temperature-forecasting-using-rnns.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "view-in-github"
   },
   "source": [
    "[View in Colaboratory](https://colab.research.google.com/github/silentvoice/dl-with-keras/blob/master/temperature-forecasting-using-rnns.ipynb)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "QFiciD62GoJs"
   },
   "source": [
    "## Check Tensorflow Version"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "AQ65ShNiVf1e"
   },
   "outputs": [],
   "source": [
    "#!kill -9 -1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "CDoELe4vHHUh"
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "QVVOM-TbHLQO",
    "outputId": "3d5eb801-cd5e-477f-9666-23a22bcfacc9"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/device:GPU:0'"
      ]
     },
     "execution_count": 5,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf.test.gpu_device_name()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "X22hst3SHV8o"
   },
   "source": [
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "NUobtRG4HoCg"
   },
   "outputs": [],
   "source": [
    "from tensorflow import keras\n",
    "from tensorflow.python.keras import models, layers\n",
    "from tensorflow.python.keras.models import Sequential\n",
    "from tensorflow.python.keras.layers import GRU, Embedding, Flatten, Dense\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import os"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "0nTVXHeFHo0T"
   },
   "source": [
    "## Data Loading"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "I2Xcyevek3UE",
    "outputId": "73b0f5c7-6833-42bf-b240-d8823624a0ff"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--2018-07-12 04:55:43--  https://s3.amazonaws.com/keras-datasets/jena_climate_2009_2016.csv.zip\n",
      "Resolving s3.amazonaws.com (s3.amazonaws.com)... 52.216.168.21\n",
      "Connecting to s3.amazonaws.com (s3.amazonaws.com)|52.216.168.21|:443... connected.\n",
      "HTTP request sent, awaiting response... 200 OK\n",
      "Length: 13568290 (13M) [application/zip]\n",
      "Saving to: ‘jena_climate_2009_2016.csv.zip’\n",
      "\n",
      "jena_climate_2009_2 100%[===================>]  12.94M  28.8MB/s    in 0.4s    \n",
      "\n",
      "2018-07-12 04:55:44 (28.8 MB/s) - ‘jena_climate_2009_2016.csv.zip’ saved [13568290/13568290]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "!wget https://s3.amazonaws.com/keras-datasets/jena_climate_2009_2016.csv.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "colab_type": "code",
    "id": "Dhde4mXdk_vq",
    "outputId": "8d25684d-66c5-4f76-8e0d-87efad7d3894"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Archive:  jena_climate_2009_2016.csv.zip\n",
      "  inflating: jena_climate_2009_2016.csv  \n"
     ]
    }
   ],
   "source": [
    "!unzip jena_climate_2009_2016.csv.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "9Ili20fWlLwJ"
   },
   "outputs": [],
   "source": [
    "data_dir = '/content/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "usF8mxrVldn7"
   },
   "outputs": [],
   "source": [
    "f_name = os.path.join(data_dir, 'jena_climate_2009_2016.csv')\n",
    "f = open(f_name)\n",
    "data = f.read()\n",
    "f.close()\n",
    "\n",
    "lines = data.split('\\n')\n",
    "header = lines[0].split(',')\n",
    "lines = lines[1:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 272
    },
    "colab_type": "code",
    "id": "d4jiC3fNmDb9",
    "outputId": "19663667-fec2-4015-f8ee-0c25ec71fa9d"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['\"Date Time\"',\n",
       " '\"p (mbar)\"',\n",
       " '\"T (degC)\"',\n",
       " '\"Tpot (K)\"',\n",
       " '\"Tdew (degC)\"',\n",
       " '\"rh (%)\"',\n",
       " '\"VPmax (mbar)\"',\n",
       " '\"VPact (mbar)\"',\n",
       " '\"VPdef (mbar)\"',\n",
       " '\"sh (g/kg)\"',\n",
       " '\"H2OC (mmol/mol)\"',\n",
       " '\"rho (g/m**3)\"',\n",
       " '\"wv (m/s)\"',\n",
       " '\"max. wv (m/s)\"',\n",
       " '\"wd (deg)\"']"
      ]
     },
     "execution_count": 9,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "header"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "PC8VkaBdmHMu",
    "outputId": "adc8c7a2-adb7-4eaf-f2a4-fdddc8b78c35"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "420551"
      ]
     },
     "execution_count": 10,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(lines)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "UpRJGQGemOE7"
   },
   "outputs": [],
   "source": [
    "data = np.zeros((len(lines), len(header) - 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "74DgPhvKmm0E"
   },
   "outputs": [],
   "source": [
    "for idx, line in enumerate(lines):\n",
    "  vals = [float(x) for x in line.split(',')[1:]]\n",
    "  data[idx, :] = vals\n",
    "  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "w55MB1AonUzj",
    "outputId": "51b47dc5-b91d-4b35-cc39-dd841d2e31ed"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(420551, 14)"
      ]
     },
     "execution_count": 13,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "an-ibBr-qGJw"
   },
   "source": [
    "## Data Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "h8RcMogSqMEu"
   },
   "outputs": [],
   "source": [
    "mean = data[:200000].mean(axis=0)\n",
    "data -= mean\n",
    "std = data[:200000].std(axis=0)\n",
    "data /= std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "lA1pw1t0rXZc"
   },
   "outputs": [],
   "source": [
    "def data_generator(data, lookback, delay, min_idx, max_idx, shuffle=False, batch_size=128, step=6):\n",
    "  if max_idx is None:\n",
    "    max_idx = len(data) - delay -1\n",
    "  idx = min_idx + lookback\n",
    "  while True:\n",
    "    if shuffle:\n",
    "      rows = np.random.randint(min_idx + lookback, max_idx, size=batch_size)\n",
    "    else:\n",
    "      if idx + batch_size >= max_idx:\n",
    "        idx = min_idx + lookback\n",
    "      rows = np.arange(idx, min(idx + batch_size, max_idx))\n",
    "      idx += len(rows)\n",
    "    samples = np.zeros((len(rows), lookback // step, data.shape[-1]))\n",
    "    targets = np.zeros((len(rows),))\n",
    "    \n",
    "    for i in range(len(rows)):\n",
    "      indices = range(rows[i] - lookback, rows[i], step)\n",
    "      samples[i] = data[indices]\n",
    "      targets[i] = data[rows[i] + delay][1]\n",
    "    yield samples, targets  \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "foRijY5Dvhj8"
   },
   "outputs": [],
   "source": [
    "lookback = 1440\n",
    "step = 6\n",
    "delay = 144\n",
    "batch_size = 128"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ZiW4WpInvnYC"
   },
   "outputs": [],
   "source": [
    "train_data_generator = data_generator(data,\n",
