg2s2jgN8BlQDHweTMr9jZVSLQA33bODQbOA+7wSbsB7gLWex0ixH4FzHHODQJG0MXbb2bZwDeAEufcUCAamOptKgkUn/bb6rP9RX12F+yzw7YYBkYDW5xz5c65w8AzwDUeZwo651yVc25Fx+0G2l9o2d6mCj4zywGuAP7odZZQMbMUYBzwCIBz7rBz7iNvU4VEDJBoZjFAElDpcR4JHN/12+qz1Wd7myokunyfHc7FcDaw66ivd+ODDuZoZtYfOAdY7G2SkPgl8B2gzesgITQAqAUe6/io8Y9mlux1qGByzlUAPwd2AlVAnXPuDW9TSQD5ut9Wn93lqc/uon12OBfD9hn3+WYdODPrBrwA3O2cq/c6TzCZ2ZVAjXNuuddZQiwGOBf4nXPuHOBjoEvPsTSzNNpHCvOBvkCymd3ibSoJIN/22+qzfUF9dhfts8O5GN4N5B71dQ5dcGj+s5hZLO2d6lPOuRe9zhMCFwBXm9l22j9WnWBmf/Y2UkjsBnY75/45ivQ87R1tVzYJ2Oacq3XONQMvAud7nEkCx5f9tvps9dldmC/67HAuhpcChWaWb2ZxtE/YfsnjTEFnZkb7fKT1zrmHvM4TCs657znncpxz/Wn/Ob/lnOty7zw/zTm3B9hlZkUdd00E1nkYKRR2AueZWVLH7/pEuvgFKD7ju35bfbb6bA8jhYIv+uwYrwMcj3Ouxcy+DrxO+9WLjzrn1nocKxQuAG4FVptZWcd9/+ace9XDTBI8dwJPdRQO5cAXPc4TVM65xWb2PLCC9qvwV9KFt/j0G5/22+qz/UV9dhfss7Uds4iIiIj4VjhPkxARERERCSoVwyIiIiLiWyqGRURERMS3VAyLiIiIiG+pGBYRERER31IxLCIiIiK+pWJYRERERHzr/wDdxKhHeHr7xwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fold mean histograms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training on fold 2\n",
      "================================================================================\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 461s 122ms/step - loss: 24.6493\n",
      "custom valid metrics\n",
      "unsmoothed mse 18.712229 \t16 smoothed mse 13.39701966477753\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 6.0087\n",
      "comma val metrics\n",
      "unsmoothed mse 11.225768 \t16 smoothed mse 7.008941729657572\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 462s 122ms/step - loss: 12.6096\n",
      "custom valid metrics\n",
      "unsmoothed mse 15.592646 \t16 smoothed mse 11.147829899549807\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 5.0743\n",
      "comma val metrics\n",
      "unsmoothed mse 12.046751 \t16 smoothed mse 8.031322478765537\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 462s 122ms/step - loss: 10.0840\n",
      "custom valid metrics\n",
      "unsmoothed mse 13.853809 \t16 smoothed mse 9.477568707661908\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 4.6772\n",
      "comma val metrics\n",
      "unsmoothed mse 10.368568 \t16 smoothed mse 7.502275323380444\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 460s 121ms/step - loss: 8.5717\n",
      "custom valid metrics\n",
      "unsmoothed mse 16.043688 \t16 smoothed mse 12.33732577107825\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 58s 113ms/step - loss: 4.8933\n",
      "comma val metrics\n",
      "unsmoothed mse 9.042676 \t16 smoothed mse 6.305488326672465\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 452s 119ms/step - loss: 7.4109\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.464027 \t16 smoothed mse 7.05003348985834\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 53s 104ms/step - loss: 4.3385\n",
      "comma val metrics\n",
      "unsmoothed mse 9.621104 \t16 smoothed mse 6.178613556607033\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 461s 122ms/step - loss: 6.8860\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.003428 \t16 smoothed mse 8.314465721357422\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 57s 112ms/step - loss: 4.1325\n",
      "comma val metrics\n",
      "unsmoothed mse 9.79061 \t16 smoothed mse 5.780359372305846\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 461s 122ms/step - loss: 6.0736\n",
      "custom valid metrics\n",
      "unsmoothed mse 13.196251 \t16 smoothed mse 9.23364393023436\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 3.7964\n",
      "comma val metrics\n",
      "unsmoothed mse 9.037233 \t16 smoothed mse 6.210241528628496\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 461s 121ms/step - loss: 5.9213\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.118659 \t16 smoothed mse 6.969753405235419\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 3.6299\n",
      "comma val metrics\n",
      "unsmoothed mse 8.637254 \t16 smoothed mse 5.2422992105473805\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 460s 121ms/step - loss: 5.3808\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.148132 \t16 smoothed mse 6.9696150540013955\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 55s 109ms/step - loss: 3.5157\n",
      "comma val metrics\n",
      "unsmoothed mse 9.771973 \t16 smoothed mse 6.3294099630710186\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 460s 121ms/step - loss: 5.0144\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.005432 \t16 smoothed mse 8.04907926222821\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 3.1970\n",
      "comma val metrics\n",
      "unsmoothed mse 10.38923 \t16 smoothed mse 6.682293070327277\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fold mean histograms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training on fold 3\n",
      "================================================================================\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 471s 124ms/step - loss: 23.5302\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.619921 \t16 smoothed mse 7.848295057292025\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 6.4401\n",
      "comma val metrics\n",
      "unsmoothed mse 9.163071 \t16 smoothed mse 5.794520096023704\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 467s 123ms/step - loss: 13.5492\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.737404 \t16 smoothed mse 7.183249542469412\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 111ms/step - loss: 5.5420\n",
      "comma val metrics\n",
      "unsmoothed mse 9.325732 \t16 smoothed mse 5.747725209387715\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 467s 123ms/step - loss: 11.0906\n",
      "custom valid metrics\n",
      "unsmoothed mse 10.83359 \t16 smoothed mse 6.631488275880662\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 5.2345\n",
      "comma val metrics\n",
      "unsmoothed mse 9.661849 \t16 smoothed mse 6.106546233467795\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 470s 124ms/step - loss: 9.4407\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.260871 \t16 smoothed mse 6.6003746699325685\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 57s 111ms/step - loss: 4.9556\n",
      "comma val metrics\n",
      "unsmoothed mse 9.275025 \t16 smoothed mse 5.095735425332127\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 469s 124ms/step - loss: 8.1442\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.6192465 \t16 smoothed mse 6.225474777210496\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 4.6376\n",
      "comma val metrics\n",
      "unsmoothed mse 9.611098 \t16 smoothed mse 5.642463758804034\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 468s 123ms/step - loss: 7.5618\n",
      "custom valid metrics\n",
      "unsmoothed mse 10.573258 \t16 smoothed mse 6.00913777077164\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 4.2919\n",
      "comma val metrics\n",
      "unsmoothed mse 10.581233 \t16 smoothed mse 5.971564405674724\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 464s 122ms/step - loss: 6.7764s \n",
      "custom valid metrics\n",
      "unsmoothed mse 9.83146 \t16 smoothed mse 4.871736808931618\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 3.7872\n",
      "comma val metrics\n",
