auto_encoder.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T09:03:29.196238Z",
     "start_time": "2017-11-17T09:03:28.644004Z"
    },
    "_cell_guid": "679e0d3e-646d-4e96-9eb0-b362d8c6e51f",
    "_uuid": "0d05e5ce89af3e25d1c1fb244d021a1cfa1a058c"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from __future__ import division\n",
    "from utils import *\n",
    "import array \n",
    "\n",
    "from pydub import AudioSegment\n",
    "import tensorflow as tf\n",
    "\n",
    "from keras.models import Model, Sequential\n",
    "from keras.layers import Input, Conv2D, Flatten, MaxPooling2D, Activation, BatchNormalization, GlobalAveragePooling2D, GlobalMaxPool2D, concatenate, Dense, Dropout\n",
    "from keras.optimizers import Adam\n",
    "from tensorflow.python.keras.utils import to_categorical\n",
    "\n",
    "\n",
    "\n",
    "from tqdm import tqdm_notebook\n",
    "from keras_tqdm import TQDMNotebookCallback\n",
    "from keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from keras.callbacks import TensorBoard\n",
    "\n",
    "\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T09:03:29.210749Z",
     "start_time": "2017-11-17T09:03:29.19832Z"
    },
    "_cell_guid": "8ab00801-08b9-44d3-a063-32e82dbf8f58",
    "_uuid": "53c19941676690454dd4b91109976b6c59cb7a40"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "12"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "POSSIBLE_LABELS = 'yes no up down left right on off stop go silence unknown'.split()\n",
    "id2name = {i: name for i, name in enumerate(POSSIBLE_LABELS)}\n",
    "name2id = {name: i for i, name in id2name.items()}\n",
    "len(id2name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### load raw auidos into memory for faster processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 18 s, sys: 12.6 s, total: 30.6 s\n",
      "Wall time: 30.6 s\n"
     ]
    }
   ],
   "source": [
    "%%time \n",
    "\n",
    "train_df = pickle.load( open(\"cache/train_df_waug.pik\",\"rb\"))\n",
    "valid_df = pickle.load( open(\"cache/valid_df.pik\",\"rb\"))\n",
    "silent_df = pickle.load(open(\"cache/silent_df.pik\",\"rb\"))\n",
    "unknown_df = pickle.load(open(\"cache/unknown_df_waug.pik\",\"rb\"))\n",
    "test_df =  pickle.load(open(\"cache/test_df.pik\",\"rb\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### generators\n",
    "train gen will also sample from test files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 368,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3382.203125"
      ]
     },
     "execution_count": 368,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(train_df.shape[0] + unknown_df.shape[0] + test_df.shape[0])/64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# there are a couple of records with all 'pixel' values set to 1.0, they will cause 'inf' loss durin training, so removing them\n",
    "means = test_df.raw.apply(lambda x: x[0].mean() ) \n",
    "test_df = test_df[  means < 0.0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T09:03:32.519795Z",
     "start_time": "2017-11-17T09:03:32.483881Z"
    },
    "_cell_guid": "144c6e60-8a83-437d-8b8a-ea065af90923",
    "_uuid": "22e0e6c718171167089fb6df36d3dc43a1029992"
   },
   "outputs": [],
   "source": [
    "def train_generator(train_batch_size):\n",
    "    while True:\n",
    "        \n",
    "        this_train = train_df.groupby('label_id').apply(lambda x: x.sample(n = 2000))\n",
    "        extra_data_size = int(this_train.shape[0]* 0.1)\n",
    "        this_train = pd.concat([test_df , #.sample(this_train.shape[0]),\n",
    "                                silent_df.sample(extra_data_size),\n",
    "                                this_train,\n",
    "                                unknown_df.sample(n = unknown_df.shape[0]//5)],axis=0 ) #.sample(this_train.shape[0])],axis=0)\n",
    "        \n",
    "        this_train.reset_index(drop=True,inplace=True)\n",
    "#         print this_train.columns\n",
    "        \n",
    "        shuffled_ids = random.sample(range(this_train.shape[0]), this_train.shape[0])\n",
    "        \n",