    "                                      lookback=lookback,\n",
    "                                      delay=delay,\n",
    "                                      min_idx=0,\n",
    "                                      max_idx=200000,\n",
    "                                      shuffle=True,\n",
    "                                      step=step, \n",
    "                                      batch_size=batch_size)\n",
    "val_data_generator = data_generator(data,\n",
    "                                    lookback=lookback,\n",
    "                                    delay=delay,\n",
    "                                    min_idx=200001,\n",
    "                                    max_idx=300000,\n",
    "                                    step=step,\n",
    "                                    batch_size=batch_size)\n",
    "test_data_generator = data_generator(data,\n",
    "                                     lookback=lookback,\n",
    "                                     delay=delay,\n",
    "                                     min_idx=300001,\n",
    "                                     max_idx=None,\n",
    "                                     step=step,\n",
    "                                     batch_size=batch_size)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "E2yg9KU8wYxu"
   },
   "outputs": [],
   "source": [
    "val_steps = (300000 - 200001 - lookback) // batch_size\n",
    "test_steps = (len(data) - 300001 - lookback) // batch_size"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "PTsEqpbkw3O0"
   },
   "source": [
    "## Calculate Mean absolute Error to Determine base line"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nYGMyvsywai7"
   },
   "outputs": [],
   "source": [
    "def evaluate_maes():\n",
    "  batch_maes = []\n",
    "  for _ in range(val_steps):\n",
    "    samples, targets = next(val_data_generator)\n",
    "    preds = samples[:, -1, 1]\n",
    "    mae = np.mean(np.abs(preds - targets))\n",
    "    batch_maes.append(mae)\n",
    "  return np.mean(batch_maes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "lbK1oNVJx1Ey",
    "outputId": "c2b30a5b-e94b-41e2-9dc9-d66b3c9e83d4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.2897359729905486"
      ]
     },
     "execution_count": 20,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "evaluate_maes()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "UyIVB3r23W70",
    "outputId": "988fd6ef-7c8a-419f-b921-192a6183c9e2"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.5645689841146786"
      ]
     },
     "execution_count": 21,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(0.2897 * std[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "scCUF_gSI9MF"
   },
   "source": [
    "## Build Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "6ocS5JqiJE7M"
   },
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(layers.GRU(32, input_shape=(None, data.shape[-1])))\n",
    "model.add(layers.Dense(1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "-IHqOA7-JIE5"
   },
   "outputs": [],
   "source": [
    "model.compile(optimizer='rmsprop', loss='mae')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 717
    },
    "colab_type": "code",
    "id": "J7VBuCsnKMJE",
    "outputId": "1b25b017-e798-46fe-9df6-906ab461d090"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "500/500 [==============================] - 381s 761ms/step - loss: 0.2985 - val_loss: 0.2690\n",
      "Epoch 2/20\n",
      "335/500 [===================>..........] - ETA: 1:27 - loss: 0.2878500/500 [==============================] - 379s 759ms/step - loss: 0.2848 - val_loss: 0.2701\n",
      "Epoch 3/20\n",
      "500/500 [==============================] - 379s 757ms/step - loss: 0.2776 - val_loss: 0.2684\n",
      "Epoch 4/20\n",
      " 11/500 [..............................] - ETA: 4:18 - loss: 0.2801500/500 [==============================] - 378s 755ms/step - loss: 0.2725 - val_loss: 0.2617\n",
      "Epoch 5/20\n",
      "392/500 [======================>.......] - ETA: 56s - loss: 0.2681500/500 [==============================] - 379s 757ms/step - loss: 0.2673 - val_loss: 0.2669\n",
      "Epoch 6/20\n",
      "500/500 [==============================] - 378s 756ms/step - loss: 0.2637 - val_loss: 0.2650\n",
      "Epoch 7/20\n",
      " 26/500 [>.............................] - ETA: 4:07 - loss: 0.2548500/500 [==============================] - 377s 755ms/step - loss: 0.2578 - val_loss: 0.2620\n",
      "Epoch 8/20\n",
      "398/500 [======================>.......] - ETA: 53s - loss: 0.2549500/500 [==============================] - 377s 755ms/step - loss: 0.2545 - val_loss: 0.2651\n",
      "Epoch 9/20\n",
      "500/500 [==============================] - 379s 757ms/step - loss: 0.2503 - val_loss: 0.2695\n",
      "Epoch 10/20\n",
      " 27/500 [>.............................] - ETA: 4:10 - loss: 0.2548500/500 [==============================] - 378s 756ms/step - loss: 0.2465 - val_loss: 0.2702\n",
      "Epoch 11/20\n",
      "396/500 [======================>.......] - ETA: 54s - loss: 0.2437500/500 [==============================] - 377s 755ms/step - loss: 0.2429 - val_loss: 0.2713\n",
      "Epoch 12/20\n",
      "500/500 [==============================] - 378s 756ms/step - loss: 0.2390 - val_loss: 0.2840\n",
      "Epoch 13/20\n",
      " 24/500 [>.............................] - ETA: 4:10 - loss: 0.2398500/500 [==============================] - 379s 757ms/step - loss: 0.2361 - val_loss: 0.2772\n",
      "Epoch 14/20\n",
      "395/500 [======================>.......] - ETA: 55s - loss: 0.2305500/500 [==============================] - 377s 755ms/step - loss: 0.2306 - val_loss: 0.2884\n",
      "Epoch 15/20\n",
      "500/500 [==============================] - 372s 743ms/step - loss: 0.2285 - val_loss: 0.2861\n",
      "Epoch 16/20\n",
      " 24/500 [>.............................] - ETA: 4:10 - loss: 0.2305500/500 [==============================] - 375s 750ms/step - loss: 0.2244 - val_loss: 0.2837\n",
      "Epoch 17/20\n",
      "395/500 [======================>.......] - ETA: 55s - loss: 0.2208500/500 [==============================] - 375s 751ms/step - loss: 0.2210 - val_loss: 0.2890\n",
      "Epoch 18/20\n",
      "500/500 [==============================] - 375s 751ms/step - loss: 0.2173 - val_loss: 0.2942\n",
      "Epoch 19/20\n",
      " 24/500 [>.............................] - ETA: 4:10 - loss: 0.2165500/500 [==============================] - 376s 751ms/step - loss: 0.2175 - val_loss: 0.3000\n",
      "Epoch 20/20\n",
      "395/500 [======================>.......] - ETA: 54s - loss: 0.2119500/500 [==============================] - 376s 751ms/step - loss: 0.2123 - val_loss: 0.2945\n"