      "unsmoothed mse 10.012157 \t16 smoothed mse 5.867360929969857\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 453s 119ms/step - loss: 6.1205\n",
      "custom valid metrics\n",
      "unsmoothed mse 9.14134 \t16 smoothed mse 4.615425054636533\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 3.8542\n",
      "comma val metrics\n",
      "unsmoothed mse 6.19258 \t16 smoothed mse 2.8758398270035763\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 467s 123ms/step - loss: 5.4442\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.330582 \t16 smoothed mse 5.822710155011012\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 57s 112ms/step - loss: 3.2675\n",
      "comma val metrics\n",
      "unsmoothed mse 7.506882 \t16 smoothed mse 3.722780174041571\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 465s 123ms/step - loss: 5.0411\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.865165 \t16 smoothed mse 7.178982190570247\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 3.2195\n",
      "comma val metrics\n",
      "unsmoothed mse 17.876596 \t16 smoothed mse 12.410796918594233\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fold mean histograms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training on fold 4\n",
      "================================================================================\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 450s 119ms/step - loss: 24.2652\n",
      "custom valid metrics\n",
      "unsmoothed mse 16.001095 \t16 smoothed mse 8.967145074733123\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 55s 108ms/step - loss: 7.2036\n",
      "comma val metrics\n",
      "unsmoothed mse 7.7305665 \t16 smoothed mse 6.0399594214321946\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 449s 118ms/step - loss: 13.8578\n",
      "custom valid metrics\n",
      "unsmoothed mse 14.448364 \t16 smoothed mse 8.109861977846327\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 5.8712\n",
      "comma val metrics\n",
      "unsmoothed mse 6.285473 \t16 smoothed mse 4.691406879005087\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 452s 119ms/step - loss: 11.0028\n",
      "custom valid metrics\n",
      "unsmoothed mse 17.50178 \t16 smoothed mse 10.39887955256498\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 5.8712\n",
      "comma val metrics\n",
      "unsmoothed mse 4.7242393 \t16 smoothed mse 2.9917056962927906\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 451s 119ms/step - loss: 9.2999\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.400234 \t16 smoothed mse 6.117657274484655\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 5.2752\n",
      "comma val metrics\n",
      "unsmoothed mse 6.2590685 \t16 smoothed mse 4.491031936174682\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 452s 119ms/step - loss: 8.3119\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.666067 \t16 smoothed mse 6.157071742894997\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 5.0054\n",
      "comma val metrics\n",
      "unsmoothed mse 5.4550724 \t16 smoothed mse 3.3498729447696913\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 453s 119ms/step - loss: 7.2905\n",
      "custom valid metrics\n",
      "unsmoothed mse 10.738752 \t16 smoothed mse 5.867254507707361\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 109ms/step - loss: 4.7622\n",
      "comma val metrics\n",
      "unsmoothed mse 4.5861135 \t16 smoothed mse 2.572440348806453\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 453s 119ms/step - loss: 6.6363\n",
      "custom valid metrics\n",
      "unsmoothed mse 11.267087 \t16 smoothed mse 7.148219458099286\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 56s 110ms/step - loss: 4.2329\n",
      "comma val metrics\n",
      "unsmoothed mse 4.9933605 \t16 smoothed mse 3.146025605751995\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 451s 119ms/step - loss: 6.1842\n",
      "custom valid metrics\n",
      "unsmoothed mse 13.93933 \t16 smoothed mse 7.197634460789605\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 53s 104ms/step - loss: 4.2586\n",
      "comma val metrics\n",
      "unsmoothed mse 4.584555 \t16 smoothed mse 2.8660770431445575\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 425s 112ms/step - loss: 5.7669\n",
      "custom valid metrics\n",
      "unsmoothed mse 9.920947 \t16 smoothed mse 5.26719196880307\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 53s 103ms/step - loss: 3.8516\n",
      "comma val metrics\n",
      "unsmoothed mse 5.962401 \t16 smoothed mse 4.17829128108734\n",
      "Epoch 1/1\n",
      "3796/3796 [==============================] - 426s 112ms/step - loss: 5.1405\n",
      "custom valid metrics\n",
      "unsmoothed mse 12.538888 \t16 smoothed mse 7.1790692250623405\n",
      "Epoch 1/1\n",
      "510/510 [==============================] - 53s 104ms/step - loss: 3.2166\n",
      "comma val metrics\n",
      "unsmoothed mse 6.2523236 \t16 smoothed mse 4.01758584745952\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fold mean histograms\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "split_count = 5\n",
    "num_epochs = 10\n",
    "steps_per_epoch = None\n",
    "kf = KFold(shuffle = False, n_splits = split_count)\n",
    "test_preds = np.zeros((split_count, len(comma_test_index_array)))\n",
    "oof_preds = np.zeros((len(index_array)))\n",
    "comma_oof_preds = np.zeros((len(comma_index_array)))\n",
    "custom_splits = list(kf.split(index_array))\n",
    "comma_splits = list(kf.split(comma_index_array))\n",
    "for fold in range(split_count):\n",
    "    train_indexes, val_indexes = custom_splits[fold]\n",
    "    comma_train_indexes, comma_val_indexes = comma_splits[fold]\n",
    "    print(\"training on fold \" + str(fold))\n",
    "    print(\"=\"*80)\n",
    "    \n",
    "    best_comma_mse = 200\n",
    "    best_custom_mse = 200\n",
    "    \n",
    "    comma_train_generator = DataGenerator(32, time_history, hdf5_path_comma, indexes = comma_train_indexes)\n",
    "    comma_val_generator =  DataGenerator(32, time_history, hdf5_path_comma, indexes = comma_val_indexes, validation_mode = True)\n",
    "    comma_test_generator = DataGenerator(32, time_history, hdf5_path_comma_test, indexes = comma_test_index_array, validation_mode = True)\n",
    "    \n",
    "    train_generator = DataGenerator(32, time_history, hdf5_path, indexes = train_indexes)\n",
    "    valid_generator = DataGenerator(32, time_history, hdf5_path, indexes = val_indexes, validation_mode = True)\n",
    "    \n",
    "    model_flat = build_model_flat(time_history)\n",
    "    comma_mse = []\n",
    "    custom_mse = []\n",
    "    \n",
    "    for i in range(num_epochs):\n",
    "        model_flat.fit_generator(train_generator, epochs = 1, steps_per_epoch=steps_per_epoch)\n",
    "        val_predictions = model_flat.predict_generator(valid_generator)[:, 0]\n",
    "        print(\"custom valid metrics\")\n",
    "        val_roll_mse, val_roll_preds = evaluate_metrics(np.array(speed_data)[val_indexes], val_predictions, title = \"custom data\")\n",
    "        custom_mse.append(val_roll_mse)\n",
    "        if val_roll_mse < best_custom_mse:\n",
    "            oof_preds[val_indexes] = val_roll_preds\n",
    "    \n",