    "        for start in range(0, len(shuffled_ids), train_batch_size):\n",
    "            x_batch = []\n",
    "            end = min(start + train_batch_size, len(shuffled_ids))\n",
    "            i_train_batch = shuffled_ids[start:end]\n",
    "            for i in i_train_batch:\n",
    "                x_batch.append(this_train.loc[i,'raw'].T)\n",
    "                \n",
    "            x_batch = 1.- np.array(x_batch)/-80.\n",
    "#             y_batch = to_categorical(y_batch, num_classes = len(POSSIBLE_LABELS))\n",
    "            yield x_batch, x_batch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(105525, 6)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T09:03:32.624289Z",
     "start_time": "2017-11-17T09:03:32.521828Z"
    },
    "_cell_guid": "59a13393-9bc3-4b27-abe2-9c78b3c32ead",
    "_uuid": "6f9a8fbf6e352b1c77b9c22dddf1c5d69382bd5b"
   },
   "outputs": [],
   "source": [
    "def valid_generator(val_batch_size):\n",
    "    while True:\n",
    "        ids = list(range(valid_df.shape[0]))\n",
    "        for start in range(0, len(ids), val_batch_size):\n",
    "            x_batch = []\n",
    "            end = min(start + val_batch_size, len(ids))\n",
    "            i_val_batch = ids[start:end]\n",
    "            for i in i_val_batch:\n",
    "                x_batch.append(valid_df.loc[i,'raw'].T)\n",
    "\n",
    "            x_batch = 1.- np.array(x_batch)/-80.\n",
    "            yield x_batch, x_batch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.31318883826536237"
      ]
     },
     "execution_count": 192,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "next(valid_generator(64))[0].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#a common practice is to choose a filter size in time which spans 2/3 o"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.layers import Input, GRU, RepeatVector, BatchNormalization, TimeDistributed, Conv1D\n",
    "from keras import backend as K\n",
    "from keras.layers import  Conv2D, MaxPooling2D, UpSampling2D, Lambda, Reshape\n",
    "\n",
    "\n",
    "timesteps, input_dim , latent_dim = 32,128, 128\n",
    "\n",
    "# inputs = Input(shape=(timesteps, input_dim))\n",
    "# conv2d = Conv1D(64,1,activation='relu')(inputs)\n",
    "# encoded = GRU(latent_dim,return_sequences=True)(conv2d)\n",
    "# encoded = GRU(latent_dim,return_sequences=True)(encoded)\n",
    "# encoded = TimeDistributed(Dense(latent_dim, activation='sigmoid'))(encoded)\n",
    "\n",
    "# # decoded = RepeatVector(timesteps)(encoded)\n",
    "# decoded = GRU(latent_dim, return_sequences=True)(encoded)\n",
    "# decoded = GRU(latent_dim, return_sequences=True)(decoded)\n",
    "# decoded = TimeDistributed(Dense(input_dim, activation='sigmoid'))(decoded)\n",
    "\n",
    "# model = Model(inputs, decoded)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_img = Input(shape=(timesteps, input_dim))  # adapt this if using `channels_first` image data format\n",
    "\n",
    "x = Reshape((timesteps, input_dim,1))(input_img)\n",
    "\n",
    "\n",
    "x = Conv2D(64, (16, 10), activation='relu', padding='same')(x)\n",
    "x = MaxPooling2D((2, 2), padding='same')(x)\n",
    "x = Conv2D(32, (8, 5), activation='relu', padding='same')(x)\n",
    "x = MaxPooling2D((2, 2), padding='same')(x)\n",
    "x = Conv2D(32, (4, 3), activation='relu', padding='same')(x)\n",
    "encoded = MaxPooling2D((2, 2), padding='same')(x)\n",
    "\n",
    "# at this point the representation is (4, 4, 8) i.e. 128-dimensional\n",
    "\n",
    "x = Conv2D(32, (4, 3), activation='relu', padding='same')(encoded)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "x = Conv2D(32, (8, 5), activation='relu', padding='same')(x)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "x = Conv2D(64, (16, 10), activation='relu', padding='same')(x)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "\n",
    "decoded = Conv2D(1, (3, 3), activation='sigmoid', padding='same')(x)\n",
    "decoded  = Reshape((timesteps, input_dim))(decoded)\n",
    "\n",
    "model = Model(input_img, decoded)\n",