     ]
    }
   ],
   "source": [
    "model_history = model.fit_generator(train_data_generator,\n",
    "                              steps_per_epoch=500,\n",
    "                              epochs=20,\n",
    "                              validation_data=val_data_generator,\n",
    "                              validation_steps=val_steps)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "fW9ynACXKvCh"
   },
   "source": [
    "## Plotting Loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 379
    },
    "colab_type": "code",
    "id": "MreDgsqrKOKX",
    "outputId": "34a874e4-4c79-4ede-9677-6bedf92fbd89"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fb1bb2fde80>"
      ]
     },
     "execution_count": 27,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAFZCAYAAACizedRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fb1b92c0eb8>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_loss = model_history.history['loss']\n",
    "val_loss = model_history.history['val_loss']\n",
    "\n",
    "epochs = range(len(train_loss))\n",
    "\n",
    "plt.plot(epochs, train_loss, 'r', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "TaapyuT7KiDP"
   },
   "source": [
    "## Increase the depth and regularize the network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "UnGBb8J9K2aw"
   },
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(layers.GRU(32,\n",
    "                     dropout=0.1,\n",
    "                     recurrent_dropout=0.5,\n",
    "                     return_sequences=True,\n",
    "                     input_shape=(None, data.shape[-1])))\n",
    "model.add(layers.GRU(64, activation='relu',\n",
    "                     dropout=0.1, \n",
    "                     recurrent_dropout=0.5))\n",
    "model.add(layers.Dense(1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ywUZk-wOLAqH"
   },
   "outputs": [],
   "source": [
    "model.compile(optimizer='rmsprop', loss='mae')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "1_GyW0uQLRep"
   },
   "source": [
    "## Plotting Loss of Regularized Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 717
    },
    "colab_type": "code",
    "id": "UchDhgsiLIKo",
    "outputId": "0fb31adb-8a99-4044-be6f-2e502ca0c933"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "500/500 [==============================] - 860s 2s/step - loss: 0.3364 - val_loss: 0.2853\n",
      "Epoch 2/20\n",
      "340/500 [===================>..........] - ETA: 3:10 - loss: 0.3135500/500 [==============================] - 857s 2s/step - loss: 0.3121 - val_loss: 0.2811\n",
      "Epoch 3/20\n",
      "500/500 [==============================] - 857s 2s/step - loss: 0.3035 - val_loss: 0.2704\n",
      "Epoch 4/20\n",
      " 18/500 [>.............................] - ETA: 9:33 - loss: 0.3028500/500 [==============================] - 857s 2s/step - loss: 0.3021 - val_loss: 0.2699\n",
      "Epoch 5/20\n",
      "402/500 [=======================>......] - ETA: 1:56 - loss: 0.2979500/500 [==============================] - 857s 2s/step - loss: 0.2978 - val_loss: 0.2708\n",
      "Epoch 6/20\n",
      "500/500 [==============================] - 860s 2s/step - loss: 0.2959 - val_loss: 0.2699\n",
      "Epoch 7/20\n",
      " 33/500 [>.............................] - ETA: 9:14 - loss: 0.2904500/500 [==============================] - 858s 2s/step - loss: 0.2933 - val_loss: 0.2681\n",
      "Epoch 8/20\n",
      "408/500 [=======================>......] - ETA: 1:49 - loss: 0.2906500/500 [==============================] - 857s 2s/step - loss: 0.2905 - val_loss: 0.2775\n",
      "Epoch 9/20\n",
      "500/500 [==============================] - 859s 2s/step - loss: 0.2888 - val_loss: 0.2646\n",
      "Epoch 10/20\n",
      " 35/500 [=>............................] - ETA: 9:11 - loss: 0.2830500/500 [==============================] - 858s 2s/step - loss: 0.2855 - val_loss: 0.2656\n",
      "Epoch 11/20\n",
      "406/500 [=======================>......] - ETA: 1:51 - loss: 0.2830500/500 [==============================] - 856s 2s/step - loss: 0.2825 - val_loss: 0.2678\n",
      "Epoch 12/20\n",
      "500/500 [==============================] - 860s 2s/step - loss: 0.2817 - val_loss: 0.2685\n",
      "Epoch 13/20\n",
      " 32/500 [>.............................] - ETA: 9:16 - loss: 0.2810500/500 [==============================] - 857s 2s/step - loss: 0.2818 - val_loss: 0.2659\n",
      "Epoch 14/20\n",
      "405/500 [=======================>......] - ETA: 1:52 - loss: 0.2779500/500 [==============================] - 855s 2s/step - loss: 0.2779 - val_loss: 0.2736\n",
      "Epoch 15/20\n",
      "500/500 [==============================] - 854s 2s/step - loss: 0.2761 - val_loss: 0.2677\n",
      "Epoch 16/20\n",
      " 33/500 [>.............................] - ETA: 9:12 - loss: 0.2790500/500 [==============================] - 856s 2s/step - loss: 0.2771 - val_loss: 0.2667\n",
      "Epoch 17/20\n",
      "405/500 [=======================>......] - ETA: 1:53 - loss: 0.2771500/500 [==============================] - 859s 2s/step - loss: 0.2763 - val_loss: 0.2656\n",
      "Epoch 18/20\n",
      "500/500 [==============================] - 861s 2s/step - loss: 0.2759 - val_loss: 0.2652\n",
      "Epoch 19/20\n",
      " 32/500 [>.............................] - ETA: 9:19 - loss: 0.2706500/500 [==============================] - 863s 2s/step - loss: 0.2734 - val_loss: 0.2732\n",
      "Epoch 20/20\n",
      "405/500 [=======================>......] - ETA: 1:53 - loss: 0.2726500/500 [==============================] - 861s 2s/step - loss: 0.2727 - val_loss: 0.2682\n"
     ]
    }
   ],
   "source": [
    "model_history = model.fit_generator(train_data_generator,\n",
    "                              steps_per_epoch=500,\n",
    "                              epochs=20,\n",
    "                              validation_data=val_data_generator,\n",
    "                              validation_steps=val_steps)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 379
    },
    "colab_type": "code",
    "id": "ox3aOUWzLaLc",
    "outputId": "cda4076c-229d-4312-e9ab-2cd40ae87980"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f6865c560f0>"
      ]
     },