    "        model_flat.fit_generator(comma_train_generator, epochs = 1, steps_per_epoch=steps_per_epoch)\n",
    "        print(\"comma val metrics\")\n",
    "        comma_val_predictions = model_flat.predict_generator(comma_val_generator)[:, 0]\n",
    "        comma_roll_mse, comma_roll_preds = evaluate_metrics(np.array(comma_speed_data)[comma_val_indexes], comma_val_predictions, title = \"comma data\")\n",
    "        comma_mse.append(comma_roll_mse)\n",
    "        if comma_roll_mse < best_comma_mse:\n",
    "            model_flat.save(\"best_model\" + str(fold) + \".ckpt\")\n",
    "            comma_test_predictions = model_flat.predict_generator(comma_test_generator)[:, 0][comma_test_index_array]\n",
    "            comma_test_predictions = runningMeanFast(comma_test_predictions, 21)\n",
    "            test_preds[fold, :] = comma_test_predictions\n",
    "            comma_oof_preds[comma_val_indexes] = comma_roll_preds\n",
    "            best_comma_mse = comma_roll_mse\n",
    "        \n",
    "#         plt.plot(comma_test_predictions)\n",
    "#         plt.plot(np.mean(test_preds[:fold+1], axis = 0))\n",
    "#         plt.title(\"comma test predictions\")\n",
    "#         plt.show()\n",
    "        \n",
    "    plt.figure(figsize=(12,4))\n",
    "    plt.subplot(1, 2, 1)\n",
    "    plt.plot(custom_mse)\n",
    "    plt.title(\"smoothed custom validation mse\")\n",
    "    plt.subplot(1, 2, 2)\n",
    "    plt.plot(comma_mse)\n",
    "    plt.title(\"smoothed comma validation mse\")\n",
    "    plt.show()\n",
    "\n",
    "    fold_means = np.mean(test_preds[:fold+1], axis = 0)\n",
    "    plt.plot(fold_means)\n",
    "    plt.title(\"best fold mean predictions\")\n",
    "    plt.show()\n",
    "    print(\"fold mean histograms\")\n",
    "    pd.DataFrame(fold_means).hist(bins = 25)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5.067375083527054\n",
      "8.25015783430225\n"
     ]
    }
   ],
   "source": [
    "#comma oof performance\n",
    "print(mean_squared_error(comma_speed_data[1:], comma_oof_preds))\n",
    "\n",
    "#custom off performance\n",
    "print(mean_squared_error(speed_data, oof_preds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x0000025D87F75438>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(speed_data).hist(bins = 25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x0000025DE77844A8>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEICAYAAABWJCMKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAEPFJREFUeJzt3W2MXNV9x/Hvv5A0kZeAkZOVZdxuGrlVCVYorAA1VbUrVDDwwkQqFRQRmyZy1IKUSLzAiRSBQlJZFaQVKUV1hBXTElZICbUFbqllZUV5QYuNCIZQwCIO9YNspXacLEGNnPz7Yq6VYb2787AzOw/n+5FWM3Pm3HvP8d2dn++5956JzESSVJ7f6HUDJEm9YQBIUqEMAEkqlAEgSYUyACSpUAaAJBXKAJCkQhkAUosi4sKIeDIi3omIH0XEn/e6TVI7zu11A6QB9BDwC2AUuBR4OiK+n5mv9rZZUmvCO4Gl5kXEMuAkcElmvlGV/RNwODM397RxUoscApJa87vAL898+Fe+D3y8R+2R2mYASK0ZAU7NKjsFnNeDtkiLYgBIrZkBPjSr7EPAz3rQFmlRDACpNW8A50bEmrqyTwCeANbA8SSw1KKImAIS+Cy1q4B2AX/oVUAaNB4BSK37K+CDwHHgceAv/fDXIPIIQJIK5RGAJBXKAJCkQhkAklQoA0CSCtXXk8GtWLEix8bG2l7+nXfeYdmyZZ1rUJ+wX4PFfg2WYejXvn37fpyZH25Ur68DYGxsjL1797a9/PT0NBMTE51rUJ+wX4PFfg2WYehXRPyomXoOAUlSoQwASSqUASBJhTIAJKlQBoAkFcoAkKRCGQCSVCgDQJIKZQBIUqH6+k5gScNpbPPTLdU/uOWGLrWkbB4BSFKhDABJKpQBIEmFMgAkqVAGgCQVygCQpEIZAJJUKANAkgplAEhSoQwASSqUASBJhTIAJKlQBoAkFaphAETE6oj4XkS8FhGvRsTnq/ILI2J3RLxZPS6vyiMiHoyIAxHxckRcVreuDVX9NyNiQ/e6JUlqpJkjgNPAXZn5+8BVwB0RcTGwGdiTmWuAPdVrgOuANdXPJuBhqAUGcA9wJXAFcM+Z0JAkLb2GAZCZRzPzxer5z4DXgFXAemB7VW07cGP1fD3waNY8D1wQESuBa4HdmXkiM08Cu4F1He2NJKlpkZnNV44YA54FLgHezswL6t47mZnLI+IpYEtmPleV7wHuBiaAD2TmV6vyLwPvZub9s7axidqRA6Ojo5dPTU213bmZmRlGRkbaXr5f2a/BYr/Otv/wqZbqr111flvbaccw7K/Jycl9mTneqF7T3wgWESPAd4AvZOZPI2LeqnOU5QLl7y3I3ApsBRgfH8+JiYlmm3iW6elpFrN8v7Jfg8V+nW1jq98Idmt722nHsO6vuTR1FVBEvI/ah/9jmfndqvhYNbRD9Xi8Kj8ErK5b/CLgyALlkqQeaOYqoAAeAV7LzK/XvbUTOHMlzwZgR135p6urga4CTmXmUeAZ4JqIWF6d/L2mKpMk9UAzQ0CfBG4D9kfES1XZl4AtwBMR8RngbeCm6r1dwPXAAeDnwO0AmXkiIu4DXqjqfSUzT3SkF5KkljUMgOpk7nwD/lfPUT+BO+ZZ1zZgWysNlCR1h3cCS1KhDABJKpQBIEmFMgAkqVAGgCQVygCQpEIZAJJUKANAkgplAEhSoQwASSqUASBJhTIAJKlQBoAkFcoAkKRCGQCSVCgDQJIKZQBIUqEMAEkqlAEgSYUyACSpUAaAJBXKAJCkQhkAklQoA0CSCmUASFKhDABJKpQBIEmFMgAkqVAGgCQVygCQpEIZAJJUKANAkgplAEhSoQwASSqUASBJhTIAJKlQBoAkFercXjegn4xtfrql+ge33NCllkhS9zU8AoiIbRFxPCJeqSu7NyIOR8RL1c/1de99MSIORMTrEXFtXfm6quxARGzufFckSa1oZgjoW8C6Ocr/NjMvrX52AUTExcDNwMerZf4hIs6JiHOAh4DrgIuBW6q6kqQeaTgElJnPRsRYk+tbD0xl5v8BP4yIA8AV1XsHMvMtgIiYqur+oOUWS5I6IjKzcaVaADyVmZdUr+8FNgI/BfYCd2XmyYj4e+D5zPznqt4jwL9Wq1mXmZ+tym8DrszMO+fY1iZgE8Do6OjlU1NTbXduZmaGkZGRpuvvP3yqpfWvXXV+q03qiFb7NSjs12BZTL/6+W9tGPbX5OTkvswcb1Sv3ZPADwP3AVk9PgD8BRBz1E3mHmqaM3kycyuwFWB8fDwnJibabCJMT0/TyvIbWz0JfGvz6+6kVvs1KOzXYFlMv/r5b21Y99dc2gqAzDx25nlEfBN4qnp5CFhdV/Ui4Ej1fL5ySVIPtHUfQESsrHv5KeDMFUI7gZsj4jcj4qPAGuC/gBeANRHx0Yh4P7UTxTvbb7YkabEaHgFExOPABLAiIg4B9wATEXEptWGcg8DnADLz1Yh4gtrJ3dPAHZn5y2o9dwLPAOcA2zLz1Y73RpLUtGauArpljuJHFqj/NeBrc5TvAna11DpJUtc4FYQkFcoAkKRCGQCSVCgngyucE+BJ5fIIQJIKZQBIUqEMAEkqlAEgSYUyACSpUAaAJBVqqC8D3X/4VMvTzkpSKTwCkKRCGQCSVCgDQJIKZQBIUqEMAEkqlAEgSYUyACSpUAaAJBXKAJCkQhkAklQoA0CSCmUASFKhDABJKpQBIEmFMgAkqVAGgCQVygCQpEIZAJJUqKH+SkhJZRpr46tgD265oQst6W8eAUhSoQwASSqUASBJhTIAJKlQBoAkFcoAkKRCGQCSVCgDQJIKZQBIUqEaBkBEbIuI4xHxSl3ZhRGxOyLerB6XV+UREQ9GxIGIeDkiLqtbZkNV/82I2NCd7kiSmtXMEcC3gHWzyjYDezJzDbCneg1wHbCm+tkEPAy1wADuAa4ErgDuORMakqTeaBgAmfkscGJW8Xpge/V8O3BjXfmjWfM8cEFErASuBXZn5onMPAns5uxQkSQtocjMxpUixoCnMvOS6vVPMvOCuvdPZubyiHgK2JKZz1Xle4C7gQngA5n51ar8y8C7mXn/HNvaRO3ogdHR0cunpqba7tzxE6c49m7bize0dtX53Vv5AmZmZhgZGenIuvYfPtVS/W72uZP96if262zd/r1rdf312xiG/TU5ObkvM8cb1ev0bKAxR1kuUH52YeZWYCvA+Ph4TkxMtN2Ybzy2gwf2d2/C04O3TnRt3QuZnp5mMf8u9Ta2OGtiN/vcyX71E/t1tm7/3rW6/vptDOv+mku7VwEdq4Z2qB6PV+WHgNV19S4CjixQLknqkXb/e7wT2ABsqR531JXfGRFT1E74nsrMoxHxDPDXdSd+rwG+2H6zyzDfnOZ3rT097/9wuj2neavzrJc4x7o0KBoGQEQ8Tm0Mf0VEHKJ2Nc8W4ImI+AzwNnBTVX0XcD1wAPg5cDtAZp6IiPuAF6p6X8nM2SeWJUlLqGEAZOYt87x19Rx1E7hjnvVsA7a11DpJUtd4J7AkFcoAkKRC+aXwi+AJUUmDzCMASSqUASBJhXIIaMi0OizVba205661p5noXlMkzWIALKF++3CWBoV/O93hEJAkFcoAkKRCOQQkNeDlvhpWHgFIUqEMAEkqlAEgSYUyACSpUAaAJBXKAJCkQnkZqAZaO3eIepmmVOMRgCQVyiMA9RXnfJGWjkcAklQoA0CSCmUASFKhDABJKpQBIEmFMgAkqVAGgCQVygCQpEIZAJJUKO8EliR+fRf6XWtPs7GJO9KHYU4pjwAkqVAGgCQVygCQpEIZAJJUKANAkgplAEhSoQwASSqUASBJhfJGMBXHr52UahZ1BBARByNif0S8FBF7q7ILI2J3RLxZPS6vyiMiHoyIAxHxckRc1okOSJLa04khoMnMvDQzx6vXm4E9mbkG2FO9BrgOWFP9bAIe7sC2JUlt6sY5gPXA9ur5duDGuvJHs+Z54IKIWNmF7UuSmhCZ2f7CET8ETgIJ/GNmbo2In2TmBXV1Tmbm8oh4CtiSmc9V5XuAuzNz76x1bqJ2hMDo6OjlU1NTbbfv+IlTHHu37cX71ugHsV99bO2q89/zemZmhpGRkR61pnsW06/9h091uDWd0+zv4ez93E8mJyf31Y3KzGuxJ4E/mZlHIuIjwO6I+O8F6sYcZWelT2ZuBbYCjI+P58TERNuN+8ZjO3hg//Cd575r7Wn71ccO3jrxntfT09Ms5ve4X9X3q/UT6/27n5v9PZy9nwfRovZCZh6pHo9HxJPAFcCxiFiZmUerIZ7jVfVDwOq6xS8Cjixm+5Iaa+eqp2GY6liNtX0OICKWRcR5Z54D1wCvADuBDVW1DcCO6vlO4NPV1UBXAacy82jbLZckLcpijgBGgScj4sx6vp2Z/xYRLwBPRMRngLeBm6r6u4DrgQPAz4HbF7FtqVhLcR9DM9to9otT1L/aDoDMfAv4xBzl/wtcPUd5Ane0uz1pWHljmnrFqSAkqVAGgCQVygCQpEIZAJJUKANAkgplAEhSoQwASSqUASBJhTIAJKlQBoAkFap/52SVBtTsqR2cM0f9yiMASSqUASBJhTIAJKlQBoAkFcoAkKRCGQCSVCgDQJIKZQBIUqEMAEkqlAEgSYUyACSpUAaAJBXKAJCkQhkAklQoA0CSCmUASFKhDABJKpQBIEmF8ishJakNs7/6s5GDW27oUkva5xGAJBXKAJCkQhkAklQoA0CSCmUASFKhvApIkpZAP1415BGAJBXKAJCkQhkAklSoJQ+AiFgXEa9HxIGI2LzU25ck1SxpAETEOcBDwHXAxcAtEXHxUrZBklSz1EcAVwAHMvOtzPwFMAWsX+I2SJKAyMyl21jEnwLrMvOz1evbgCsz8866OpuATdXL3wNeX8QmVwA/XsTy/cp+DRb7NViGoV+/nZkfblRpqe8DiDnK3pNAmbkV2NqRjUXszczxTqyrn9ivwWK/Bsuw9msuSz0EdAhYXff6IuDIErdBksTSB8ALwJqI+GhEvB+4Gdi5xG2QJLHEQ0CZeToi7gSeAc4BtmXmq13cZEeGkvqQ/Ros9muwDGu/zrKkJ4ElSf3DO4ElqVAGgCQVaigDYFinm4iIgxGxPyJeioi9vW7PYkTEtog4HhGv1JVdGBG7I+LN6nF5L9vYjnn6dW9EHK7220sRcX0v29iOiFgdEd+LiNci4tWI+HxVPtD7bIF+Dfw+a8bQnQOoppt4A/gTapedvgDckpk/6GnDOiAiDgLjmTnoN6kQEX8MzACPZuYlVdnfACcyc0sV3Msz8+5etrNV8/TrXmAmM+/vZdsWIyJWAisz88WIOA/YB9wIbGSA99kC/fozBnyfNWMYjwCcbmIAZOazwIlZxeuB7dXz7dT+EAfKPP0aeJl5NDNfrJ7/DHgNWMWA77MF+lWEYQyAVcD/1L0+xPDs0AT+PSL2VVNmDJvRzDwKtT9M4CM9bk8n3RkRL1dDRAM1TDJbRIwBfwD8J0O0z2b1C4Zon81nGAOg4XQTA+yTmXkZtdlU76iGG9T/HgY+BlwKHAUe6G1z2hcRI8B3gC9k5k973Z5OmaNfQ7PPFjKMATC0001k5pHq8TjwJLXhrmFyrBqTPTM2e7zH7emIzDyWmb/MzF8B32RA91tEvI/ah+Rjmfndqnjg99lc/RqWfdbIMAbAUE43ERHLqpNURMQy4BrglYWXGjg7gQ3V8w3Ajh62pWPOfEBWPsUA7reICOAR4LXM/HrdWwO9z+br1zDss2YM3VVAANUlW3/Hr6eb+FqPm7RoEfE71P7XD7UpPL49yP2KiMeBCWpT7x4D7gH+BXgC+C3gbeCmzByoE6rz9GuC2lBCAgeBz50ZNx8UEfFHwH8A+4FfVcVfojZePrD7bIF+3cKA77NmDGUASJIaG8YhIElSEwwASSqUASBJhTIAJKlQBoAkFcoAkKRCGQCSVKj/BwbUO8bbjkFLAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(comma_speed_data).hist(bins = 25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x0000025DF4F6C0B8>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(fold_means).hist(bins = 25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([13.41369704, 14.08682982, 14.11967446, ...,  2.60811919,\n",
       "        1.80955536,  0.97247431])"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "with open(\"test.txt\", \"w\") as f:\n",
    "    for line in [13] + list(fold_means):\n",
    "        f.write(str(line) + \"\\n\")\n",
    "fold_means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model(history_size):\n",
    "    k.clear_session()\n",
    "    frame_inp = Input(shape=(history_size, 224, 224, 3))\n",
    "    op_flow_inp = Input(shape=(history_size, 224, 224, 3))\n",
    "    filter_size = (3,3)\n",
    "    #frame = TimeDistributed(BatchNormalization())(frame_inp)\n",
    "    frame = TimeDistributed(SpatialDropout2D(.2))(frame_inp)\n",
    "    frame = TimeDistributed(Conv2D(4, filter_size, activation = \"relu\", data_format = \"channels_last\"))(frame)\n",
    "    frame = TimeDistributed(MaxPool2D())(frame)\n",
    "    frame = TimeDistributed(Conv2D(4, filter_size, activation = \"relu\", data_format = \"channels_last\"))(frame)\n",
    "    frame = TimeDistributed(MaxPool2D())(frame)\n",
    "    frame = TimeDistributed(Conv2D(8, filter_size, activation = \"relu\", data_format = \"channels_last\"))(frame)\n",
    "    frame = TimeDistributed(MaxPool2D())(frame)\n",
    "    frame = TimeDistributed(Conv2D(16, filter_size, activation = \"relu\", data_format = \"channels_last\"))(frame)\n",
    "    #frame = TimeDistributed(SpatialDropout2D(.2))(frame)\n",
    "    frame = TimeDistributed(MaxPool2D())(frame)\n",
    "    frame = TimeDistributed(Conv2D(32, filter_size, activation = \"relu\", data_format = \"channels_last\"))(frame)\n",
    "    frame = TimeDistributed(MaxPool2D())(frame)\n",
    "    frame = TimeDistributed(GlobalMaxPool2D())(frame)\n",
    "    \n",
    "    op_flow = TimeDistributed(BatchNormalization())(op_flow_inp)\n",
    "    op_flow = TimeDistributed(Dropout(.3))(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(4, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(8, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(32, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(64, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(Dropout(.3))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(128, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow_max = TimeDistributed(GlobalMaxPool2D())(op_flow)\n",
    "    op_flow_avg = TimeDistributed(GlobalAvgPool2D())(op_flow)\n",
    "    \n",
    "    conc = concatenate([op_flow_max, op_flow_avg,\n",
    "                       #frame\n",
    "                       ])\n",
    "    #conc = BatchNormalization()(conc)\n",
    "    conc = SpatialDropout1D(.2)(conc)\n",
    "    conc = CuDNNGRU(256)(conc)\n",
    "    conc = Dense(100, activation = \"relu\")(conc)\n",
    "    conc = Dropout(.2)(conc)\n",
    "    conc = Dense(50, activation = \"relu\")(conc)\n",
    "    conc = Dropout(.1)(conc)\n",
    "    result = Dense(1, activation='linear')(conc)\n",
    "    \n",
    "    model = Model(inputs=[frame_inp, op_flow_inp], outputs=[result])\n",
    "    print(model.summary())\n",
    "    model.compile(loss=\"mse\", optimizer='adam')\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_2 (InputLayer)            (None, 1, 224, 224,  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_13 (TimeDistri (None, 1, 224, 224,  12          input_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_14 (TimeDistri (None, 1, 224, 224,  0           time_distributed_13[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_15 (TimeDistri (None, 1, 222, 222,  112         time_distributed_14[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_16 (TimeDistri (None, 1, 111, 111,  0           time_distributed_15[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_17 (TimeDistri (None, 1, 109, 109,  296         time_distributed_16[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_18 (TimeDistri (None, 1, 54, 54, 8) 0           time_distributed_17[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_19 (TimeDistri (None, 1, 52, 52, 32 2336        time_distributed_18[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_20 (TimeDistri (None, 1, 26, 26, 32 0           time_distributed_19[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_21 (TimeDistri (None, 1, 24, 24, 64 18496       time_distributed_20[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_22 (TimeDistri (None, 1, 24, 24, 64 0           time_distributed_21[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_23 (TimeDistri (None, 1, 12, 12, 64 0           time_distributed_22[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_24 (TimeDistri (None, 1, 10, 10, 12 73856       time_distributed_23[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_25 (TimeDistri (None, 1, 5, 5, 128) 0           time_distributed_24[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_26 (TimeDistri (None, 1, 128)       0           time_distributed_25[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_27 (TimeDistri (None, 1, 128)       0           time_distributed_25[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_1 (Concatenate)     (None, 1, 256)       0           time_distributed_26[0][0]        \n",
      "                                                                 time_distributed_27[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "spatial_dropout1d_1 (SpatialDro (None, 1, 256)       0           concatenate_1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "cu_dnngru_1 (CuDNNGRU)          (None, 256)          394752      spatial_dropout1d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 100)          25700       cu_dnngru_1[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "dropout_3 (Dropout)             (None, 100)          0           dense_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 50)           5050        dropout_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_4 (Dropout)             (None, 50)           0           dense_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_3 (Dense)                 (None, 1)            51          dropout_4[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 520,661\n",