    "# autoencoder.compile(optimizer='adadelta', loss='binary_crossentropy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_3 (InputLayer)         (None, 32, 128)           0         \n",
      "_________________________________________________________________\n",
      "reshape_5 (Reshape)          (None, 32, 128, 1)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_19 (Conv2D)           (None, 32, 128, 64)       10304     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_9 (MaxPooling2 (None, 16, 64, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_20 (Conv2D)           (None, 16, 64, 32)        81952     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_10 (MaxPooling (None, 8, 32, 32)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_21 (Conv2D)           (None, 8, 32, 32)         12320     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_11 (MaxPooling (None, 4, 16, 32)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_22 (Conv2D)           (None, 4, 16, 32)         12320     \n",
      "_________________________________________________________________\n",
      "up_sampling2d_9 (UpSampling2 (None, 8, 32, 32)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_23 (Conv2D)           (None, 8, 32, 32)         40992     \n",
      "_________________________________________________________________\n",
      "up_sampling2d_10 (UpSampling (None, 16, 64, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_24 (Conv2D)           (None, 16, 64, 64)        327744    \n",
      "_________________________________________________________________\n",
      "up_sampling2d_11 (UpSampling (None, 32, 128, 64)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_25 (Conv2D)           (None, 32, 128, 1)        577       \n",
      "_________________________________________________________________\n",
      "reshape_6 (Reshape)          (None, 32, 128)           0         \n",
      "=================================================================\n",
      "Total params: 486,209\n",
      "Trainable params: 486,209\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#original\n",
    "# x = Conv2D(64, (10, 10), activation='relu', padding='same')(x)\n",
    "# x = MaxPooling2D((2, 2), padding='same')(x)\n",
    "# x = Conv2D(32, (5, 5), activation='relu', padding='same')(x)\n",
    "# x = MaxPooling2D((2, 2), padding='same')(x)\n",
    "# x = Conv2D(32, (3, 3), activation='relu', padding='same')(x)\n",
    "# encoded = MaxPooling2D((2, 2), padding='same')(x)\n",
    "\n",
    "# # at this point the representation is (4, 4, 8) i.e. 128-dimensional\n",
    "\n",
    "# x = Conv2D(32, (3, 3), activation='relu', padding='same')(encoded)\n",
    "# x = UpSampling2D((2, 2))(x)\n",
    "# x = Conv2D(32, (5, 5), activation='relu', padding='same')(x)\n",
    "# x = UpSampling2D((2, 2))(x)\n",
    "# x = Conv2D(64, (10, 10), activation='relu', padding='same')(x)\n",
    "# x = UpSampling2D((2, 2))(x)\n",
    "# decoded = Conv2D(1, (3, 3), activation='sigmoid', padding='same')(x)\n",
    "# decoded  = Reshape((timesteps, input_dim))(decoded)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def root_mean_squared_error(y_true, y_pred):\n",
    "        return K.sqrt(K.mean(K.square(y_pred - y_true), axis=-1)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(optimizer=\"adam\", loss=root_mean_squared_error, metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model.load_weights('weights/starter_{}.hdf5'.format(exp_name))\n",
    "# model.load_weights('weights/starter_ae_lstm_large_withdense_128_with_test.hdf5')\n",
    "# model.load_weights('weights/starter_ae_wtest_conv_rmse_c5.hdf5'.format(exp_name))\n",
    "# "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Epoch 8/100\n",
    "329/329 [==============================] - 30s - loss: 0.5700 - acc: 0.1437 - val_loss: 0.5567 - val_acc: 0.1522\n",
    "\n",
    "RMSE\n",
    "latent_dim = 128\n",
    "Epoch 16/100\n",
    "329/329 [==============================] - 48s - loss: 0.0966 - acc: 0.2346 - val_loss: 0.0859 - val_acc: 0.2259\n",
    "\n",
    "latent_dim = 256\n",