     "execution_count": 25,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAFZCAYAAACizedRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzs3Xd4FOXax/HvbE0nhQQIokKkRpFe\nDFITCAJyRIEIEj0qqICIoB5AMCiCoKAIR1EEPDQliBFUSmgiSJGSSAcpShNe0pNN3/L+EYxGUihJ\nZndzf67Li2yZ2fvejfntzDPzjGKz2WwIIYQQwm5o1C5ACCGEEEVJOAshhBB2RsJZCCGEsDMSzkII\nIYSdkXAWQggh7IyEsxBCCGFnJJyFU4uKiiI8PJzw8HCCg4Pp0qVL4W2TyXRT6woPDycxMbHU58ya\nNYsvv/zydkoud0899RQxMTHlsq6GDRty5coVNm3axPjx42/r9VauXFn48428tzdq3LhxfPzxx+Wy\nLiHUolO7ACEq0ptvvln4c9euXXn33Xdp1arVLa1rw4YNZT5n7Nixt7RuRxMWFkZYWNgtL5+QkMCC\nBQsYMGAAcGPvrRBViWw5iyptyJAhfPDBB/Ts2ZO4uDgSExN55plnCA8Pp2vXrnz++eeFz/1zq/Hn\nn39m4MCBzJo1i549e9K1a1f27t0LFN1q69q1KytWrOCxxx6jQ4cOTJ8+vXBdn3zyCe3bt+fRRx9l\n+fLldO3atdj6vvrqK3r27En37t0ZPHgwly5dAiAmJoZRo0YxYcIEevTowUMPPcSpU6cAuHDhAv37\n9yc0NJSxY8disViuW++PP/5Inz59itzXt29ftm/fXup78KeYmBieeuqpMl9vy5Yt9OnThx49etCv\nXz+OHz8OQEREBH/88Qfh4eHk5eUVvrcAS5Ys4aGHHiI8PJwXXniB5OTkwvd2zpw5/Pvf/6ZLly78\n+9//Jjs7u6SPFoATJ04QERFBeHg4ffv2ZceOHQBkZmYyYsQIevbsSbdu3Zg4cSL5+fkl3i9EZZNw\nFlXekSNHWLt2LS1atGDevHnccccdbNiwgcWLFzNr1iwuX7583TLHjh3j/vvvZ/369QwaNIh58+YV\nu+59+/YRHR3N119/zbJly7hy5QqnTp1iwYIFrFmzhi+++KLErcakpCTeeustPv/8czZu3Midd95Z\nZHft9u3bGTRoELGxsbRt25bFixcDMHPmTNq3b8/mzZt58skniYuLu27d7du358qVK1y4cAEoCNgr\nV67wwAMP3PB78KeSXs9sNjNu3DimTJlCbGwsXbt2ZcaMGQBMmzaNWrVqsWHDBgwGQ+G6fvnlFxYu\nXMjSpUvZsGEDgYGBzJo1q/DxDRs28MEHH7Bp0yaSk5PZtGlTiXVZrVbGjBnDE088wYYNG3j77bcZ\nO3YsJpOJ1atX4+Xlxfr164mNjUWr1XL69OkS7xeiskk4iyqvU6dOaDQF/ytMnDiRSZMmAVCnTh38\n/f25ePHidcu4u7sTGhoKQHBwMH/88Uex6+7Tpw9arZYaNWrg5+fH5cuX2bdvH23atCEgIACj0cij\njz5a7LJ+fn4cOHCAmjVrAtCqVavCMAUICgri3nvvBaBJkyaFAbp//34eeughAJo2bUq9evWuW7fB\nYKBLly5s3boVgM2bNxMaGopOp7vh9+BPJb2eTqdj165dNGvWrNj6i7Nt2zZ69OiBn58fAP3792fn\nzp2Fj3fq1Alvb290Oh0NGjQo9UvDxYsXSUxMpFevXgDcd999BAYGcvjwYXx9fYmPj+enn37CarXy\n5ptv0rhx4xLvF6KyyZizqPKqVatW+PPhw4cLtxQ1Gg0JCQlYrdbrlvH09Cz8WaPRFPscAA8Pj8Kf\ntVotFouF9PT0Iq9Zo0aNYpe1WCzMmTOHrVu3YrFYyMzMpG7dusXW8Oe6AdLS0oq8rpeXV7Hr79Gj\nB0uWLOHJJ59k8+bNDB8+/Kbegz+V9npLly7lm2++IS8vj7y8PBRFKXE9AMnJyQQEBBRZV1JSUpk9\nl7QuT0/PIq/p5eVFcnIyvXr1Ii0tjQ8//JCzZ8/y8MMPM378eHr27Fns/X/fuheiMsiWsxB/8+qr\nr9KjRw9iY2PZsGEDPj4+5f4aHh4eZGVlFd6+evVqsc9bt24dW7duZdmyZcTGxjJq1KgbWr+Xl1eR\nI9H/HLP9pwcffJATJ07w+++/8/vvv9OuXTvg5t+Dkl4vLi6Ozz77jHnz5hEbG8vbb79dZu3Vq1cn\nNTW18HZqairVq1cvc7ni+Pn5kZaWxt+v7ZOamlq4VR4REcFXX33FunXrOHr0KKtXry71fiEqk4Sz\nEH+TlJTEvffei6IofPPNN2RnZxcJ0vLQtGlTfv75Z5KTk8nLyyvxj39SUhK1a9fG19eXlJQU1q9f\nT2ZmZpnrb9asWeFYbFxcHOfPny/2eQaDgQ4dOvDee+/RrVs3tFpt4evezHtQ0uslJyfj5+dHYGAg\n2dnZfPPNN2RlZWGz2dDpdGRlZWE2m4usq3PnzmzatImUlBQAVqxYQadOncrsuTh33HEHNWvWZN26\ndYW1JSYm0rRpUz766CNWrVoFFOy5uOOOO1AUpcT7hahsEs5C/M1LL73EiBEj6NOnD1lZWQwcOJBJ\nkyaVGHC3omnTpjzyyCM88sgjREZG0qVLl2Kf17t3b1JTUwkLC2Ps2LGMHj2aK1euFDnquzivvvoq\nP/zwA6GhoSxfvpwHHnigxOf26NGDzZs307Nnz8L7bvY9KOn1HnzwQQICAggNDeXpp5/mySefxNPT\nk1GjRtGwYUOqVatGSEhIkfH6pk2bMmzYMAYPHkx4eDgZGRm8/PLLpfZbEkVReP/991m2bBk9e/bk\n7bff5sMPP8TNzY2+ffuyZs0aevToQXh4OHq9nr59+5Z4vxCVTZHrOQtR+Ww2W+EW2bZt25g9e7bs\nPhVCFJItZyEqWXJyMu3atePSpUvYbDbWr19feESzEELADYbztGnTGDhwIBERERw6dKjIYytXrmTA\ngAFEREQwefLkIgdf5OTkEBoaWm5TBwrhDHx9fRk9ejRPPfUUPXr0IC0tjRdffFHtsoQQdqTMU6n2\n7t3LuXPniI6O5syZM0yYMIHo6GgAsrOzWbt2LcuXL0ev1xMZGUl8fDwtWrQAYN68eUVOGRFCFHj8\n8cd5/PHH1S5DCGGnytxy3r17d+FkC0FBQaSlpRWeNuHq6srixYvR6/VkZ2djMpnw9/cH4MyZM5w+\nfZrOnTtXXPVCCCGEEyoznBMTE4uc5+jr60tCQkKR58yfP5+wsDDCw8OpU6cOADNmzGDcuHHlXK4Q\nQgjh/G76gLDiDu4eNmwYmzdvZseOHRw4cIDVq1fTrFmzwqC+EWZzyTP9CCGEEFVJmWPOAQEBRa6z\nevXq1cJd16mpqZw6dYrWrVvj4uJCx44diYuL4+jRo1y4cIFt27Zx5coVDAYDNWvWLPV8y5SU8p3o\nwd/fk4SEjHJdpz1wxr6kJ8fhjH1JT47D2fry9/cs8bEyt5xDQkKIjY0F4OjRowQEBBTOo/vnVWf+\nnLXo8OHD1K1bl9mzZ/P111+zcuVK+vfvz/Dhw0sNZiGEEEL8pcwt5xYtWhAcHExERASKohAVFUVM\nTAyenp6EhYUxYsQIIiMj0el0NGzYkG7dulVG3UIIIYTTspsZwsp7V4Wz7f74kzP2JT05DmfsS3py\nHM7W123t1hZCCCFE5ZJwFkIIIeyMhLMQQghhZySchRBCCDtT5tHaQgghxM2aO/cDTp48TnJyEjk5\nOQQG1sbLqxrTpr1X5rLr1n2Hu7sHnToVf63zDz+cRf/+EQQG1r6l2kaOHMaYMa9Rr949t7R8ZZBw\nFkIIUe5efPFloCBoz549w8iRo2942Yce6lPq4y+9NPa2anMEEs5CCCEqTVzcflasWEZWVhYjR75M\nfPwBtm3bgtVqpX37EJ5+ehgLF36Kt7c3desGEROzEkXRcO7cb/Tq9RADBz5ZuOX7ww9byMw0cf78\nOS5dusioUWNp3z6EZcv+x+bNGwkMrI3ZbCYiYjAtWrS6rhaTycTUqZMxmTIwm82MHv0qDRs2Yvbs\n9zhx4jgWi4VHHnmMhx7qU+x9Fckpw1l7/BgcMkHTNmqXIoQQqnOfPBHjd6vLdZ25ff5F5uS3b2nZ\nM2dO8+WXMRgMBuLjD/DxxwvQaDQMGNCXgQMHFXnusWNH+eKLr7FarQwY8DADBz5Z5PGrV/+PmTPn\nsGfPLtas+Zrg4HuJifmKL7/8mszMTCIi+hERMbjYOr766kuCg+/liSee4sSJY8yd+z7Tpr3Hrl0/\nsXLlGsxmM+vWfUd6etp191U0pwxn93fegi2bUE7+js2j5JO8hRBCVL577qmPwWAAwMXFhZEjh6HV\naklNTSU9Pb3Icxs2bISLi0uJ62ratBlQcB0Ik8nExYsXqFcvCKPRBaPRhcaNg0tc9sSJY0RGPgNA\no0ZNuHjxAl5e1ahT5y7GjRtDly6hhIf3wmAwXHdfRXPKcLbc0wA2rEN3YD/5JRxQIIQQVUXm5Ldv\neSu3Iuj1egCuXLlMdPRyFi1ajpubG0OGDLjuuVqtttR1/f1xm82GzQYazV8nIilKycsqilLkSotW\nqxWAWbPmcPLkCTZt2sCGDWv54IOPir2vIjnlqVT5bdsDoP95t8qVCCGEKElqaio+Pj64ublx8uQJ\nrly5Qn5+/m2ts1atWpw9ewaz2UxKSgonThwv8bmNGjUhPn4/AEeOHKZu3SAuX/6Dr75aQcOGjRg5\ncjRpaWnF3lfRnHLLOb91wVizfu/PKlcihBCiJPXrN8DV1Y0XXnia++5rRt++/Zg1awZNm95/y+v0\n9fUjLCycoUMjueuuujRpElzi1veAAY8zbdqbjBr1PFarlTFj/kP16v4cOXKQLVs2otfr6dXr4WLv\nq2jOe+GLTm2xnTtP4ukLoHOe7yDONvE7SE+OxBn7kp4cx432tW7dd4SFhaPVaomMjOD99+cSEFCj\nEiq8OaVd+MJ5UuufQkJQjh9Hd+wI5msHDAghhHB+SUlJDBv2JHq9ge7dw+0ymMvivOHcoQMsWID+\n590SzkIIUYUMGfIUQ4Y8pXYZt8UpDwgDICQEAJ2MOwshhHAwzhvOQUFY/QMKjti2j2F1IYQQ4oY4\nbzgrCvlt2qG9chnNhfNqVyOEEELcMOcNZyC/bTtAzncWQgjhWJw7nNtcC2cZdxZCiEr13HP/vm4C\nkE8++S9ffrms2OfHxe1n4sTXABg3bsx1j3/9dTRz584t8fVOnz7F+fPnAIiKGk9ubs6tls5jj/Uh\nKyvrlpcvD04dzub77sfm6op+r2w5CyFEZQoL68HWrZuK3Ldt21ZCQ7uXuez06e/f9Ov9+ONWLlwb\nwnzzzXcwGkuej9sROO+pVAB6PfktWqHf9RNKago2bx+1KxJCiCqhW7fuvPDCMwwfPgqAEyeO4+/v\nj79/APv2/cyCBZ+g1+vx9PTkrbemF1m2V69urF27hf379zJnzix8ff3w86tO/fr1MJvNTJ06mYSE\nq2RnZ/P008OoWbMWa9bE8OOPW/Hx8eGNN8azZEk0JlMG77zzFvn5+Wg0GsaNm4SiKEydOpnAwNqc\nPn2KBg0aMm7cpGJ7uHr1/65bPiCgBm+9NYmkpETy8vJ45pnnaNWqzXX3tWv3wG29f84dzkB+m7YY\ndu5Av38veaE91C5HCCEq3eTJRr77rnz/3PfpY2by5NwSH/fx8SUwsDbHjh2hSZN72bp1E2Fh4QBk\nZGQQFfU2gYG1mTLlDX7+eTdubm7XrePTT//LpElTqF+/Aa+8Murasum0adOOnj17c+nSRSZNGsei\nRcto27Y9nTt3o0mTewuXX7DgE3r37ku3bt354YfNLFo0n2eeeY6TJ4/z5pvT8PHx5ZFHHiIjIwNP\nz+tn6ypu+f79HyctLZWPPvqMjIwMdu/eyZkzp6+773Y59W5t+OsiGHK+sxBCVK6wsHC2bCnYtb1z\n53Y6d+4GgLe3NzNmvM3IkcOIjz9AenrxF5K4fPky9es3AKBZsxYAeHp6cfz4UV544WmmTp1c4rIA\nJ08ep3nzlgC0aNGKU6dOAlC7dh38/Kqj0WioXt2fzEzTDS9/1113k5WVyZQpk4iL20doaPdi77td\nTr/lbG7VBpuiyBHbQogqa/Lk3FK3citKp05dWLJkEWFhPahT5068vLwAeOedKbz33mzuvrsu778/\no8Tl/37px4LLQChs2rSB9PR0PvpoAenp6Tz77JBSKvjrkpD5+WYUpWB9/7wQRsmXmLh+eRcXFz79\n9H8cPnyI9eu/Y+fOHUyYEFXsfbfD6becbV7VsDQORh9/APLy1C5HCCGqDDc3d4KC6rNkyeeFu7QB\nMjNN1KhRk4yMDOLiDpR4mcjq1f05f/53bDYb8fEHgILLTNaqFYhGo+HHH7cWLqsoChaLpcjyjRs3\nIS6u4JKQv/xygEaNGt9U/cUt/+c1ne+/vxmvvDKe33//rdj7bpfTbzlDwbiz7tgRdId+wdyqjdrl\nCCFElREWFs7bb0cRFTWl8L5+/frzwgvPUKfOnQweHMmiRfMZNmz4dcsOGzaciRP/Q82atQovXtG5\nc1fGjRvDsWNH6NXrYQICAvj888+4//7mzJ79XpGx62effZ533pnCd9+tRqfTM378JMxm8w3XXtzy\nRqMLn376EWvWxKDRaBg0aAi1agVed9/tct5LRv7t0mLGr1fi9cKzmCZPJXv4i+X6OpXNGS8FJz05\nDmfsS3pyHM7WV2mXjHT63drwt8lIZNxZCCGEA7ih3drTpk3j4MGDKIrChAkTaNq0aeFjK1euZNWq\nVWg0Gho1akRUVBQ5OTmMGzeOpKQkcnNzGT58OF26dKmwJspivaMOlsDa6PftKbgIhqKoVosQQghR\nljLDee/evZw7d47o6GjOnDnDhAkTiI6OBiA7O5u1a9eyfPly9Ho9kZGRxMfHc+XKFe69916GDh3K\npUuXePrpp1UN54KLYLTFZXUM2rOnsQTVV68WIYQQogxlhvPu3bsJDQ0FICgoiLS0NEwmEx4eHri6\nurJ48WKgIKhNJhP+/v60aNGicPnLly9To0aNCir/xuW3bY/L6hh0e3+WcBZCCGHXyhxzTkxMxMfn\nr2kvfX19SUhIKPKc+fPnExYWRnh4OHXq1Cm8PyIigldeeYUJEyaUY8m3xlx4EYw9KlcihBBClO6m\nT6Uq7uDuYcOGERkZydChQ2nZsiUtWxbMqLJixQqOHz/Oq6++yrfffotSylivj48bOp22xMdvRZEj\n4Tq2A09PXPf/jGspR8g5gtKO8HNU0pPjcMa+pCfH4ax9/VOZ4RwQEEBiYmLh7atXr+Lv7w8UnAx+\n6tQpWrdujYuLCx07diQuLg6j0Yifnx+1atWicePGWCwWkpOT8fPzK/F1UlLK9/JcxR1yX61FKww/\n/kDi8d+wVa9erq9XWZztVAKQnhyJM/YlPTkOZ+vrtk6lCgkJITY2FoCjR48SEBCAh4cHAGazmXHj\nxpGZmQnA4cOHqVu3Lvv372fRokVAwW7xrKysIrvG1fLnPNv6fTLPthBCCPtV5pZzixYtCA4OJiIi\nAkVRiIqKIiYmBk9PT8LCwhgxYgSRkZHodDoaNmxIt27dyM3N5fXXX2fQoEHk5OTwxhtvFJkjVS35\nfxt3zuvZS+VqhBBCiOJViRnCCplMVK9fB3PzlqSu21yur1dZnG23DkhPjsQZ+5KeHIez9VXlZwgr\n5OGB+d6m6A7GQ3a22tUIIYQQxapa4Qzkt22Hkp+P/mC82qUIIYQQxap64Xxt3Fkn5zsLIYSwU1Uu\nnM1yEQwhhBB2rsqFs7VmLSx33l1wOpXVqnY5QgghxHWqXDhDwbizJjUV7a8n1S5FCCGEuE7VDGeZ\nZ1sIIYQdq5rh/OdMYTLuLIQQwg5VyXC2NGiItZq3bDkLIYSwS1UynNFoyG/TFu2539H83xW1qxFC\nCCGKqJrhjJzvLIQQwn5V2XA2y7izEEIIO1Vlwzn//ubY9HoZdxZCCGF3qmw44+qKuWkzdIcPgcmk\ndjVCCCFEoaobzhScUqVYLOjjD6hdihBCCFGoaoezzLMthBDCDkk4IzOFCSGEsC9VOpxt1atjDroH\n3f59YLGoXY4QQggBVPFwhoJxZ40pA+2xo2qXIoQQQgASzn9d33mvjDsLIYSwD1U+nPPbyrizEEII\n+1Llw9lS7x6s1auj3/uz2qUIIYQQgIQzKAr5rduhvXQRzcULalcjhBBCSDiDnFIlhBDCvkg487dx\nZ5mMRAghhB2QcAbMTZthc3GRcWchhBB2QcIZwGAgv3lLtMeOoKSnqV2NEEKIKk7C+Rpzm3YoNlvB\nbGFCCCGEiiScr/nrfGcZdxZCCKEu3Y08adq0aRw8eBBFUZgwYQJNmzYtfGzlypWsWrUKjUZDo0aN\niIqKQlEU3n33XQ4cOIDZbOa5556je/fuFdZEechv1QZAxp2FEEKorsxw3rt3L+fOnSM6OpozZ84w\nYcIEoqOjAcjOzmbt2rUsX74cvV5PZGQk8fHx5OXlcerUKaKjo0lJSeGRRx6x+3C2eftgbtwE/YF9\nkJ8Per3aJQkhhKiiygzn3bt3ExoaCkBQUBBpaWmYTCY8PDxwdXVl8eLFQEFQm0wm/P39CQwMLNy6\n9vLyIjs7G4vFglarrcBWbl9+63bojh9Dd+QQ5uYt1S5HCCFEFVXmmHNiYiI+Pj6Ft319fUlISCjy\nnPnz5xMWFkZ4eDh16tRBq9Xi5uYGwKpVq+jYsaPdBzPI+c5CCCHsww2NOf+dzWa77r5hw4YRGRnJ\n0KFDadmyJS1bFmx1bt68mVWrVrFo0aIy1+vj44ZOV74B7u/veXML9CzYQ+Bx8AAeN7tsJbrpvhyA\n9OQ4nLEv6clxOGtf/1RmOAcEBJCYmFh4++rVq/j7+wOQmprKqVOnaN26NS4uLnTs2JG4uDhatmzJ\njh07+OSTT1iwYAGenmW/mSkpWbfRxvX8/T1JSMi4uYXc/fCtWQtl+w6SrqaDopRrTeXhlvqyc9KT\n43DGvqQnx+FsfZX2RaPM3dohISHExsYCcPToUQICAvDw8ADAbDYzbtw4MjMzATh8+DB169YlIyOD\nd999l08//RRvb+/y6KFyKAr5bdqhSbiK5vff1K5GCCFEFVXmlnOLFi0IDg4mIiICRVGIiooiJiYG\nT09PwsLCGDFiBJGRkeh0Oho2bEi3bt1YuXIlKSkpjB49unA9M2bMIDAwsEKbKQ/mtu3g22/Q/7yb\n3Lr11C5HCCFEFaTYihtEVkF576q41d0fuoPx+IR1InvIU5hmzSnXmsqDs+3WAenJkThjX9KT43C2\nvm5rt3ZVYw6+D5ubuxyxLYQQQjUSzv+k05HfsjW6X0+iJCepXY0QQogqSMK5GIXnO+/bq3IlQggh\nqiIJ52Lkt/nzIhh7VK5ECCFEVSThXAxzq9bYNBoZdxZCCKEKCedi2Dw8MQffh+6XOMjJUbscIYQQ\nVYyEcwnMbdqi5OWhO/iL2qUIIYSoYiScS5Dftj0g485CCCEqn4RzCf46KEzGnYUQQlQuCecSWANr\nY6lzJ/p9P4N9TKImhBCiipBwLkV+67ZokpPRnj6ldilCCCGqEAnnUhSOO8spVUIIISqRhHMpZDIS\nIYQQapBwLoWlUWOsXtXQyZazEEKISiThXBqtFnOr1uh+O4ty9ara1QghhKgiJJzLIOc7CyGEqGwS\nzmWQcWchhBCVTcK5DPnNW2LT6WQyEiGEEJVGwrksbm6Ym96P7tBByMpSuxohhBBVgITzDchv0x7F\nbEYff0DtUoQQQlQBEs43QMadhRBCVCYJ5xtQGM5yvrMQQohKIOF8A2wBAZjr1kO3fx9YLGqXI4QQ\nwslJON8gc9v2aNLT0J44rnYpQgghnJyE8w3K69ARALdP/qtyJUIIIZydhPMNyu3Xn/ymzXCJ/gLD\nlo1qlyOEEMKJSTjfKJ2OjNkfYdPp8HhlNEpGutoVCSGEcFISzjfBcu99ZI16Ge2li7hPiVK7HCGE\nEE5KwvkmZb38GuaGjXD930L0u35SuxwhhBBO6IbCedq0aQwcOJCIiAgOHTpU5LGVK1cyYMAAIiIi\nmDx5MjabDYBff/2V0NBQli1bVv5Vq8loLNi9rdHg8fJImdJTCCFEuSsznPfu3cu5c+eIjo5m6tSp\nTJ06tfCx7Oxs1q5dy/Lly1mxYgVnz54lPj6erKwspkyZQvv27Su0eLWYW7Yme9hwdL+dxf3daWqX\nI4QQwsmUGc67d+8mNDQUgKCgINLS0jCZTAC4urqyePFi9Ho92dnZmEwm/P39MRgMfPbZZwQEBFRs\n9SrKHDcRy911cf3kv+hkzm0hhBDlSFfWExITEwkODi687evrS0JCAh4eHoX3zZ8/nyVLlhAZGUmd\nOnUKVqwrc9VF+Pi4odNpb2qZsvj7e5br+oryhM8XQZcu+Ix9EQ4cAIOhAl/vLxXblzqkJ8fhjH1J\nT47DWfv6p5tLUCgcU/67YcOGERkZydChQ2nZsiUtW7a86UJSUsp37Nbf35OEhIxyXed1glviEfk0\nrksWkfl6FFmvTajY16OS+qpk0pPjcMa+pCfH4Wx9lfZFo8zd2gEBASQmJhbevnr1Kv7+/gCkpqay\nb98+AFxcXOjYsSNxcXG3W69DyYx6C0tgbdxmz0R77Kja5QghhHACZYZzSEgIsbGxABw9epSAgIDC\nXdpms5lx48aRmZkJwOHDh6lbt24Flmt/bJ5emGbORjGb8Rw9HMxmtUsSQgjh4Mrcrd2iRQuCg4OJ\niIhAURSioqKIiYnB09OTsLAwRowYQWRkJDqdjoYNG9KtWzeOHDnCjBkzuHTpEjqdjtjYWObOnYu3\nt3dl9FTp8kJ7kPPYQFxWReP6yUdkj3xJ7ZKEEEI4MMVW3CCyCsp7HKGyxyaU5CR8O7RBMWWQ8sNO\nLEH1K+R1nG3MBaQnR+KMfUlPjsPZ+rqtMWdxY2y+fmRMn4mSk4PHyy+C1ap2SUIIIRyUhHM5yuvz\nL3If6oNhzy5c/rdQ7XKEEEI4KAnn8qQomGbMwlrNG/cpUWgunFe7IiGEEA5IwrmcWWvUxDTlHTSZ\nJjxfeQnsY0hfCCGEA5FwrgBwjwzxAAAgAElEQVS5AweR1zUUww9bMEZ/oXY5QgghHIyEc0VQFDJm\nfojV3QOPSePR/N8VtSsSQgjhQCScK4j1jjpkTnoTTVoqHv8ZK7u3hRBC3DAJ5wqU89Qz5LUPwbju\nOwzfrVa7HCGEEA5CwrkiaTSYPpiLzcUFz3GvoCQnqV2REEIIByDhXMEs9e4h87XX0SQm4DFxnNrl\nCCGEcAASzpUg+/kR5DdrjsuqaAybY9UuRwghhJ2TcK4MOh0Zsz/Gptfj8cpolIx0tSsSQghhxySc\nK4mlSTBZL41F+8cl3N98Q+1yhBBC2DEJ50qUNfoVzI2b4LpkEfqdO9QuRwghhJ2ScK5MBgMZsz/C\nptHg+fJIyMpSuyIhhBB2SMK5kpmbtyT7hRfR/v4b7tPfVrscIYQQdkjCWQWZr03AXC8I1/kfozuw\nT+1yhBBC2BkJZzW4umL64L8oViueo0dAbq7aFQkhhLAjEs4qyW8fQva/n0V38gTVBj2Gy4JP0B49\nAlar2qUJIYRQmU7tAqqyzElvoos/gGHHjxh2/AiA1ceH/LYPkP9ACPkPdMAcfB9otSpXKoQQojJJ\nOKvI5uFJauw2NOd+R797J4ZdP6HfvRPjhrUYN6wFwOpVjfy27chv34H8B0Kg24MqVy2EEKKiSTir\nTVGw3l2X3Lvrkvv4EwBoLl5Afy2o9bt+wrgpFuOma9N+enhQrXVb8h7oQH77DpibNQeDQcUGhBBC\nlDcJZztkvaMOuQMeJ3fA4wBoLv9xLah34rp3F4YftmD4YQsANjc38lu2KdwNnt+iFRiNapYvhBDi\nNkk4OwBrrUBy+/Unt19/XP09STx6Bv2enRh27ijYHb5jG4Yd2wCwGY3kt2xNXveeZA99HvR6dYsX\nQghx0yScHZAtIIC8hx8h7+FHAFCSkgq2rHf/hGHXzsLxa+P670mf/znWWoEqVyyEEOJmyKlUTsDm\n50de74fJnPouKT/sJOnYWXL69kP/8258uoag37ZV7RKFEELcBAlnJ2Tz8yNj/udkvDMTJT2dagMf\nwe3daWCxqF2aEEKIGyDh7KwUhZxnhpH6/Uasd9TBfeZ0qg3sh5KQoHZlQgghyiDh7OTMzVuSsnk7\nud3DMWz/AZ9uHdDv2aV2WUIIIUpxQ+E8bdo0Bg4cSEREBIcOHSry2MqVKxkwYAARERFMnjwZm81W\n5jKictl8fElfsgLTpLfQJFyl2iO9cJ07G659VkIIIexLmeG8d+9ezp07R3R0NFOnTmXq1KmFj2Vn\nZ7N27VqWL1/OihUrOHv2LPHx8aUuI1Si0ZD94mjSYr7HWt0fjylv4BUZgZKaonZlQggh/qHMcN69\nezehoaEABAUFkZaWhslkAsDV1ZXFixej1+vJzs7GZDLh7+9f6jJCXfntQ0jZupO8BztjjF2PT2hH\ndPEH1C5LCCHE35R5nnNiYiLBwcGFt319fUlISMDDw6Pwvvnz57NkyRIiIyOpU6fODS3zTz4+buh0\n5XuBB39/z3Jdn7247b78PeGHzTBlCtq33sKnTw94/30YPhwUpXyKvNmSnPCzcsaewDn7kp4ch7P2\n9U83PQmJrZhxymHDhhEZGcnQoUNp2bLlDS3zTykpWTdbSqn8/T1JSMgo13Xag3Lta8RY9E2a4TX8\nWTQjR5Kz5QdMs+Zg86jcX35n/KycsSdwzr6kJ8fhbH2V9kWjzN3aAQEBJCYmFt6+evUq/v7+AKSm\nprJv3z4AXFxc6NixI3FxcaUuI+xLfpdupGz5ifw27XD55mu8wzqhPXZU7bKEEKJKKzOcQ0JCiI0t\nuCLS0aNHCQgIKNw9bTabGTduHJmZmQAcPnyYunXrlrqMsD/WwNqkfrOWrOGj0J05jU/PrhhXLFe7\nLCGEqLLK3K3dokULgoODiYiIQFEUoqKiiImJwdPTk7CwMEaMGEFkZCQ6nY6GDRvSrVs3FEW5bhlh\n5/R6Mie/TX7b9ni++Dxeo14ge88uTNPeAzc3tasTQogqRbHdyIBwJSjvcQRnG5v4U2X0pfn9N7yG\nPoX+YDzmxsGkL1qCJah+hb2eM35WztgTOGdf0pPjcLa+bmvMWVQ91rvrkvpdLNn/fhbd8aN4h3bC\nuCZG7bKEEKLKkHAWxXNxwTTjfdI/WYhis+E19Ck8xr8ik5YIIUQlkHAWpcrt15+UTT9ibtQY14Xz\n8Qu+B6/HH8W4YrkEtRBCVJCbPs9ZVD2W+g1IWb8V14XzMa6JwbhlE8Ytm7Dp9eR17Exu337khT+E\nzdtH7VKFEMIpSDiLG+PuTvaol8ke9TKas2cwfrca47erJaiFEKICSDiLm2atF0T2S2PJfmmsBLUQ\nQlQACWdxWySohRCi/Ek4i3IjQS2EEOVDwllUiJsJaiIfR04cEEKIv8hfRFHh/gzq1C07SNoTj+n1\nKMyNmmDcsgmvUS/AHXfgHvU6mj8uqV2qEELYBQlnUan+GdSZ4yaCpydu8+bi2+o+PF98Hu3xY2qX\nKYQQqpJwFqqx1gsia8xr8NtvZMz+CEvderhEf4Fvp3Z4De6PfvdOsI+p34UQolJJOAv1GY3kDBpC\nyo69pC2NJr9te4ybYvHu2xPvh7ph+G4NWCxqVymEEJVGwlnYD42GvB49Sf0ulpTvN5Hbszf6A/up\n9swQfEJa4bLkc8jJUbtKIYSocBLOwi6Z27QlffEXJO/cT/YTT6K9eAHPV17Cr0Uwbh+8h5KSrHaJ\nQghRYSSchV2z1G+A6f25JB84QtaoMZCXh/s7U/BrHoz7pHFoLl5Qu0QhhCh3Es7CIVhr1CRz4mSS\n449imjwVa7VquH36Mb5t7sdz+FC0R4+oXaIQQpQbCWfhUGyeXmQPf5HkfYdInzMPyz31cVkVjW+X\nB6gW0Q/9T9vlCG8hhMOTcBaOyWAgN2IwKdt2k7Z8JXntQzBs3Yx3v9549+iM4dtvwGpVu0ohhLgl\nEs7CsWk05IWFk7ZmPSnrNpPb62F0B3+h2rNP4tO5vYS0EMIhSTgLp2Fu1Yb0z5eRsms/ORGD0Z76\ntSCkuzyA4bvVEtJCCIch4SycjiWoPhlz5pG8cz85Ax5He/IE1Z6JxKdLSMGEJhLSQgg7J+EsnJa1\nXhAZ//2UlJ37yOkfgfbk8YIJTbp2wPD9txLSQgi7JeEsnJ4lqD4ZH80n5ad95Dw2EO2JY1R7+gl8\nuj2IYe13cnS3EMLuSDiLKsNyT30yPv6MlB17yenXH+2xI1T792C8uz2IYf1aCWkhhN2QcBZVjqV+\nAzI+WXgtpB9Dd/Qw1Z58HO/Qjhg2rJOQFkKoTsJZVFmWBg3J+GQRKdt/JueRR9EdOUS1yAi8wzph\niF0vIS2EUI2Es6jyLA0bkfHp56T8uIecvv3QHT5ItSED8e7eGcNGCWkhROW7oXCeNm0aAwcOJCIi\ngkOHDhV5bM+ePQwYMICIiAjGjx+P1WrFarUyadIkIiIiGDJkCGfOnKmQ4oUoT5ZGjcn47H+kbNtN\nzsOPoD8YT7UnBhbMOLZpg4S0EKLSlBnOe/fu5dy5c0RHRzN16lSmTp1a5PE33niDOXPmsGLFCjIz\nM9mxYwdbtmwhIyODFStWMHXqVN59990Ka0CI8mZp3ISMBYtJ3rab3D7/Qv9LPNUGD8C7Z1dclnyO\n5vIfapcohHByurKesHv3bkJDQwEICgoiLS0Nk8mEh4cHADExMYU/+/r6kpKSQkJCAk2bNgXgzjvv\n5I8//sBisaDVaiuqDyHKnaVJMOkLl6A9egT3WTMwfr8GfdwBAPKbNiMvrAd53cMx398cNDJCJIQo\nP2X+RUlMTMTHx6fwtq+vLwkJCYW3/wzmq1evsnPnTjp16kSDBg346aefsFgsnD17lgsXLpCSklIB\n5QtR8SzB95K+aClJew9imjqDvI5d0B0/ivusGfj06ILffQ3wGD0Cw7rvwWRSu1whhBMoc8v5n2zF\njLslJSXx/PPPExUVhY+PD506dSIuLo7BgwfTsGFD6tWrV+xyf+fj44ZOV75b1v7+nuW6PnvhjH05\nRE/+TaF1U5jwGqSnw6ZN8P33aNauxfWLpbh+sRQMBujSBXr3xr93b7j7brWrLncO8VndJOnJcThr\nX/+k2MpIzblz5+Lv709ERAQA3bp1Y82aNYVbzCaTicjISEaPHk3Hjh2LXUdoaCgbN25EU8quv4SE\njFvtoVj+/p7lvk574Ix9OXxPViu6uP0YNm3AuDEW3dHDhQ+ZGzchLyyc3LBwzK1ag4MP7Tj8Z1UM\n6clxOFtfpX3RKHO3dkhICLGxsQAcPXqUgICAwmAGmD59Ok8++WSRYD5x4gTjx48HYPv27TRp0qTU\nYBbCoWk0mFu1IWv8G6T8sJOk+GMwbx65od3Rnj2D25z38enTHb/gIDxHDMO4JgYlPU3tqoUQdqzM\n3dotWrQgODiYiIgIFEUhKiqKmJgYPD096dChA6tXr+bcuXOsWrUKgN69e9O/f39sNhuPPfYYRqOR\nmTNnVngjQtgLa+074PnnSX90MGRmYvhpO4aNGzBs2oDLVytw+WoFNp2O/HYPkNc9nNw+/ypYRggh\nrilzt3Zlkd3aN8YZ+6oyPdls6A4fLAxqfXxcwd1GI1nDXyRr1Fhwd1eh2htXZT4rB+eMPYHz9XVb\nu7WFEOVEUTA3bUbWK+NIjd1G4uFTZLz7AVZfP9w/mInvAy0xxnwlk50IISSchVCLrUYNcp56huRd\nB8h8+RU0yUl4Pf8M3g+Hozt8UO3yhBAqknAWQm3u7mSNf4PkHXvJ7dkb/c+78Q7tiMcro1GSktSu\nTgihAglnIeyE9e66pC/+gtSVq7HUb4DrkkX4tmuOy4JPwGxWuzwhRCWScBbCzuR37krKD7swTXkH\nbDY8J7yGT7cO6Hf8qHZpQohKIuEshD3S68l+bgTJu+PIfuJJtCeO4/1oH7yeHoLm/Dm1qxNCVDAJ\nZyHsmM3fH9P7c0nduI38Vm0wfr8G3w6tcZsxFbKy1C5PCFFBJJyFcADm+5uTunYT6R9/hrWaN+6z\nZuAb0grDt9/IqVdCOCEJZyEchaKQ+9hAUnYfIGvUGDQJV6n27JNUe6QX2qNH1K5OCFGOJJyFcDA2\nD08yJ04mefvP5PboiWHXT/h064DHf8agJMupV0I4AwlnIRyUtV4Q6UujSV3xNZZ6Qbh+vgDf9i1w\n+XwBWCxqlyeEuA1OGc6ZmXD1qtpVCFE58ruGkbJtN6bJUyHfjOd/xuDT7UEMm2NlPFoIB+WU4Tx2\nrAt33glLl+rlb5OoGgwGsoe/WHDq1eNPoD1+lGqD+uPdpwf6XT+pXZ0Q4iY5ZTgPGZKPm1tBSI8Y\n4YLJpHZFQlQOW40amD78mJQfdpEb3gv93j14/+shqg34F7pf4tQuTwhxg5wynENCLMTHQ4sWFlat\n0hMe7sbJk07ZqhDFsjQJJn3Jl6Ss30Jexy4Ytm3Fp3tnvJ4ajPbEcbXLE0KUwWkT66674Ntvs3ju\nuTx+/VVLjx5urFypU7ssISqVuWVr0latITXm+4JJTNZ9h0+ndngOH4rmt7NqlyeEKIHThjOAwQBT\npuSycGE2Wi2MHOnK2LFGsrPVrkyIypXfoSOpazeRtiwaS5N7cVkVjW9IKzxefRnN5T/ULk8I8Q9O\nHc5/6tPHzKZNmdx7r4WlSw089JAbZ88qapclROVSFPK69yRlyw7S53+O5c67cF28EN+2zXCPel0u\nTymEHakS4QxQr56NdeuyiIzM4+hRLaGh7nz7rezmFlWQRkPuvx4l5ad9ZMz+CKtfddzmzcW31X24\nzZiKkp6mdoVCVHlVJpwBXFxg5sxcPv44G6sVnn3WlQkTjOTlqV2ZECrQ6cgZNITkPfFkTHsXXF0L\n5uxu3RTXubPlwhpCqKhKhfOfHnvMzMaNWTRsaGHBAgMPP+zG+fOym1tUUUYjOc8+T9K+Q5gmTgar\nDY8pb+Db5n5cFs5Hvr0KUfmqZDgDNGhgZcOGLPr3zycurmA398aNWrXLEkI97u5kjxpD8v5DZL78\nChqTCc/xr+D7QEuMK5bLlKBCVKIqG84A7u7w3//m8MEHOeTkwBNPuPHWWwby89WuTAj12Kp5kzX+\nDZL2HSLrueForlzGa9QL+HRsC6+/jjH6C3Rx+2VsWogKpNhs9jHBZUJCRrmuz9/f86bWeeSIhmef\ndeXsWQ1t25qZPz+HWrXs4q0p4mb7cgTSk33TXLqI2/vv4vLFUpR/bD1bAmpgqd8AS1B9LPXrY772\ns/WOOqB1jD1RzvRZ/ckZewLn68vf37PExySc/yYjA15+2YVvv9Xj52dl3rwcOne2r115zvbLCdKT\no1ASE6l+5Xcy9h9Ee/pXdKdPoT19Cs2F8yj/+DNic3HBUjeoIKzvqV/wX/0GmIPqg4eHSh0Uzxk/\nK2fsCZyvr9LCWc4l+htPT/jssxzat7fwxhtGBg50ZcyYPF55Jc9RNgKEqDC26tWhcV1y7m1V9IHs\nbLRnz6A7/Sva06fQnir4V3f6FLrjR69bj6VWIJZ7GmC55x7M9Rtg86sOubkoubkouTmQU/CvkpcH\nOTkF9117TMnNg9ycgufmXLs/728/5+ag5ORic3Uh89UJ5Pz7WVDkYE/heGTLuQTx8QW7uS9c0PDg\ng2bmzcshIED9t8rZvjmC9ORIbqovmw3N5T8KA7swvE+fQnvp4m3XYtPpwGDE5mLEZnQBoxGbiws2\ngxHtud/RpKWSG9qdjNkfYwsIKJ+eHIQz9gTO15dsOd+C5s2tbNmSyYsvuhIbq6NbNzfmzy/YqhZC\n3ABFwRpYG2tgbfI7di76WGYmurOn0Z76FSUtDZvLtXA1usC1sLUZjH/9bDRee/yvIEZX8p8vzZXL\neL74PMbNG9F3bkfGBx+R16NnxfYrRDmSLecy2Gzw8cd63n7biM0GX32VzYMPqhfQzvbNEaQnR+JQ\nfVmtuC74BPcpUSi5uWRHPo3pzakFp2n8jUP1dIOcsSdwvr5K23K+oVOppk2bxsCBA4mIiODQoUNF\nHtuzZw8DBgwgIiKC8ePHY7VayczMZOTIkQwZMoSIiAh27Nhxex2oSFFgxIh8Vq7MxmpVmDzZiNWq\ndlVCiDJpNGQPG05K7DbMjYNxXbIIn9AH5brWwiGUGc579+7l3LlzREdHM3XqVKZOnVrk8TfeeIM5\nc+awYsUKMjMz2bFjB9988w1169Zl6dKlfPjhh9ct44gefNBCv375HD6s5bvvZDRACEdhaRJMSuwP\nZD0/Et2Z03g/FIrb7JkyqYqwa2WG8+7duwkNDQUgKCiItLQ0TCZT4eMxMTHUrFkTAF9fX1JSUvDx\n8SE1NRWA9PR0fHx8KqL2Svef/+Si09l45x2jTFQihCNxcSHzrWmkrvoWq38A7tPewvtfD6E5f07t\nyoQoVpnhnJiYWCRcfX19SUhIKLztce2cxatXr7Jz5046depEr169+OOPPwgLC+OJJ57gP//5TwWU\nXvnq1rUxeHA+Z89qWLFCr3Y5QoiblN+xMynbdpHz8CPof96NT5cQWLas4OASIezITe+fLe74saSk\nJJ5//nmioqLw8fFhzZo1BAYGsnDhQk6cOMGECROIiYkpdb0+Pm7odOV7MnFpg+23ato0WLkS3n/f\nhRdecMHVtdxfokwV0ZfapCfH4fB9+XvC6q9h6VI0I0fCkCH4DxwI8+aBk+zlAyf4nErgrH39U5nh\nHBAQQGJiYuHtq1ev4u/vX3jbZDIxdOhQRo8eTYcOHQCIi4sr/LlRo0ZcvXoVi8WCtpSZPFJSyvfy\ndBV1VJ9eD88+a2DuXCMzZuQwYkTl7t92tqMVQXpyJE7VV89H0Gxtjt9Lz0N0NJYdP5Hx30/J79BR\n7cpum1N9Tn/jbH3d1tHaISEhxMbGAnD06FECAgIKd2UDTJ8+nSeffJKOHf/6hb7rrrs4ePAgAJcu\nXcLd3b3UYHY0L76Yh5eXjTlzjKSnq12NEOJWWe+6G378kcxxE9H83xWqPdoH98kTITdX7dJEFXdD\n5znPnDmT/fv3oygKUVFRHDt2DE9PTzp06EDr1q1p3rx54XN79+5N7969mTBhAklJSZjNZl566SXa\nt29f6mvY63nOJfnwQwNTpxoZMyaXceMq73q3zvbNEaQnR+KMff3Zky5uP57Dh6I7ewZz8H2kz1uA\npVFjtcu7Jc74OYHz9SUXvqgAmZnQpo07mZkKe/dmVtrUns72ywnSkyNxxr6K9GQy4RE1Adel/8Pm\n4oLpjbfIeeY5h5uf2xk/J3C+vm57EhJxPXd3GDMmj6wshQ8/NKhdjhCiPHh4YJo1h7TFX2Jzd8dz\nwmtUi+iH5v+uqF2ZqGIknG/DkCH53HmnlcWL9Zw/71jfrIUQJcvr2YvkbXvI6xqK4Yct+HRqh2Hd\n92qXJaoQCefbYDAUTEySl6fw3ntGtcsRQpQjW40apH35NRnvzETJyqLaU4Oo1q83Lgs+QXPxgtrl\nCScn4Xyb+vUz07ixha++0nHihLydzspigVWrdKSkqF2JqFSKQs4zw0jZvIO89iEYftqO54TX8GsR\njHfXDri99w7aw4dkEhNR7iRNbpNWCxMm5GK1Krzzjow9O6tPP9UzfLgrkye7qF2KUIGlQUPS1qwn\n6dBJMt79gLyuoehOHsf9vXfw7dYB31b34f76a+h3/IjM7SvKgxytXQ5sNujVy439+7WsX59Jy5YV\nd9kqZztaEey/p8uXFR54oODIfIPBRnx8Jv7+pf9vY+893Spn7OtWe1Iy0jFs3Yxh/VoMmzeiSU8D\nwOrtTV5oD3J79iKvSyj8bV6Icmezobn8B9oTx9AdP47u5HG0J46hv3Aec0BNLHXrYakXVPDvtf+s\ntQJB45jbZc72+1fa0dpyeaVyoCgwaVIuffu6MXWqka+/zna0My9EKaKijGRmKrRvb2b3bh2LF+t5\n5ZXKO7dd2Cebpxe5ffuR27cf5OWh370T4/rvMWxYh8uqaFxWRWMzGsl7sBN5PXuT270ntho1bvn1\nlMREdCeO/SOIjxd+KSisy2iEu+5Cc/4cuuNHr6/bxQXL3XWx3F2vSGhb6gVhDaxdsDtQqE62nMtR\nRIQrW7fqWLkyi86dK+ZydM72zRHsu6ft27U89pgbLVta+OqrLJo398BgsBEXl4mxlGMA7bmn2+GM\nfZV7TzYbukO/YFi/FuP6tYUBaVMUzC1akduzN3k9e2Gp36DYxZX0NLQnTvwVxCdPoDt+DE1iQpHn\n2bRaLEH3YG7UBEvDRgX/Nm6C5e66+NfyIeFqOkpCAtrfzqL97Qza38+iPXsG7W+/oT17Bk3G9dMb\n2gwGLHfd/bfQ/mur23pHHdCpuz3nbL9/MglJJTl8WEO3bu40a2YhNjarQraene2XE+y3p7w86NLF\njTNnNGzcmEXTplbefNPIRx8ZmDs3m4EDzSUua6893S5n7Kuie9L8/hvG2HUY1q9Fv2cXirVg2Msc\ndA95PXtjuac+2l9Poj15HN2J42gvXbxuHZa77sbcqDGWRk0wN2pcEMT31Kekb4hl9mSzoSQnoz17\n+lp4X/vvWoBrrl3yt8giOh1W/wBsPr5YfX2xeftg9fHF5uuL1dun4D6fgp9tvr4Fj3l7F1yQoJw4\n2++fhHMlGjbMhdWr9SxcmE2fPiX/8b5VzvbLCfbb05w5Bt5+28jTT+cxfXrBXMsXLyq0bu1OkyZW\nNm8u+QuYvfZ0u5yxr8rsSUlKwrBpA8YN6zBs24KSVfSCP5aatbBcC19z4yYFP9dveNPj1rfbk5KS\n/Fdgnz1T+LMm4SpKSkqxW90lsXp6FYS2jw82H5/rgt1yd92Csfkb2Cp3tt8/CedKdPasQkiIO3Xr\nWtm+Pavc9wI52y8n2GdPFy8qdOjgjpubjV27MvH2/uuxZ5914dtv9axencUDDxQ/fGGPPZUHZ+xL\ntZ6yszFs34bm/65gbtAIS6NG2LzL55KVFd5Tfj5KaiqalGSU5GQ0qSkoKclo/vw5ObngsdQUNMnJ\nBY+lpqBkZxe7OkudO8l+5jlynojE5lWtxJd1tt8/OSCsEtWrZ2PQoHyWLjWwcqWOQYPKf+tZVLxJ\nk4xkZSlMn55TJJgBhg3L49tv9Xz6qb7EcBaiTK6u5PXoqXYVt0avx+bvj+Vvlw++IdnZBaGdknLt\n32QMO37EJfoLPCa/jtt775AzeAjZzz6P9e66FVO7g3DM4+nt3Nixebi42Hj3XSM5OWpXI27W1q1a\n1q7V07atudhx5datrTRvbmHDBh2//y6H5Qtxw1xdsQbWxhJ8L/kdOpLX51+Y3v2ApF+OY5o4GZuX\nF27z5+HbrjleTw1Gv2dXlZ3gRcK5AgQG2nj66Xz++EPD//5XfgdDiIqXkwPjx7ug1dqYPj232DFl\nRSnYerbZFBYulIlnhLhdNh9fskeNIXn/YdLnLcDc9H6M677D++FwvLt3xrgquspN7iLhXEFGjcrF\n09PGhx8ayHCeIRKn9/HHBn77TcOzz+YTHFzyZDJ9+pipWdPK8uV6+XyFKC96PbmPDiA1dhsp38aS\n2+thdIcP4jV8KL4t74V33kFJSVa7ykoh4VxBfH1hxIg8kpI0zJsnW1eO4Nw5hdmzDQQEWHn11dxS\nn2swwNNP52MyKXz5pewdEaJcKQrmdu1J/3wZyXviyXpuOIrJBBMm4NesMR6vvYz29Cm1q6xQEs4V\naNiwPKpXtzJvnoHERBmbtHcTJ7qQk6Pw5pu5eHmV/fwhQ/JxcbHx2WcGLHJcmBAVwnp3XTKnTCf5\nl2MwaxbW6v64/m8hvg+0xGtwf/TbtznluLSEcwXy8IAxY/LIzFT48EPZerZnsbFaYmN1hISY6dfv\nxo6w9/Oz0b9/PufOadi4UU58EKIi2byqwZgxJP/8C2kLl5Dfui3GTbF4P/YwPl1CMH65DHJL3+Pl\nSCScK9iQIfnUqWPl88/1XLwoW8/2KDsbXn/dBZ2u5IPASjJ0aMFBKvPny65tZ5aVBaNGubBihXwJ\nU51OR16ff5G6dhMpGzhUA7IAAB8OSURBVLaS88ijaE8ex+ul4fg1b4Lbe++gJCSUvR47J+FcwYxG\nePXVXPLyFGbOlK1nezRnjoHz5zU891w+DRve3BXFGjWy0qmTmZ07dRw+LP87OauJE42sWKFn1ChX\nli2TL2L2wtyiFRmffk7y/sNkjRwN+fm4vDeDPU1fRtuuE9X698Vj9Ajc3p2Gy/Il6LdtRXvqV8jM\nVLv0MskMYZXAYoHOnd04dUrD9u1ZNGhw65eUtKe+youaPZ09q9Cpkzu+vjZ27sy8pav7bd6sZdAg\nNyIi8pkzp+DEdmf8nMA5+yqrp9WrdQwb5kqjRhauXlVISVH4739z6N/fficYcsbPCW6gL5OJGU+f\nZ9a2ttyrOcY6aw/qcP1c5VBwaU9r4B1YatfGGngH1tq1sQTWxlr7joJ/A2uXOHd5eZEZwlSm1cL4\n8Xk89ZQr06cbWPT/7d15XFTl/sDxzzmzMkQKiHsu10zBNNcsy7UwxZ+VWv7EtXst961rCXotU1Mr\nrauRVJhaZlamuKVpiy3eFDPzhynm2qJYIrggMDDb+f1xLoMjoOxzZnzer9e8mJkzA+fLc875nvOc\nZ1khRibRAkVRq7Pz8iTmzs0t87S7PXo4adLERWKinpkzJWrW1MT5rlABfv9dYupUMxaLwooVVqxW\nif79LUycaMZszq2U8fOFsks6VI3Xvr0bi0XhUE4EHWv/xtqFR7nTfAL5bCq61DMeP+Xff0OfcqjY\n3+eqEYazXn01YderR96jA3B06FglsYjkXEV693bQrp2TTz81cOCAjTZtyn71LFSMbdv0fPWVni5d\nHOU6yMoyPPWUjdhYM++9Z+DZZ8Vcz/7AbocxYwK4ckXi9det3H67Aih89FEOjz9uYfRoMyaTlZ49\nRVN9LcjMhPHjzUgSrF2bw759OmbPNhM1Lpx3321E5+giyklRkDIvI6emojt7Bjk1FfnsGXSpqWry\nTj2D/pcUpOQDAOj++J3M9z+uknhEtXYV+s9/dPTvb6FLFwfr1hU9APyNaDGu8vJGTNnZcP/9gaSl\nSXz7bfZ/D7xll5UFbdrcgsGgcOBANvXr+185wc21/c2dayQuzsSAAXbi43M9GgomJen43/8NwOmE\n1autlTZ/e1n5YznB9eMaO9bM+vUGpk7NIyZGPUHesEHPxIlmFAWWLMnlscfKcBKuKEgZGej+TMXZ\nsNF1J+YoretVa4sWLFXo/vuddO3q4Lvv9Hz3nc7bq3NTW7zYSGqqzPjxtnInZlC7zQ0daic9XWbD\nBlEh5eu++UZHXJyJRo1cLFyYW6gF/z33OFm1yookwYgRAezZI/Znb0pM1LN+vYF27Zz8858FNVf9\n+jlYu9aKxQLjxgXw+uvG0neJliSUGjVwtLyrQhPzjYjkXMX+9S+1H978+SZ/7DfvE44fl4mPN1K/\nvovJkyuuCnrkSBs6nUJCQhkOAIJmpKVJjB9vxmBQSEiwFtsWoWtXJytWWHE4YPDgAH78URxOveHM\nGYlp09R2AUuXWjFc05i+UycnW7bkUK+eixdfNBETY/KJQYPE1lTFWrd20bevnZ9+0rFtm7jCqmqK\nAtOnm7DbJV58MY/AwIr73fXrK/Tp4+DQIR3ffltxv1eoOi4XTJxo5vx5mZkz82jd+vptQyIjnbz1\nVi65uTBokIWDB8UhtSo5nTBhgpnMTIl58/L429+KPitu3tzFZ5/l0KKFk3ffNfL3v5vJyanilS0l\nsSV5QWysDVlWWLBADPtY1TZv1vPdd3oeeMBB794V39J21Cj1Snzx4gr/1UIViI838PXX6vYxenTJ\nZkHq29fBG2/kcuUKDBwYwJEj4rBaVeLjjezerScqys7gwdcvr9q1FTZvzqFLFwfbtxsYMMCi6WGV\nxVbkBU2buhg0yM6xYzo++UT7V882Gxw9KvPVVzoOH5Z9do7qrCx47jkTJpPC/PmF7yNWhPy5njdv\nhl9/1e6OLxT2008y8+ebqFnTxeuv5yKX4ug4YICDf/87lwsXZB57LICTJ0XZV7aDB2VeeslIrVou\nXn21ZCP7BQXBmjVWBg60s3+/jj59LJw6pc2yKlFmmD9/PsnJyUiSxIwZM2jVqpV7WVJSEq+99hqy\nLNO4cWPmzZvH+vXr2bx5s/szhw4d4sCBAxW/9j7smWdsrF9v4JVXTPTr56jsvu4lkpmp3o/1fOj4\n7TcJp7NgA5ZlhQYNFJo2ddG0qYs77nBy++0u7rjDRfXqXgzgBhYtMvHXXzJTp+bRuHHl3BTOn+t5\n7NgAli838uKL/jPWrz+7cgVGj1ZbX8fH5xIWVvrtY/BgB1ZrLtOnm+nf38KmTTk0aiQaH1SGnBy1\ndbbdrnZzCw0t+f/ZaIS4uFzq13fx2msm+vSxsHq1lXbttNW99YZdqX744QeWL1/O22+/zcmTJ5kx\nYwYff1zQz6tnz56sWrWK2rVrM2nSJAYMGEDXrl09vv/ZZ58xa9as667IzdCV6lrPPWfi7beN9Opl\nJyLCRWioQmioQkiI50+zueA75Y1LUeDPPyWOHZM5cUL2+JmWVvhSIThYoWlTJ02buqhfX+HPPyV3\n4k5PL/z5sDDXVUm74HndukqxZ7ZVUVa//CLTo4eFunUVdu3KJiCg8v6W3Q4dOgSRmamQnJxFUPG9\nJXyOL+xXpVWjRhADBtjZsMHAlCl5zJhRvkaCS5camD3bTIMGLjZtyqFevapP0P5YTlAQV0yMiZUr\njYwaZSvXCfCqVQamTTNhMsHbb1vp1atq7zOWa4SwPXv28OCDDwLQpEkTLl++TFZWFrf8twljYmKi\n+3lISAgXL170+P7SpUtZtGhRmVfen02ZYmPLFj3btxvYvr34z1ksBcm6Th245RZzkUk8/3lwsILT\nCb/+eu1VsJqIs7M9s6QkKdx2m0KPHg53Ms1/hIYWn1QvXIDjx3UcP+6Z5Pfs0bF7t+emFRioXmnn\nX2Hn//7GjSv/bFVRIDbWhMMhsWCBtVITM4DBABMmwIwZEmvWGEp871LwjpUrYcMGA+3bOytkAJnx\n4+3k5kq8/LKJAQPUK+hatcQVdEX58ksdK1caCQ93MnNm+Wqmhg+3U6eOi6eeCuCJJwJYsCCPv/9d\nG/vrDZNzeno6LVq0cL8OCQnh/Pnz7oSc/zMtLY3vv/+eyZMnuz978OBB6tSpQ1hYWEWvt18IDVXY\nvTubP/6QyciQuHBB8viZ/zz/9dGjMsnJANcfeF+S1ITqcnlmVZNJoUkTl0eSvP12F02auLBYSr/+\nISHQsaOTjh09zzZzcuDkyYJknX9icOSIzP/9n2d/UL1eoX176NnTQFSUo9jWluWxfr2e3bv19Opl\nJzKyas6MR42COXPUuZ6ffNKOTnSD1aRjx2QmToRbb1V4663C3XDK6p//tGG1wuuvm3jssQA2bLBS\no4ZI0OWVlqbODmY0KsTH53rUKpZVZKSTjRtzGDw4gJgYM6mpEjNm2ErV5qAylLo1UlG14BkZGYwZ\nM4ZZs2YRHBzsfn/dunX069evRL83ONiCXl+xR7DrVRloScOGJf9sTg6kp6uP8+cLnnu+lnC5oFkz\naN5cfYSHQ8OGEjqdDqj8TNGwIfTo4fmeejUPv/wCR46oj8OHJX74AZKSzMyZA61aQf/+0K8ftGxJ\nuRttXb4Mc+aA2QxvvmkgLKzqZhQaPlwiIUEiKSmIRx+tsj9b6Xxlv7qR3FwYN07dpz75RKJduzIO\nrl6MxYvV7XfJEh2DB9/Czp1w1eGx0vlLOeVTFHj4YUhPl3n1VejWreL6QUZGwt690KuXekKVkWFi\nxQr1/rS33PCec1xcHGFhYQwaNAiABx54gE2bNrmvmLOyshg+fDhTpkyhS5cuHt996KGH2LJlC8YS\nRHgz3nMuC3+MS5KCWLPGytatBr75RofNpmbkRo1c9OnjoE8fO23busp0JjtzpomEBCOxsXkeIwdV\ntrCwIP7zn2w6dw6kUycHGzeWbbhWrfGn7W/6dBPLlxsZPRrmzq2cmBQFnnnGxPvvG2nb1sknn+RU\nSRsEfyqnfOq49Wa6dFFH/aqMK9uMDIlhwwL48UcdnTs7WLnSyq23VvzfyVeu4Tvvu+8+duzYAcDh\nw4epWbOmOzEDvPTSS4wYMaJQYj537hyBgYElSszCza1GDYiOdrB6tZUjR7JISLDyyCN20tIkli41\nEhUVSOvWgcTGmti1S4ejhN2TDx2SeecdA40buxg/vuono2jWzEW3bg527xZzPWvNtm16li830ry5\nk3//u/L+jiTBwoV5DByoDjw0ZEiAL0wlrDknTkg8/7yJ4GC1pXVlVTmHhiqsX59DVJSdXbv09O1r\n4exZ73S1umGIbdu2pUWLFgwaNIgXX3yRWbNmkZiYyBdffIHVamXjxo2sW7eOYcOGMWzYMHdL7vPn\nzxMSElLpAQj+JSgIHn3UwbJlufzySxbvv59DdLSdvDyJFSuMDBhgoUWLW5g82cyOHbpi+1y7XGoj\nMJdLYsGCXK91VRs9Wj0pSEgQJ6lakZoqMWWKGbNZISEht9IbCMoyLF6cyyOP2ElK0jNiRIDPjhXg\nDTYbjB0bgNUqkZAAdepU7r37gABYvjyXkSNtHDmio3dvCykpVX9yLWal8jH+GFdJYnI4YM8eHVu3\n6tm2Tc9ff6k7S2CgwoMPOujTx8GDDzrc4yB/9JGeSZMC+J//sXtl/uz8mFwuuP9+C3/8IbN/f7bP\nt9qtiK58Fy/C77/L/PGHzO+/ywQHKwwYYC9To8TScjigf/8AkpL0LFyYy4gR9irbp+x2GDnSzPbt\nBiIj1SrTyqpY9KfjxLx5RpYsMTFokJ0PPzRUWVyKoo4YN3u2maAghXfftdK5c8U2KL1etbZIzj7G\nH+MqbUwulzqa07Ztej791MBvv6mJ2mRS6NrVyUMPOViwwEhOjsT332d7vZ/pypUGYmLMPPNMHtOm\n+fZczyUpK5tNnYzgt9/U5Ks+JPfzK1cKVxOGhroYOdLOP/5hozIr3F5+2cirr5ro29fOO++oo8RV\n5T6VlwfDhgXwzTd6+vSxs2xZLvpKGCTQX44TSUk6HnkkgAYNFL7+OpvGjas+rqunnUxIyKVPn4ob\n9lckZz/ij3GVJyZFgSNHZLZu1bN1q56UlIKW6DNn5jFpkneS4dUxZWdD69bqXM8//ZRdId0/vCUs\nLIi0tCukp0seCffqBHz2rISiFE7AFotCw4au/z7U5w0auDhwQMfy5UYuXZKwWBSGDrUzZoyN+vUr\n9tC0e7eO/v0DqFdPYefObKpVK4ipKvepnBwYMiSA77/X07+/naVLcyu8q50/HCcyM6F790BSUyW2\nbMmhQweX1+LavVvHiBEB3Huvg1WrKq4mTiRnP+KPcVVkTL/+KrFtm56MDInYWJvXukJcG9OcOUbe\neMPE669bGTSo4ifcqEx2O3z5pZ7ERD2nThk4eVIhJ6dw8pUkhbp1FY/ke3UyrlGj+AFtsrLggw8M\nvPmmkbNnZfR6hf79HUyYYKN58/IPVHPhgnqgT0uT2LxZPdDn88Y+lZUFAwda+PFHHb162end20FE\nhDr2QEXcA/eH48TYsWbWrzcwdWoeMTHqSbY348rMVAcYqsg2CiI5+xF/jOtmiOnMGYkOHQJp3tzF\nzp05lTLpRkU7flxmzRoDa9fqOX9evXUQFAQNGzqLTMD16yvlbnhns0Fiop6lS40cPapeTj70kJqk\nrx3spqQUBYYPD2DHDj0zZuQxZYpnbYq3tr/MTHj8cQsHDhRcNsuyOlBQeLiLiAgXERFOIiJc3HZb\n8Sc2RfH1fSoxUc+YMQG0a6fOxZxf9e/rcV1LJGc/4o9x3SwxPfWUmU2bDCQm5nD//dqcKzQrS51W\nc80aAz/8oB4Rg4MVHnvMTnS0nW7dAklPr/yycrng8891xMWZ2LdPTV533+1g4kQbkZHOUnWlWbbM\nwL/+ZaZzZ7V/7LVVyN7c/ux2SE6WSUnRkZIi//ehIzPTMxMHBSmEh6uJ+urEXVyfaV/ep86ckejW\nLRCHA3buzPYYNdCX4ypKucbWFgShYowaZWPTJgMJCQZNJWdFgX37ZD780MDGjQaysyUkSaFbNweD\nB9vp1cvhvk9eVVf8sgy9ejnp1SuHpCQdcXFGvvhCz7Bhepo3dzJ+vI3+/R03HG7z559lZs82UaOG\ni/j4ir+3W14GA7Rv76J9+4JqdkWBs2cld6LOT9r79+vcJ0z5GjQouLrOT9x/+5u2ZlcqDacTJkww\nk5kpsXixtVKG8/UV4srZx/hjXDdLTIoCvXtbOHBAZs+ebK8feNLSJNau1fPhhwaOH1ez1m23qXON\nDxpk57bbCq+fN8sqJUVm6VIjiYl6nE6JevVcjB1rY8gQO4FFjOSYlQWRkYGcPCnz4Yc5PPBA0SdE\nvrL95eaqtxoOH5Y5cqQgaeffcshnNiu0aCEREWGjVSsXd93lJDzcpYlpaW8kLs7I3LkmoqLsrFxZ\neM51XymrkhLV2n7EH+O6mWLKv5f21FM25s2r+rme1apCHR98YOCLL/Q4HBImk0JUlHqV3Lnz9auM\ntVBWp09LvPWWkQ8+MJCTIxEcrPCPf9h48km7x7y+kyaZ+egjA2PH2pg9u/j/tRZiKo+0NIkjR9RE\nnZ+0jx7VkXdVyHq9QvPmaqLOT9gRES5N9Rw4eFCmd28LISEK336bXWSXOl8vq2uJ5OxH/DGumykm\nux3atQvkyhWJ5OSsSh2392qnTqnTV378sYFz59Tse+edToYMsdO/v73EEzJoqawyMiSWLzewfLmR\nixclAgIUhgyxM3asjb17dYwbF0Dr1k4+/TTnuq32tRRTRaleXR3b/eBBmeRkHQcP6jh8WCY3t+BS\nVK9XaNbMRatWLlq1cnLXXU5atKiY1uKllZMDkZEWjh/XsXZtDt26+XYtR0mJ5OxH/DGumy2mJUuM\nzJtnYs6cXMaMqby5Y7OzYcsWtXFXUpJ6r7JaNXU0rsGD7bRqVfp7k1osq+xsWLPGQHy8kdRUGZ1O\nwWAAnQ6++urGtw+0GFN5FRWTwwFHj8r8/LOasJOT1YRttRYkbJ1O4Y47XNx1l3p13bKlkzvvLNuU\nsqURE2Ni5Uojo0fbmDvXf2s5riWSsx/xx7hutpguXFAHJalZU2Hv3uwKa6TkcMCpU2r15q5dOjZs\nMJCVpR54O3dWq62johzlujLSclnZ7epoTm+8YeToUZn4+FwGDLhxn3Itx1RWJY3J4VDvYycnyxw8\nWJCwr+7HLstqwm7VSh00pnp1hVtvVaheXaFaNfWEL/+9wMDSNxr88ksdgwdbCA93smNHznWr2v2t\nrERrbUHQkJAQePxxO++/b2T7dn2ZhgNMS5PcDYLy7zMeOyaTl1dwZKxb18WoUTaio+00bKiJc/BK\nZTDAwIEOHnvMQUaGRFiY/8dcXno9hIerrbzzB8dxOuHEiasTtszPP+v45Zcbn0Xq9fmJmmuSeP4D\nj9cmk9o2wGhUiI/P1dQ9cG8TyVkQvGDUKDU5JyQYrpucrVY4dkz26FZz5IhMenrhFrrNmxf0f23Z\n0kXHjk7NdR2qCrKMSMzloNOp0502a+Zi4MCChH3qlMy5cxKXLklcvixx+TJcvnz164L3L12SOHNG\nds/NfiNz5uTSooXvdgGrDCI5C4IX5M/1/M03eg4elLnzThenT0seCTglRebUKRmXy/MA17Chi/bt\n7e6+rRERTho3Vm7KRCxUDZ0OmjZ10bRp6b5nteJO2pcuSWRm4pHML12SqFFDYdSoymt74atEchYE\nLxk92sY33+gZPDiA7GyJ7GzPJFytmsLdd189wITaXzV/WkxB0LqAAAgIUKhdW9RklJZIzoLgJd27\nO2nd2smhQzJNmxYMy5g/TGPduqUbT1kQBP8hkrMgeIksw/btOTgceG32LEEQtKkUw8cLglDRZFkk\nZkEQChPJWRAEQRA0RiRnQRAEQdAYkZwFQRAEQWNEchYEQRAEjRHJWRAEQRA0RiRnQRAEQdAYkZwF\nQRAEQWNEchYEQRAEjRHJWRAEQRA0RiRnQRAEQdAYkZwFQRAEQWMkRVHEXF6CIAiCoCHiylkQBEEQ\nNEYkZ0EQBEHQGJGcBUEQBEFjRHIWBEEQBI0RyVkQBEEQNEYkZ0EQBEHQGL23V6C85s+fT3JyMpIk\nMWPGDFq1auVetnv3bl577TV0Oh1dunRh/PjxXlzT0nnllVfYv38/DoeD0aNH07NnT/eyHj16ULt2\nbXQ6HQCLFi2iVq1a3lrVEtm7dy+TJ0+madOmANxxxx0899xz7uW+WlaffPIJmzdvdr8+dOgQBw4c\ncL9u0aIFbdu2db9+99133eWmRceOHWPcuHE88cQTDB06lD///JNp06bhdDoJCwtj4cKFGI1Gj+9c\nbx/UgqJimj59Og6HA71ez8KFCwkLC3N//kbbqhZcG1NsbCyHDx+mevXqAIwcOZJu3bp5fEfr5QSF\n45o0aRIXL14E4NKlS7Ru3Zq5c+e6P5+YmMiSJUto0KABAJ06dWLs2LFeWfcKp/iwvXv3KqNGjVIU\nRVFOnDihDBw40GN57969lbNnzypOp1OJjo5Wjh8/7o3VLLU9e/YoTz75pKIoinLhwgWla9euHsu7\nd++uZGVleWHNyi4pKUmZOHFisct9tayutnfvXuWFF17weO/uu+/20tqUXnZ2tjJ06FBl5syZyvvv\nv68oiqLExsYq27ZtUxRFUV599VXlgw8+8PjOjfZBbysqpmnTpilbt25VFEVRVq9erbz88sse37nR\ntuptRcUUExOj7Ny5s9jvaL2cFKXouK4WGxurJCcne7y3fv165aWXXqqqVaxSPl2tvWfPHh588EEA\nmjRpwuXLl8nKygLg9OnTVKtWjTp16iDLMl27dmXPnj3eXN0S69ChA0uWLAHg1ltvxWq14nQ6vbxW\nlceXy+pqS5cuZdy4cd5ejTIzGo0sW7aMmjVrut/bu3cvDzzwAADdu3cvVC7X2we1oKiYZs2axUMP\nPQRAcHAwly5d8tbqlUlRMd2I1ssJrh/XqVOnuHLliiav9iuLTyfn9PR0goOD3a9DQkI4f/48AOfP\nnyckJKTIZVqn0+mwWCwArFu3ji5duhSqCp01axbR0dEsWrQIxUcGeTtx4gRjxowhOjqa77//3v2+\nL5dVvoMHD1KnTh2P6lEAm83G1KlTGTRoECtXrvTS2pWMXq/HbDZ7vGe1Wt3V2KGhoYXK5Xr7oBYU\nFZPFYkGn0+F0OlmzZg19+/Yt9L3itlUtKComgNWrVzN8+HCefvppLly44LFM6+UExccFsGrVKoYO\nHVrksh9++IGRI0cyYsQIUlJSKnMVq5TP33O+mq8kqZL68ssvWbduHStWrPB4f9KkSXTu3Jlq1aox\nfvx4duzYQa9evby0liXTqFEjJkyYQO/evTl9+jTDhw/n888/L3T/0letW7eOfv36FXp/2rRpPPzw\nw0iSxNChQ2nfvj0tW7b0whqWX0n2L1/ZB51OJ9OmTeOee+7h3nvv9Vjmi9vqI488QvXq1QkPDych\nIYE33niD559/vtjP+0o5gXqCu3//fl544YVCy+666y5CQkLo1q0bBw4cICYmhi1btlT9SlYCn75y\nrlmzJunp6e7XaWlp7iuXa5edO3euVNVA3rZr1y7eeustli1bRlBQkMeyRx99lNDQUPR6PV26dOHY\nsWNeWsuSq1WrFlFRUUiSRIMGDahRowbnzp0DfL+sQK3+bdOmTaH3o6OjCQwMxGKxcM899/hEWV3N\nYrGQm5sLFF0u19sHtWz69Ok0bNiQCRMmFFp2vW1Vq+69917Cw8MBtcHotduZr5YTwL59+4qtzm7S\npIm74VubNm24cOGC39wC9OnkfN9997Fjxw4ADh8+TM2aNbnlllsAqF+/PllZWZw5cwaHw8HXX3/N\nfffd583VLbErV67wyiuv8Pbbb7tbX169bOTIkdhsNkDdcPNblWrZ5s2bWb58OaBWY2dkZLhbmPty\nWYGatAIDAwtdWZ06dYqpU6eiKAoOh4OffvrJJ8rqap06dXLvY59//jmdO3f2WH69fVCrNm/ejMFg\nYNKkScUuL25b1aqJEydy+vRpQD1RvHY788Vyyvfzzz/TvHnzIpctW7aMTz/9FFBbeoeEhGi6N0Rp\n+PysVIsWLeLHH39EkiRmzZpFSkoKQUFBREZGsm/fPhYtWgRAz549GTlypJfXtmQ+/vhj4uLiaNy4\nsfu9jh070qxZMyIjI3nvvffYuHEjJpOJiIgInnvuOSRJ8uIa31hWVhbPPPMMmZmZ2O12JkyYQEZG\nhs+XFajdpxYvXsw777wDQEJCAh06dKBNmzYsXLiQpKQkZFmmR48emu7mcejQIV5++WVSU1PR6/XU\nqlWLRYsWERsbS15eHnXr1mXBggUYDAaefvppFixYgNlsLrQPFncg9YaiYsrIyMBkMrmTU5MmTXjh\nhRfcMTkcjkLbateuXb0cSYGiYho6dCgJCQkEBARgsVhYsGABoaGhPlNOUHRccXFxxMXF0a5dO6Ki\notyfHTt2LG+++SZ//fUXzz77rPsEWKtdxMrC55OzIAiCIPgbn67WFgRBEAR/JJKzIAiCIGiMSM6C\nIAiCoDEiOQuCIAiCxojkLAiCIAgaI5KzIAiCIGiMSM6CIAiCoDEiOQuCIAiCxvw/24BeZADdQ3AA\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f68729125c0>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_loss = model_history.history['loss']\n",
    "val_loss = model_history.history['val_loss']\n",
    "\n",
    "epochs = range(len(train_loss))\n",
    "\n",
    "plt.plot(epochs, train_loss, 'r', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "rV50DLDybfnM"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "include_colab_link": true,
   "name": "timeforecasting-using-rnns",
   "provenance": [],
   "version": "0.3.2"
  },
  "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}