      "Trainable params: 520,655\n",
      "Non-trainable params: 6\n",
      "__________________________________________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "model = build_model(time_history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "1061/1061 [==============================] - 619s 584ms/step - loss: 40.8469 - val_loss: 141.1516\n",
      "Epoch 2/10\n",
      "1061/1061 [==============================] - 614s 579ms/step - loss: 17.7586 - val_loss: 91.3322\n",
      "Epoch 3/10\n",
      "1061/1061 [==============================] - 620s 584ms/step - loss: 14.0745 - val_loss: 91.2476\n",
      "Epoch 4/10\n",
      "1061/1061 [==============================] - 636s 600ms/step - loss: 12.2106 - val_loss: 77.1058\n",
      "Epoch 5/10\n",
      "1061/1061 [==============================] - 621s 586ms/step - loss: 10.8576 - val_loss: 98.6059\n",
      "Epoch 6/10\n",
      " 334/1061 [========>.....................] - ETA: 5:58 - loss: 10.2859"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-87-1f2592fe14c1>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m      1\u001b[0m hist = model.fit_generator(train_generator, validation_data=valid_generator, epochs = 10, \n\u001b[1;32m----> 2\u001b[1;33m                            callbacks=[EarlyStopping(patience=3), ModelCheckpoint(filepath=\"rnn_model.ckpt\", save_weights_only=False)])\n\u001b[0m",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\legacy\\interfaces.py\u001b[0m in \u001b[0;36mwrapper\u001b[1;34m(*args, **kwargs)\u001b[0m\n\u001b[0;32m     89\u001b[0m                 warnings.warn('Update your `' + object_name + '` call to the ' +\n\u001b[0;32m     90\u001b[0m                               'Keras 2 API: ' + signature, stacklevel=2)\n\u001b[1;32m---> 91\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     92\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     93\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\engine\\training.py\u001b[0m in \u001b[0;36mfit_generator\u001b[1;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[0;32m   1416\u001b[0m             \u001b[0muse_multiprocessing\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1417\u001b[0m             \u001b[0mshuffle\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mshuffle\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1418\u001b[1;33m             initial_epoch=initial_epoch)\n\u001b[0m\u001b[0;32m   1419\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1420\u001b[0m     \u001b[1;33m@\u001b[0m\u001b[0minterfaces\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mlegacy_generator_methods_support\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\engine\\training_generator.py\u001b[0m in \u001b[0;36mfit_generator\u001b[1;34m(model, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[0;32m    215\u001b[0m                 outs = model.train_on_batch(x, y,\n\u001b[0;32m    216\u001b[0m                                             \u001b[0msample_weight\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 217\u001b[1;33m                                             class_weight=class_weight)\n\u001b[0m\u001b[0;32m    218\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    219\u001b[0m                 \u001b[0mouts\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mto_list\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\engine\\training.py\u001b[0m in \u001b[0;36mtrain_on_batch\u001b[1;34m(self, x, y, sample_weight, class_weight)\u001b[0m\n\u001b[0;32m   1215\u001b[0m             \u001b[0mins\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mx\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0my\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1216\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_make_train_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1217\u001b[1;33m         \u001b[0moutputs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mins\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1218\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0munpack_singleton\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1219\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py\u001b[0m in \u001b[0;36m__call__\u001b[1;34m(self, inputs)\u001b[0m\n\u001b[0;32m   2713\u001b[0m                 \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_legacy_call\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2714\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2715\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_call\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2716\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2717\u001b[0m             \u001b[1;32mif\u001b[0m \u001b[0mpy_any\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mis_tensor\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[1;32min\u001b[0m \u001b[0minputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py\u001b[0m in \u001b[0;36m_call\u001b[1;34m(self, inputs)\u001b[0m\n\u001b[0;32m   2653\u001b[0m                 array_vals.append(\n\u001b[0;32m   2654\u001b[0m                     np.asarray(value,\n\u001b[1;32m-> 2655\u001b[1;33m                                dtype=tf.as_dtype(tensor.dtype).as_numpy_dtype))\n\u001b[0m\u001b[0;32m   2656\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfeed_dict\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2657\u001b[0m             \u001b[1;32mfor\u001b[0m \u001b[0mkey\u001b[0m \u001b[1;32min\u001b[0m \u001b[0msorted\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfeed_dict\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mkeys\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\numpy\\core\\numeric.py\u001b[0m in \u001b[0;36masarray\u001b[1;34m(a, dtype, order)\u001b[0m\n\u001b[0;32m    499\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    500\u001b[0m     \"\"\"\n\u001b[1;32m--> 501\u001b[1;33m     \u001b[1;32mreturn\u001b[0m \u001b[0marray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0ma\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mcopy\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mFalse\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0morder\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0morder\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    502\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    503\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "hist = model.fit_generator(train_generator, validation_data=valid_generator, epochs = 10, \n",
    "                           callbacks=[EarlyStopping(patience=3), ModelCheckpoint(filepath=\"rnn_model.ckpt\", save_weights_only=False)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "850/850 [==============================] - 64s 75ms/step - loss: 10.2186 - val_loss: 7.8661\n",
      "Epoch 2/10\n",
      "850/850 [==============================] - 54s 63ms/step - loss: 6.9825 - val_loss: 10.1133\n",
      "Epoch 3/10\n",
      "850/850 [==============================] - 52s 61ms/step - loss: 6.1236 - val_loss: 14.2986\n",
      "Epoch 4/10\n",
      "850/850 [==============================] - 52s 61ms/step - loss: 5.2178 - val_loss: 12.8146\n",
      "Epoch 5/10\n",
      "850/850 [==============================] - 52s 61ms/step - loss: 4.7625 - val_loss: 9.2727\n",
      "Epoch 6/10\n",
      "329/850 [==========>...................] - ETA: 26s - loss: 4.4986"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-26-2474291f4df1>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mhist2\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmodel_flat\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit_generator\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtrain_generator\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mvalidation_data\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mvalid_generator\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mepochs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m10\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\legacy\\interfaces.py\u001b[0m in \u001b[0;36mwrapper\u001b[1;34m(*args, **kwargs)\u001b[0m\n\u001b[0;32m     89\u001b[0m                 warnings.warn('Update your `' + object_name + '` call to the ' +\n\u001b[0;32m     90\u001b[0m                               'Keras 2 API: ' + signature, stacklevel=2)\n\u001b[1;32m---> 91\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     92\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     93\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\engine\\training.py\u001b[0m in \u001b[0;36mfit_generator\u001b[1;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[0;32m   1416\u001b[0m             \u001b[0muse_multiprocessing\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1417\u001b[0m             \u001b[0mshuffle\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mshuffle\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1418\u001b[1;33m             initial_epoch=initial_epoch)\n\u001b[0m\u001b[0;32m   1419\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1420\u001b[0m     \u001b[1;33m@\u001b[0m\u001b[0minterfaces\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mlegacy_generator_methods_support\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\engine\\training_generator.py\u001b[0m in \u001b[0;36mfit_generator\u001b[1;34m(model, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[0;32m    199\u001b[0m                 \u001b[1;31m# build batch logs\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    200\u001b[0m                 \u001b[0mbatch_logs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m{\u001b[0m\u001b[1;33m}\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 201\u001b[1;33m                 \u001b[1;32mif\u001b[0m \u001b[0mx\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m \u001b[1;32mor\u001b[0m \u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    202\u001b[0m                     \u001b[1;31m# Handle data tensors support when no input given\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    203\u001b[0m                     \u001b[1;31m# step-size = 1 for data tensors\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "hist2 = model_flat.fit_generator(train_generator, validation_data=valid_generator, epochs = 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "1020/1020 [==============================] - 61s 59ms/step - loss: 9.1252 - val_loss: 6.3036\n",
      "Epoch 2/10\n",
      "1020/1020 [==============================] - 60s 59ms/step - loss: 5.7693 - val_loss: 5.7843\n",
      "Epoch 3/10\n",
      "1020/1020 [==============================] - 60s 59ms/step - loss: 5.0111 - val_loss: 8.4845\n",
      "Epoch 4/10\n",
      "1020/1020 [==============================] - 60s 59ms/step - loss: 4.5018 - val_loss: 9.0041\n",
      "Epoch 5/10\n",
      "1020/1020 [==============================] - 60s 58ms/step - loss: 4.5581 - val_loss: 6.3973\n",
      "Epoch 6/10\n",
      "1020/1020 [==============================] - 60s 58ms/step - loss: 4.3193 - val_loss: 7.4391\n",
      "Epoch 7/10\n",
      "1020/1020 [==============================] - 59s 58ms/step - loss: 4.0120 - val_loss: 17.6570\n",
      "Epoch 8/10\n",
      "1020/1020 [==============================] - 60s 59ms/step - loss: 3.4943 - val_loss: 9.9679\n",
      "Epoch 9/10\n",
      "1020/1020 [==============================] - 60s 58ms/step - loss: 3.6226 - val_loss: 10.5491\n",
      "Epoch 10/10\n",
      "1020/1020 [==============================] - 60s 58ms/step - loss: 3.4438 - val_loss: 6.0805\n"
     ]
    }
   ],
   "source": [
    "hist2 = model_flat.fit_generator(train_generator, validation_data=valid_generator, epochs = 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model_frame(history_size):\n",
    "    k.clear_session()\n",
    "    frame_inp = Input(shape=(history_size, 224, 224, 3))\n",
    "    op_flow_inp = Input(shape=(history_size, 224, 224, 3))\n",
    "    filter_size = (3,3)\n",
    "    base_mod = Xception(weights='imagenet', include_top=False, input_shape = (224,224,3))\n",
    "    for l in base_mod.layers: \n",
    "        l.trainable=False\n",
    "    frame = TimeDistributed(BatchNormalization())(frame_inp)\n",
    "    frame = TimeDistributed(base_mod)(frame)\n",
    "    #frame = Reshape(target_shape=(history_size, 2048))(frame)\n",
    "    frame = TimeDistributed(Conv2D(128, (1,1), activation = \"relu\"))(frame)\n",
    "    frame_max = TimeDistributed(GlobalMaxPool2D())(frame)\n",
    "    frame_avg = TimeDistributed(GlobalAvgPool2D())(frame)\n",
    "    \n",
    "    op_flow = TimeDistributed(BatchNormalization())(op_flow_inp)\n",
    "    op_flow = TimeDistributed(Dropout(.3))(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(4, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(8, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(32, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(64, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(Dropout(.3))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(128, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow_max = TimeDistributed(GlobalMaxPool2D())(op_flow)\n",
    "    op_flow_avg = TimeDistributed(GlobalAvgPool2D())(op_flow)\n",
    "    \n",
    "    conc = concatenate([\n",
    "        #frame_max, frame_avg,\n",
    "        op_flow_max, op_flow_avg\n",
    "    ])\n",
    "    conc = BatchNormalization()(conc)\n",
    "    #conc = SpatialDropout1D(.2)(conc)\n",
    "    #conc = CuDNNGRU(256)(conc)\n",
    "    conc = Flatten()(conc)\n",
    "    conc = Dropout(.3)(conc)\n",
    "    conc = Dense(500, activation = \"relu\")(conc)\n",
    "    conc = Dropout(.2)(conc)\n",
    "    conc = Dense(100, activation = \"relu\")(conc)\n",
    "    conc = Dropout(.1)(conc)\n",
    "    result = Dense(1, activation='linear')(conc)\n",
    "    \n",
    "    model = Model(inputs=[frame_inp, op_flow_inp], outputs=[result])\n",
    "    print(model.summary())\n",
    "    model.compile(loss=\"mse\", optimizer='adam')\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_2 (InputLayer)            (None, 3, 224, 224,  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_6 (TimeDistrib (None, 3, 224, 224,  12          input_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_7 (TimeDistrib (None, 3, 224, 224,  0           time_distributed_6[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_8 (TimeDistrib (None, 3, 222, 222,  112         time_distributed_7[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_9 (TimeDistrib (None, 3, 111, 111,  0           time_distributed_8[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_10 (TimeDistri (None, 3, 109, 109,  296         time_distributed_9[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_11 (TimeDistri (None, 3, 54, 54, 8) 0           time_distributed_10[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_12 (TimeDistri (None, 3, 52, 52, 32 2336        time_distributed_11[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_13 (TimeDistri (None, 3, 26, 26, 32 0           time_distributed_12[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_14 (TimeDistri (None, 3, 24, 24, 64 18496       time_distributed_13[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_15 (TimeDistri (None, 3, 24, 24, 64 0           time_distributed_14[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_16 (TimeDistri (None, 3, 12, 12, 64 0           time_distributed_15[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_17 (TimeDistri (None, 3, 10, 10, 12 73856       time_distributed_16[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_18 (TimeDistri (None, 3, 5, 5, 128) 0           time_distributed_17[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_19 (TimeDistri (None, 3, 128)       0           time_distributed_18[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_20 (TimeDistri (None, 3, 128)       0           time_distributed_18[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_1 (Concatenate)     (None, 3, 256)       0           time_distributed_19[0][0]        \n",
      "                                                                 time_distributed_20[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_7 (BatchNor (None, 3, 256)       1024        concatenate_1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "flatten_1 (Flatten)             (None, 768)          0           batch_normalization_7[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "dropout_3 (Dropout)             (None, 768)          0           flatten_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 500)          384500      dropout_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_4 (Dropout)             (None, 500)          0           dense_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 100)          50100       dropout_4[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_5 (Dropout)             (None, 100)          0           dense_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_3 (Dense)                 (None, 1)            101         dropout_5[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 530,833\n",
      "Trainable params: 530,315\n",
      "Non-trainable params: 518\n",
      "__________________________________________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "model3 = build_model_frame(time_history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "1020/1020 [==============================] - 163s 159ms/step - loss: 28.2054 - val_loss: 12.9030\n",
      "Epoch 2/10\n",
      "1020/1020 [==============================] - 160s 157ms/step - loss: 13.6002 - val_loss: 18.9242\n",
      "Epoch 3/10\n",
      "1020/1020 [==============================] - 158s 155ms/step - loss: 10.7904 - val_loss: 11.0065\n",
      "Epoch 4/10\n",
      "1020/1020 [==============================] - 159s 156ms/step - loss: 8.6554 - val_loss: 13.5720\n",
      "Epoch 5/10\n",
      "1020/1020 [==============================] - 160s 156ms/step - loss: 7.8862 - val_loss: 5.3888\n",
      "Epoch 6/10\n",
      "1020/1020 [==============================] - 161s 157ms/step - loss: 6.3103 - val_loss: 7.8890\n",
      "Epoch 7/10\n",
      "1020/1020 [==============================] - 160s 157ms/step - loss: 5.9375 - val_loss: 7.0740\n",
      "Epoch 8/10\n",
      "1020/1020 [==============================] - 160s 156ms/step - loss: 5.4968 - val_loss: 9.8594\n",
      "Epoch 9/10\n",
      "1020/1020 [==============================] - 160s 157ms/step - loss: 4.8999 - val_loss: 8.6684\n",
      "Epoch 10/10\n",
      "1020/1020 [==============================] - 160s 157ms/step - loss: 4.6090 - val_loss: 6.5605\n"
     ]
    }
   ],
   "source": [
    "hist3 = model3.fit_generator(train_generator, validation_data=valid_generator, epochs = 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model_frame(history_size):\n",
    "    k.clear_session()\n",
    "    frame_inp = Input(shape=(history_size, 224, 224, 3))\n",
    "    op_flow_inp = Input(shape=(history_size, 224, 224, 3))\n",
    "    filter_size = (3,3)\n",
    "    base_mod = Xception(weights='imagenet', include_top=False, input_shape = (224,224,3))\n",
    "    for l in base_mod.layers: \n",
    "        l.trainable=False\n",
    "    #frame = TimeDistributed(BatchNormalization())(frame_inp)\n",
    "    frame = TimeDistributed(base_mod)(frame_inp)\n",
    "    #frame = Reshape(target_shape=(history_size, 2048))(frame)\n",
    "    frame = TimeDistributed(Conv2D(128, (1,1), activation = \"relu\"))(frame)\n",
    "    frame_max = TimeDistributed(GlobalMaxPool2D())(frame)\n",
    "    frame_avg = TimeDistributed(GlobalAvgPool2D())(frame)\n",
    "    \n",
    "    op_flow = TimeDistributed(BatchNormalization())(op_flow_inp)\n",
    "    op_flow = TimeDistributed(Dropout(.3))(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(4, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(8, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(32, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(64, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(Dropout(.3))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow = TimeDistributed(Conv2D(128, filter_size, activation = \"relu\", data_format = \"channels_last\"))(op_flow)\n",
    "    op_flow = TimeDistributed(MaxPool2D())(op_flow)\n",
    "    op_flow_max = TimeDistributed(GlobalMaxPool2D())(op_flow)\n",
    "    op_flow_avg = TimeDistributed(GlobalAvgPool2D())(op_flow)\n",
    "    \n",
    "    conc = concatenate([\n",
    "        frame_max, frame_avg,\n",
    "        #op_flow_max, op_flow_avg\n",
    "    ])\n",
    "    conc = BatchNormalization()(conc)\n",
    "    #conc = SpatialDropout1D(.2)(conc)\n",
    "    #conc = CuDNNGRU(256)(conc)\n",
    "    conc = Flatten()(conc)\n",
    "    conc = Dropout(.3)(conc)\n",
    "    conc = Dense(500, activation = \"relu\")(conc)\n",
    "    conc = Dropout(.2)(conc)\n",
    "    conc = Dense(100, activation = \"relu\")(conc)\n",
    "    conc = Dropout(.1)(conc)\n",
    "    result = Dense(1, activation='linear')(conc)\n",
    "    \n",
    "    model = Model(inputs=[frame_inp, op_flow_inp], outputs=[result])\n",
    "    print(model.summary())\n",
    "    model.compile(loss=\"mse\", optimizer='adam')\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            (None, 3, 224, 224,  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_1 (TimeDistrib (None, 3, 7, 7, 2048 20861480    input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_2 (TimeDistrib (None, 3, 7, 7, 128) 262272      time_distributed_1[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_3 (TimeDistrib (None, 3, 128)       0           time_distributed_2[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "time_distributed_4 (TimeDistrib (None, 3, 128)       0           time_distributed_2[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_1 (Concatenate)     (None, 3, 256)       0           time_distributed_3[0][0]         \n",
      "                                                                 time_distributed_4[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_6 (BatchNor (None, 3, 256)       1024        concatenate_1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "flatten_1 (Flatten)             (None, 768)          0           batch_normalization_6[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "dropout_3 (Dropout)             (None, 768)          0           flatten_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 500)          384500      dropout_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_4 (Dropout)             (None, 500)          0           dense_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 100)          50100       dropout_4[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_5 (Dropout)             (None, 100)          0           dense_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_3 (Dense)                 (None, 1)            101         dropout_5[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 21,559,477\n",
      "Trainable params: 697,485\n",
      "Non-trainable params: 20,861,992\n",
      "__________________________________________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "model4 = build_model_frame(time_history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "1020/1020 [==============================] - 278s 273ms/step - loss: 13.5584 - val_loss: 86.6192\n",
      "Epoch 2/10\n",
      " 146/1020 [===>..........................] - ETA: 3:08 - loss: 7.3752"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-61-5f6e5c70b877>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mhist4\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmodel4\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit_generator\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtrain_generator\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mvalidation_data\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mvalid_generator\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mepochs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m10\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\legacy\\interfaces.py\u001b[0m in \u001b[0;36mwrapper\u001b[1;34m(*args, **kwargs)\u001b[0m\n\u001b[0;32m     89\u001b[0m                 warnings.warn('Update your `' + object_name + '` call to the ' +\n\u001b[0;32m     90\u001b[0m                               'Keras 2 API: ' + signature, stacklevel=2)\n\u001b[1;32m---> 91\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     92\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     93\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\engine\\training.py\u001b[0m in \u001b[0;36mfit_generator\u001b[1;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[0;32m   1416\u001b[0m             \u001b[0muse_multiprocessing\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1417\u001b[0m             \u001b[0mshuffle\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mshuffle\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1418\u001b[1;33m             initial_epoch=initial_epoch)\n\u001b[0m\u001b[0;32m   1419\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1420\u001b[0m     \u001b[1;33m@\u001b[0m\u001b[0minterfaces\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mlegacy_generator_methods_support\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\engine\\training_generator.py\u001b[0m in \u001b[0;36mfit_generator\u001b[1;34m(model, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[0;32m    215\u001b[0m                 outs = model.train_on_batch(x, y,\n\u001b[0;32m    216\u001b[0m                                             \u001b[0msample_weight\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 217\u001b[1;33m                                             class_weight=class_weight)\n\u001b[0m\u001b[0;32m    218\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    219\u001b[0m                 \u001b[0mouts\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mto_list\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\engine\\training.py\u001b[0m in \u001b[0;36mtrain_on_batch\u001b[1;34m(self, x, y, sample_weight, class_weight)\u001b[0m\n\u001b[0;32m   1215\u001b[0m             \u001b[0mins\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mx\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0my\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1216\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_make_train_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1217\u001b[1;33m         \u001b[0moutputs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mins\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1218\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0munpack_singleton\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1219\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py\u001b[0m in \u001b[0;36m__call__\u001b[1;34m(self, inputs)\u001b[0m\n\u001b[0;32m   2713\u001b[0m                 \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_legacy_call\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2714\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2715\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_call\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2716\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2717\u001b[0m             \u001b[1;32mif\u001b[0m \u001b[0mpy_any\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mis_tensor\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[1;32min\u001b[0m \u001b[0minputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py\u001b[0m in \u001b[0;36m_call\u001b[1;34m(self, inputs)\u001b[0m\n\u001b[0;32m   2673\u001b[0m             \u001b[0mfetched\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_callable_fn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0marray_vals\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrun_metadata\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2674\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2675\u001b[1;33m             \u001b[0mfetched\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_callable_fn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0marray_vals\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2676\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mfetched\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2677\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mc:\\users\\magic\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\tensorflow\\python\\client\\session.py\u001b[0m in \u001b[0;36m__call__\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1437\u001b[0m           ret = tf_session.TF_SessionRunCallable(\n\u001b[0;32m   1438\u001b[0m               \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_handle\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mstatus\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1439\u001b[1;33m               run_metadata_ptr)\n\u001b[0m\u001b[0;32m   1440\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1441\u001b[0m           \u001b[0mproto_data\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "hist4 = model4.fit_generator(train_generator, validation_data=valid_generator, epochs = 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}