    "Epoch 10/100\n",
    "329/329 [==============================] - 47s - loss: 0.0946 - acc: 0.2367 - val_loss: 0.0844 - val_acc: 0.2312\n",
    "\n",
    "latdim: 256, mse\n",
    "Epoch 11/100\n",
    "330/329 [==============================] - 50s - loss: 0.0113 - acc: 0.1773 - val_loss: 0.0093 - val_acc: 0.2245\n",
    "\n",
    "latdim: 256, bxent\n",
    "330/329 [==============================] - 51s - loss: 0.4866 - acc: 0.1917 - val_loss: 0.5078 - val_acc: 0.1525\n",
    "Epoch 00011: early stopping\n",
    "\n",
    "\n",
    "Epoch 00012: reducing learning rate to 0.000999999977648. adadelta bix, conv tiramisu\n",
    "990/989 [==============================] - 93s - loss: 0.4627 - acc: 0.1618 - val_loss: 0.4635 - val_acc: 0.1526\n",
    "\n",
    "\n",
    "ae_wtest_conv_rmse_c3, adam, root_mean_squared_error\n",
    "Epoch 7/100\n",
    "3300/3300 [==============================] - 388s - loss: 0.1300 - acc: 0.4638 - val_loss: 0.1303 - val_acc: 0.4960\n",
    "\n",
    "ae_wtest_conv_rmse_c6\n",
    "Epoch 23/100\n",
    "3300/3300 [==============================] - 386s - loss: 0.0260 - acc: 0.5283 - val_loss: 0.0263 - val_acc: 0.5537"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "exp_name = \"ae_wtest_conv_rmse_larger_convs\"\n",
    "callbacks = [EarlyStopping(monitor='val_acc',\n",
    "                           patience=5,\n",
    "                           verbose=1),\n",
    "             \n",
    "             ReduceLROnPlateau(monitor='val_acc',\n",
    "                               factor=0.1,\n",
    "                               patience=3,\n",
    "                               verbose=1,\n",
    "                               epsilon=0.01,\n",
    "                               min_lr=1e-5),\n",
    "             \n",
    "             ModelCheckpoint(monitor='val_acc',\n",
    "                             filepath='weights/starter_{}.hdf5'.format(exp_name),\n",
    "                             save_best_only=True,\n",
    "                             save_weights_only=True)\n",
    "             \n",
    "#             ,TensorBoard(log_dir='./logs/logs_{}'.format(exp_name), histogram_freq=0, batch_size=64, write_graph=True)\n",
    "            ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T09:42:31.48233Z",
     "start_time": "2017-11-17T09:03:33.355603Z"
    },
    "_cell_guid": "5f3d1b09-500f-410e-820a-8eaab24b6ebb",
    "_uuid": "528ec66a0a6caca952273ab916e609625839b19e",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      " 662/3300 [=====>........................] - ETA: 442s - loss: 0.0696 - acc: 0.1164"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-29-e95113f569db>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      5\u001b[0m                               \u001b[0mshuffle\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mTrue\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m                               \u001b[0mvalidation_data\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mvalid_generator\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m64\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m                               validation_steps=int(np.ceil(valid_df.shape[0]/64)))\n\u001b[0m",
      "\u001b[0;32m/home/ubuntu/anaconda2/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda2/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;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[1;32m   2040\u001b[0m                     outs = self.train_on_batch(x, y,\n\u001b[1;32m   2041\u001b[0m                                                \u001b[0msample_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2042\u001b[0;31m                                                class_weight=class_weight)\n\u001b[0m\u001b[1;32m   2043\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2044\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda2/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mtrain_on_batch\u001b[0;34m(self, x, y, sample_weight, class_weight)\u001b[0m\n\u001b[1;32m   1760\u001b[0m             \u001b[0mins\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1761\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_train_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1762\u001b[0;31m         \u001b[0moutputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1763\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1764\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0moutputs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda2/lib/python2.7/site-packages/keras/backend/tensorflow_backend.pyc\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2271\u001b[0m         updated = session.run(self.outputs + [self.updates_op],\n\u001b[1;32m   2272\u001b[0m                               \u001b[0mfeed_dict\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2273\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2274\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2275\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    887\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    888\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 889\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    890\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    891\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1118\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1119\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1120\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1121\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1122\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1315\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1316\u001b[0m       return self._do_call(_run_fn, self._session, feeds, fetches, targets,\n\u001b[0;32m-> 1317\u001b[0;31m                            options, run_metadata)\n\u001b[0m\u001b[1;32m   1318\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1319\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1321\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1322\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1323\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1324\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1325\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(session, feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1300\u001b[0m           return tf_session.TF_Run(session, options,\n\u001b[1;32m   1301\u001b[0m                                    \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1302\u001b[0;31m                                    status, run_metadata)\n\u001b[0m\u001b[1;32m   1303\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1304\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "history = model.fit_generator(generator=train_generator(64),\n",
    "                              steps_per_epoch=3300, #train_df.shape[0]*3/64,\n",
    "                              epochs=100,\n",
    "                              callbacks=callbacks,\n",
    "                              shuffle=True,\n",
    "                              validation_data=valid_generator(64),\n",
    "                              validation_steps=int(np.ceil(valid_df.shape[0]/64)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.load_weights('weights/starter_{}.hdf5'.format(exp_name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {},
   "outputs": [],
   "source": [
    "t = next(valid_generator(64))[0][18].T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### actual"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAxIAAAEYCAYAAADIwMCEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsvXmcZHdd7/35nlNb78tsmTWTQBISIBAIiYhAAiIvES+G\nK4iikitevN6rggsGvfdRnusG6uN+9XFUTLgi4CuCeuERiMAQwxLJMpJlspDJJDPTs/S+1HrqnO/z\nR50mTdPd093zqa7q6s87r3mlu6vqU79T59Q557ubu0MIIYQQQggh1kLQ6gUIIYQQQgghNh8yJIQQ\nQgghhBBrRoaEEEIIIYQQYs3IkBBCCCGEEEKsGRkSQgghhBBCiDUjQ0IIIYQQQgixZmRICCG2FGZ2\ng5mdbPU6hBBCiM2ODAkhxKbBzI6bWc3Mti/6+/1m5mZ2sDUr61zM7GD62WZavRYhhBDthQwJIcRm\n40kAPzj/i5k9H0B365bTPDbLzTt7nZtlu4UQYqsjQ0IIsdn43wB+dMHvbwPwwYVPMLO8mf2umT1t\nZmfN7P81s67ViKfe958xs2NmNmZmv2NmQfpYYGb/w8yeMrNzZvZBMxtIH7vNzH4+/XlvqvPf0t+f\nZWYTC3Reb2ZHzGzKzL5kZlcveP/jZnaLmX0NQHHxTbU1+P30/WfM7AEze1762K3ptt5hZrNm9gUz\nu3jBa5+TPjZhZo+a2ZsXPNZlZv9Pum3TZnZX+pndmT5lyszmzOylZnazmX0xXcc4gPeu9Nmk+j+a\nPjZuZv9Xup3fmT72XjO73cz+xsxmANxsZteZ2ZfTz+i0mf2JmeUW7af/amaPp9v6a+nn/KX0c/m7\nhc8XQgjBR4aEEGKz8RUA/WZ2pZmFAN4C4G8WPed9AC4H8EIAzwawF8CvrOE9bgJwLYAXAXgDgB9L\n/35z+u9GAJcC6AXwJ+ljXwBwQ/rzKwEcA/CKBb//q7snZnYNgA8A+AkA2wD8OYB/MrP8gvf/QQDf\nA2DQ3euL1vZdqe7lAAYAvBnA+ILH3wrg1wBsB3AEwIcAwMx6ANwB4G8B7ETjc/tTM7sqfd3vAngx\ngG8HMAzgFwEkC7Zh0N173f3L6e/Xp9u4C8BvrPTZpO/xp+nadqfr3rtou94A4HYAg+maYwA/m27H\nSwG8GsB/XfSa16Zr/rZ0vYcA/DCA/QCehwWRKyGEEHxkSAghNiPzUYnXADgK4NT8A2ZmAN4B4Gfd\nfcLdZwH8Jho3zqvl/elrnwbwB3jmhvStAH7P3Y+5+xyAXwLwljRq8AUA35FGHV4B4LcBvCx93SvT\nx5Gu7c/d/W53j939NgBVNG6G5/kjdz/h7uUl1hYB6APwHADm7kfd/fSCxz/p7ne6exXAfwfwUjPb\nD+D1AI67+1+7e93d7wfw9wDelK75xwC8091Ppev6UqqxHCPu/sepVvk8n833A/g/7n6Xu9fQMOp8\nkd6X3f0f3D1x97K73+vuX0n1j6NhcL1y0Wt+291n3P0hAA8C+Ez6/tMA/hnANSusXwghxAUiQ0II\nsRn53wB+CA0P+AcXPbYDjZqJe9O0mCkAn0r/vlpOLPj5KQB70p/3pL8vfCwDYJe7PwGgiEYU5OUA\nPgFgxMyuwDcbEhcD+Pn5taXr27/gPRa//zfh7p9Dw9P/vwCcM7NDZta/1GvTG/qJVPtiANcvet+3\nArgIDa9/AcATK38s38TiNS772aSPLVxXCd8cRfkWPTO73Mw+YWZn0nSn30zXuZCzC34uL/F77+o2\nRQghxHqQISGE2HS4+1NoFF2/DsDHFj08hsZN5HPdfTD9N+Dua7mp3L/g5wMARtKfR9C4IV/4WB3P\n3MB+AQ3ve87dT6W/vw3AEBppRkDjhvk3Fqxt0N273f3DCzdxpcW5+x+5+4sBXIVGitO7l1q7mfWi\nkaY0kr7vFxa9b6+7/yQan1kFwLOWervllrHo95U+m9MA9i1YVxcaaV0r6f0ZgEcAXObu/QB+GYAt\nsxYhhBAtQIaEEGKz8nYAr3L34sI/unsC4C8A/L6Z7QS+Ufz82jVov9vMhtKUoHcC+Gj69w8D+Fkz\nuyS9Sf9NAB9dUMfwBQA/hWcKlA+nv9/l7nH6t78A8F/M7Pq0cLrHzL7HzPpWszAze0n62iwaEZAK\nGrUM87zOzL4jLTT+NQBfcfcTaERILjezHzGzbPrvJWZ2ZfqZfQDA75nZHjML06LqPIDRVP/S8yxt\npc/mdgDfa2bfnq7rvTi/UdAHYAbAnJk9B8BPrubzEUIIsXHIkBBCbErc/Ql3v2eZh28B8HUAX0nT\nYv4FwBVrkP9HAPeiEUX4JIC/Sv/+ATTSqu5EIyJSAfDTC173BTRugOcNibvQSLOa/x3pmv8zGulJ\nk+k6b17D2vrRMEYm0UgfGgfwOwse/1sAv4pGStOL0Sg+Rlor8l1o1IqMADgD4P0A5ou8fwHAAwC+\nmr72/QCCNA3pNwB8MU2JWljLsZBlP5u0huGnAXwEjejEHIBzaNSGLMcvoJG+Nptu70dXeK4QQogW\nYO4rRtCFEGJLYWaORjrN11u9lrViZrcCOOnu/6PVa1mJNGIxhcbn/GSr1yOEEGJ9KCIhhBCi6ZjZ\n95pZd9qG9nfRiH4cb+2qhBBCXAgyJIQQQmwEb0AjpWoEwGUA3uIKiQshBA0ze5OZPWRmiZldu8rX\n3Gpm35/+fNgaw0qPmNlRM3vH+V6fOd8ThBBiK+Hum7YzkLvf3Oo1LIe7/ziAH2/1OoQQYrNjZjcA\nuHmJc/6DAN6Ixtyd9fJWd7/HzIYBPGFmt6bzf5ZEhoQQQgghhBCbHHc/CgCNuaxLkw5t/WM0Brqe\nALCckdCLRmfAeJnHAXSwIZEWTAohxBaBHUhh6yXnf4oQQqydMXdfy8DRDeW1r73Ox8enV/Xce+99\n7CE0Ot7Nc8jdD5GXdBMaXQyvQmNg6MNodN2b50NmVkUjBfVdC1qXL0nHGhINOnzzhBAixYxb8tYY\nU8EjSVbq9CqEEOul/lSrV7AS4+PTuPvfVpdplAlvrLj7srUNZnY3Gi27ewEMm9n8oNNb3P3Tq1zS\nKwB8ODUQRszsc4sen09t2gHgS2b2qXQI7NJrXuWbCiHEpiMg3wx3FfZS9eJk2bTTNRPVV+fxWi1x\nXDz/k4QQQqyMA0g4EVl3vx5YsUaChruPmtl9AK5HY2bRkqhrkxBCCCGEEE3BgXp9df82hjsB/ICZ\nhWa2G8CNSz3JzLoBXAPgiZXEOjgiYTDjbF4Q5M//pDUQBl1UPTa1aKzVSxBNIgz7qXrP67+Jqteb\n9FH1xsIzVD02gfN8OXMYp2kBwLniQ1Q9dsQkScpUPSGEaAoOYIM6XZvZTWgUUu8A8EkzO+Lur130\ntI8DeBUatRFPA/jyosc/ZGZlNFKobnX3e1d6zw42JIQQQgghhGglTktt+oai+2EAh5f4+8fRMBRW\neq0D+KllHrthrWvpWEMiDLow2HMVRSsmFwmyiyLduQeoIhKdy8t7f4yq95JBbgQhF3A7BT09N8DV\nq8xS9bZlumla+eDZNC0AmOi+hqo3Dm5E4uvRXVS9qeJRqh4bVoR9HvcNS6MQQpANiXaiYw2JrBWw\nK/McilZkXENiINlO1QvIpS4P5Lk3S+XqSareVoKdinRxVw9Vb1uOKocE3PDvYI773Yi9l6p3UVdI\n09peoEkBAGYjnpEDACMlbopoZNdR9WYrT1P19vZ/O1WvlsxR9c5ML85mEEI0BWKxdTvSsYaEEEII\nIYQQrYWf2tROdKwhESKDgWSYohVZRNGZJ+9cz1yWvBu7ckNUPUUk1k8cz3D1yOeyKlmPnNlE50Av\nL4IAANuJp4JCyI3msFv6lfNcxd4yN60uDLjROnakONFAPyE2J+6wuHNTCTvWkKgjwlh4mqPl3NSm\ngnHTI4x8wSqWR6l6on2Yibgns5mIm9vUFXItiRz55pp54w8AfRne+mbq3M8uIjcZGcxx17c95N74\ns2vXYnC/axkjH3xCiI1DEQkhhBBCCCHEmnAAyca0f20FHWtIOBJaJIHt8U+Ma5mGxF70AJDLcFMG\nqrX27uW/lXgSI1S9/fFBqt5Qjn2y5XrBi/X2XV8h4K6tEnM/u/EKd32lhOvxj5MKVY89h4N9XhZC\nbBSqkdiUZJHHRclBiha7RmICnJSreargri8TkFvxiHWTz11E1bsi2EfVO8DNLkE3ORWJfTP85Cy5\nDoG4vBw5Laxc5174zlS4KaIn7QRVj+0wCsnn0TzZkChalqqXOPc6JETHoK5NQgghhBBCiHVBnvfV\nTjTVkDCznwXw42jYYw8A+E/p7+8C8CwAO9x9LH3uEIAPpH+vAPgxd3/QzPYD+CCAXanOIXf/w/O9\ndwBDDziN1efI3TIGsIOqx2bCj7d6CSIlE3KHA4zWi1S9r89w51z0ZNnpNDFVr0pue9WV4XnBy+S0\nK3ZEYjDL9dDP1XZS9TIhd25GQB4gF5AjCBZ0UfUQKyIhxJK4A3XutaidaJohYWZ7AfwMgKvcvWxm\nfwfgLQC+COAT+NbR3r8M4Ii732RmzwHwvwC8GkAdwM+7+31m1gfgXjO7w90fXun9EziK4OS8GrkG\nIQ9u9406OvcA3epE9RJVr5qvUfXYFQPk7Bx0Z7iCPVnuKbOHKBeTd0Zo3M8uS+4nW4m5eXWDXRdT\n9WLy5OiccW/82V2qhBDL0dk1Es0+k2QAdJlZBkA3gBF3v999SZf3VQA+BwDu/giAg2a2y91Pu/t9\n6d9nARwFsLfJ6xZCCCGEEOLCSZLV/duENC0i4e6nzOx3ATwNoAzgM+7+mRVe8u8A3gjgX83sOgAX\nA9gH4Oz8E8zsIIBrANy9lICZvQPAuwEMZq2LltrEZtwmqXpjeJqqV6vPUvXE+onqY61ewoqUyOkv\nFfIsBPbcjOE8N72EmY40U+NGJofy3MuDkSMcUZtfdEs17jwebmwSqNenyIpCiCVxwNr8fHUhNC0i\nkdY8vAHAJQD2AOgxsx9e4SXvAzBoZkcA/DSA+4FncnbMrBfA3wN4l7svOe7X3Q+5+2XuvqMQcHO3\nhRBCCCGEWBveqJNYzb9NSDOLrb8TwJPuPgoAZvYxAN8O4G+WenJqHPyn9LkG4EkAx9Lfs2gYER9y\n94+t5s0dQJ1UJB2RJ5TOGdcTVKlPU/WqVW57WrF+nNzp4WmsWFq0ZgrVF1H1BkJu/RDZCY59PeR6\nKWK72/Eqed4N+ZrGLgafde7ch1Hy3IfhniuoenHCbZ87R1UTQqxIB0ckmmlIPA3g28ysG43UplcD\nuGe5J5vZIICSu9fQ6Ox0p7vPpEbFXwE46u6/t9o3dySoGOdCk3dutxEnd4GaKD5G1XN6Ca1oF0am\n7qTq1Qe4Nze7/HKq3kXO7ewzE3HPBT3Os3SYhdsAcLbMPU99rcRN9TkTHKPqsVN92KlNCbl4Wwix\nQXR416ampTa5+90AbgdwHxqtXwMAh8zsZ8zsJBr1D18zs79MX3IlgAfN7FEA3w3gnenfXwbgRwC8\nysyOpP9e16x1CyGEEEIIQWF+IJ2KrdeOu/8qgF9d9Oc/Sv8tfu6XAXyLO9Ld7wKwZrddI7WJYwFm\nyR76mDwBNI6XLBkRoulMlZ6g6hV6ubVNl9geqh5zEjUAFIipTWWyw4tdvD1LbjIxG41Q9dqdXMht\nd8su3hZCrMAmNRJWQ8dOtq6jhjHjdDPKB30UnXlm62eoekK0ioGug1S9/Qk3rzwkD6Zgz7nwtftI\nlmWyyr1QnYq43duqAffWdbrETW1ik89wjWJ2jYQQYqPYvIXUq6FjDQkhhBBCCCFaynxqU4fSsYZE\n4hHmYk6xWxHjFJ15ytEEVc+IXk1AxdZi9fSFF1H1EvLJtpJwC1TduXMkjPhd681yzwOX5Aeoel1V\nbiF9qY97Xh6d43Zt2p15LlUvBjcldqbMjegkipgIsTzsNnhtRAcbEjHthp3dgrNW59Y06MZfrJa+\n7stavYQNZW83dyjlMLc7LbLEdhe1hGtI5Mh5XN0B1wjrxjaqXhhwjxVrXi8TIcRmwh2od27XtY41\nJIQQQgghhGg5ikhsPhyOOKlRtKq1cxSdeVz9wEWL2J9/MVXvOcFBqt5ggXtK2lHgeoX7styLQT7g\n6RXr5EpwMr3ZkKo3VNtF1Qv6XkrVu8L2UfVKMfe6cbJwgKo3W3qcqidE5+AAObOlnehYQwJIkJAM\niSDg5jOwa25kmHQuZtwb4Yude3Nz+QA3XWUbOXWoEHC/bH2Z9vUqdRNbyQLA9gLXMCnWuYbE3mg7\nVc8S7netv4u7vSHZUCwk3BoYbo8vIToIhyISQgghhBBCiHWgrk1bG3Y3CkUQxGphF/qXnBOlmye0\nbqpeX4a7vaFxvUCzZK/wdMTTK8fctc1E3M+uXOfu29GE6wN3466PPbwwa1zBwcx+qt4o7qHqCdEx\nKCIhhBBCCCGEWDsO1ONWL6JpdKwh4V5HLRpr9TKEaCvKxo2unS2zaxC4Xlcje3HZXv8MOWLCJEfu\nXloj7wt2BOGJ6EtUvULx1VS9DLmdbESeNC6EWAaHiq2FEJ1BljxQjc0EN/MKMfk+PXauYI3opMqR\ni63Jm4oCeS7FTuMWCx/L9FP1upCj6rGNYiHERuFKbRJCCCGEEEKsAxkSQohWMNT7XK5ewC2OrpJd\n/ufKVDk6w3muV7hGvLhE5Mg5O4LQQw6G9We5l6+gyk0dKoMbXss6d3vzQR9Vj92qmt1oQoiW4VDX\nJiFEa9ieu5yq9+w+7qCGvVy7BFly9sZYlSs4VuUaThliax92VyR21yGQO17NRNzud+wb4W5wv2s1\nj6h6swl70Grn3igJccEoIiGEEEIIIYRYE66uTUKIFvF8u4Kqd7CHKoehHPfkSG7ahELI9TLvKlDl\nkIC3wecq3EnKzLQrAChyHeqYIc/36Q93U/V2ZbgHSyXmFm8/FM1R9YQQK8DuXtFGyJAQoo2Zjcnp\nDHVuovoObvYGcgE3PcLYLTOda+l0EzstsffFmQpXr5dcI7Evz7WKu2oHqXoX93ENu9mInHo1u42q\nJ4RYhg0cSGdmbwLwXgBXArjO3c87KdLMbgXwCXe/3cwOA9gNoAwgD+D33f3QSq8ndwoXQgghhBBC\nNEjbv67m3yoxsxtSA2AxDwJ4I4A7L2DBb3X3FwJ4GYD3m9mK4VBFJIRoY+ac28aoVOd6cWfrXF9E\njlzhW0m4elnyADnm6qYi8mdH7sgVk2txi+ScY6PuDe6+BfhpfwO4iCsohFieDera5O5HgZXnzljj\nwT8G8BoAJ4BlW8z1AigCWPFkK0NCiDZmyLg3/sN57s3hMLlGokBObeom10iwDaeIaOiMVrj7drLK\n3bfsG/8iOe1vxopUvbEqt0bCyTnWdXBrTIQQy7C21KbtZrYwHenQ+VKL1sFNAK4AcBWAXQAeBvCB\nBY9/yMyqAC4D8C53lyEhhBBCCCHEhuMOrL4995i7X7vcg2Z2Nxq1C70Ahs3sSPrQLe7+6VW+xysA\nfDg1EEbM7HOLHn+ru99jZjsAfMnMPuXuTy0nJkNCiDam4txe+f0ZrlfzQE+JqlcIudt7qkhuU0WG\nGZEYynFP56Fxi4WHnKt3usTN9SnWudXl7ALEkJz2tzPhpjYdpaoJ0Vk4qdja3a8HGjUSAG5295sp\nwku/16iZ3QfgegDLGhIqthZCCCGEEKJZuK/u38ZwJ4AfMLPQzHYDuHGpJ5lZN4BrADyxkpgiEi0g\nmxmm6uWy/VS9Yvk4VU+snyzZKzxLni48F3F7erLPo6WY6ysZrXL3B7MYvEpuL1jgbiqdgFzOXAnY\njQ24+6NGLtYsGnd7M5lBql69PkXVE6JlbGz715vQKKTeAeCTZnbE3V+76GkfB/AqNGojngbw5UWP\nf8jM5tu/3uru9670njIkWkA95p4gHRvTDUBsPIWAezc3kOUeK71ZbsHrUIGbXtKT5aZK1ZIBqt5p\n8hA5JnPkAXIhuetQlpzqU/JZqt65SnsXM+edO3gkl+E6tGRIiI6CbEi4+2EAh5f4+8fRMBRWeq0D\n+KllHrthrWuRISGEEEIIIUQz8LXNiNhsyJBoAe5cr7A8N53LzgI3dWggy23ByU5FCshzGjIhd3vZ\n14Ia8VTAnsEREaduA0CZnOrDbiebBddD35flfnfZEZ1KlRutY1/XhOgkfPVdmzYdMiSEINKV30fV\n68ty7x7KMVevGnNTc+aqKw7QXDPVhLu+KnnAHZM+ckeuHLkVRxd5pkfi3Bv1kVI3Va8vy91e9n3I\nZMB1QFVrZ6l6QnQMG1gj0QpkSAghhBBCCNEsZEgIIVZDQp77kGvfWlwAwLkqNx2kTI5wRAnXKzwZ\ncfVKxC5a2YB7oRrj1r2j4ZbjMVbhftciq3H1yDcOtZir15v0UvXyuV1UvXL1JFVPiJahGokLw8xC\nAPcAOOXurzezSwB8BMA2APcC+BF3r5nZxWiM6N4BYALAD7v7yVTjAIC/BLAfjavR69z9eLPXLsRa\nqdbOUPWmuPc2iJ2bmpMLuHnqc3WuITFa5Z7iiuT2uTXixWWG3GXJyFlc/eQ0vVrC3bfn5vqoevSa\nFfKNSMm4wySDQH5JIZZl42ZEbDgbMZDunfjmoZfvB/D77v5sAJMA3p7+/XcBfNDdrwbwPwH81oLX\nfBDA77j7lQCuA3Cu6asWQgghhBDiAnAAnqzu32akqS4EM9sH4HsA/AaAnzMzQ2MIxg+lT7kNwHsB\n/BmAqwD8XPr3zwP4h1TjKgAZd78DANx9rplrFuJCMON+pdgFr+zoKtsTkSen57DTfQpkPebhwo5I\n9HNrmenMMFteAdgW9FD1BvPswnzut60WccOdldoEVU+IjsHB75bQRjQ7FvkHAH4RwHzMeBuAKfdv\nJJKfBLA3/fnfAbwRwB8CuAlAn5ltA3A5gCkz+xiASwD8C4D3uPu35FSY2TsAvBsAd8QmmTDkXrCy\nGe6QrEp1hKq3lXByjQT73DNe496MdIXcVKQ9XdwhXj0Z7v5IvIuqVyR20Zomp8EVyP1GyZk+9FSf\ni3u5HcMu6eGub5ycGvZwkXvdcG/vAXxCtJLNGm1YDU1LbTKz1wM4d77R2gv4BQCvNLP7AbwSwCkA\nMRrGzsvTx18C4FIANy8l4O6H3P0yd99xgcsXQgghhBDiwphv/7qaf5uQZkYkXgbgP5jZ6wAUAPSj\nEW0YNLNMGpXYh4bBAHcfQSMiATPrBfAf3X3KzE4COOLux9LH/gHAtwH4qyauvakkSZmqF8fczjli\n/Ri4XkN2wWue7DrIBVw3S2hcvdi5EZPteW6EIx/x1jeV5W5rjpzG1e6XSHJTJLAdkOxEqSz58p8k\nikgIsSyKSKwdd/8ld9/n7gcBvAXA59z9rWjUP3x/+rS3AfhHADCz7WY2v55fQqODEwB8FQ3jYz7K\n8CoADzdr3UIIIYQQQlBwhyer+7cZaUW/tlsAfMTMfh3A/XgmsnADgN8yMwdwJ4D/BgDuHpvZLwD4\nbFqsfS+Av9jwVRNxcrJcVFeRW7sQhNwWkuQ0dewke9R35LmJ+bmQ+92YjrgVw7PECAIATNd5vpxC\n2N4XIfY1sivD9YNVyCGJehtPQQeAbuN+N9iNJtj1ZkK0lA6OSGyIIeHuhwEcTn8+hkYL18XPuR3A\n7cu8/g4AVzdvhUJweCaoxmEbOWttV4F74z9c4KYzZELuXAp26+5KzN2/EXF9NfLwvYkq90a4TL5R\nz5OtbLbRzty3AH+GSZFcHK0bfyGWwQFn5062EZogI4QQQgghRLNQREIIsRpymX6qHrtlZinmpuYU\nI+4pZIAckWCnSrUzFe5Hh5gczqmR1zcbcfftUJ773SB3awV5efRi60J+D1WvFo3TtFQILlpNJ7d/\nlSHRAWQy3LEZ9foUVW8rsa3rMqoeeyBdRM7bDo17sznQU6Hq9TtXb7zCzTU7UeblqbNvXIe5YxWQ\nI1vFJWJ9CQDMkQfczZLXR7abkCWnYdaiSaqebv5Fx+BQREIIIYQQQgixNhyKSIg2RxGE9mG0dJSq\n1x2+lKo3nIuoeoVMmxdHV7mdaebq3PwSZiejUbIDtyfDjSDUyTt3ssY9li/p44ZgLipw7xwqxCno\nAHCgp0DVG4qvoOqNzx6h6gnRMhSREEKsljDg3oxEzr15iNl65FSpXJ5rmPT0c7tUXVrl7t+JGi8t\nkW3kVMk1DQVyW6Rtea6ROF4htx7ubtqYJgqVmLu9ubCHqidEJ5GQz6fthAwJIYQQQgghmoEiEkKI\n1VKNuGlmXznH7c3eHXZR9fbUuaeQcIyb/tKV5aa/jFe46SDM7KGeDPezYxfml+rc9bE96uyuTQNZ\n7vqK5OLt2YjrIo01R0KIZVGNhBBiVRi4F3v2NOATJe76ekLuzde5MtfQyZFTkWbJhhPz2sLu6jOY\n4x58sxHXMKmxB9yRM5ESchohu71vAO768kEvVY9JEHC7rYUBN40rqk9Q9UT7wa7/aydkSAghhBBC\nCNEMHAA5wttOyJAQggjbs1SMuak5tYQ8hIo88G1nV5mqxy4Gf7LIjZicqfDc4DPcQwUZa+8LX3+O\nGw0rsQfm1bmf3xxZjz1wcC46Q9VjYsY979VjdUoUq0ftX4UQQgghhBBrxw0JuX1zOyFDQog2JiRP\nn91GnlbMhl3MzG6JWiNHOIaJdQgz5BoENvmQXEhPnnMxWuYWC4/luJdXdrH6ZMyd+l6Li1Q9JnEb\nr01sDRSREEK0hN1d3Dv/y/u4cxWGctybr8t2jFP1QvLN6wMjO6h690/xUqWMXDzbS+4CVeDadDhJ\n7kQwW+fmhlVi7uV1Lmrvas3+/F6qXrF8nKonRKtwAE5uvtBOyJAQQgghhBCiGTjgKrYW7UwY9lP1\n4niGqifah6mI+5XvzbT3uM6EfPKOEm6qGbOlp4PrsR6tUuWQC7j7YqrGPfaqztVjz7XOkSeDV8GN\nThrIISchOgi1fxVUCvk9VL0+st5U6RhVTz2y1880+WYpS+7E00ce+HZycoCqN17h9o+fJddcXNbL\nSw0bqXBP55M17rHCnolSIN9Yv3CQOwfhBUPc7+5MRDZi68NUvVz5hVS9EdxJ1ROilSi1SQghhBBC\nCLEm3KHFG761AAAgAElEQVSuTYJLpTpC1avVRql6iZMb0ot1s6OL+xUdzpHzVchcvJ3bn/2SgOsG\nf5hcbD1Z4+1fdnH0yRJVDjlyrk9Ijq6Rx1IgIqfVFelzJLhtZM4F7TtHQojWYopIiPZGN/6dyxgz\niR7A2WqWqretwDVM2DUNlQp3e6fJNSYzRL1pcvtXcuYQPRUpSzYS2V2v2CnRGbIhlifvj3P1x6h6\nQnQS7GtbOyFDQgghhBBCiGbgKrYWZHLZ7VS9Qm4bVW+m+ChVT6yfx+qnqXrf6dzC/LmI6/F//By3\nADQw7tm7VOe6hUcqPL2IPPCIHUFgr2+mxhW8tI+7vbvy3BkroXEv1+wITC7soeoJ0Sls5BwJM3sT\ngPcCuBLAde5+zypecyuAT7j77WZ2GMBuAGUAeQC/7+6HVnq9DIkWEAbc6b3ZgDfUSrQXB2wnVa8v\nw7352tlVpurt2z1J1WMTPLWLqldLumlaYzWukVPjdgdFOSbXcNRmqXr7qoNUvbE8ebI1uViTbdiF\n4DoVmIYOuzWyEGuFbUiY2Q0Abnb3mxc99CCANwL48wuQf6u732NmwwCeMLNb3X3ZK4IMCSGEEEII\nIZqBG+KYPVlmmbdyPwoAtkIzCms8+McAXgPgBLDsUJleAEUAKxZrypBoAYlzQ96l2hhVT7QPT/s5\nql4h5EY42BGE7r1cN2lEDnD0ZLiNDXbkeXqVJEfTAoCzFaocfY5Et3FnhExWuQucIEeIitzLBs5V\nuDt4tPQIVQ9GbKNFvuYKsRYaqU2rfvp2M1uYjnTofKlF6+AmAFcAuArALgAPA/jAgsc/ZGZVAJcB\neJf7ytM6ZUgIIYQQQgjRJJLVpzaNufu1yz1oZnejUbvQC2DYzI6kD93i7p9e5Xu8AsCHUwNhxMw+\nt+jx+dSmHQC+ZGafcvenlhOTIdECMiG3RqKQ4eb21upcN24cF6l6W4mX9+2j6r1490mqXqaLG0Eo\nnyZ7cWe4Xvp8luvZzAa8z68n5O6LXQXuYAVyiQS9Noy9vn1d3GMlJudYnyxyIzp9CbeRw3jEjcYK\n0UpYNRLufj2wYo0EDXcfNbP7AFwPQIZEOzFceBZVbwAXUfUqOe5QsGJZhsR62UOuozdyFyMnD8nK\n9XNvhsM8N32jfIJrmOSIhkSBbEjkA65Rx55zMV3jHss7CtztZc+5IMthdzd3e59b/A6q3r/aQzQt\n9/YexCk6G8eaIhIbwZ0AfsLMbgOwE8CNAP528ZPMrBvANQB+eyUxGRJCCCGEEEI0A9/Q9q83oVFI\nvQPAJ83siLu/dtHTPg7gVWjURjwN4MuLHv+Qmc23f73V3e9d6T1lSLSALLhu5m7vper15XZT9Yrl\n41S9rcQs2eM/Os09VnZeze0RmrlqB1UPJe76qrNzVL1Ts7z9USF3BamQJ7HOcOvUcabM3beX9XNT\nfZjRJgCYqnEv11PkiM6ZcISqV8jxWi2XKk/TtIRYO0ZPTXT3wwAOL/H3j6NhKKz0WgfwU8s8dsNa\n1yJDogU8x6+k6g1kuf27S/VLqXrngiPnf9IqSRLu3IJ2ZzbiXuxzmRWbL6yZpMy9WfJxbhpcPMo9\nXsYn+ql6lZhXh1Al3/hnyQ60QW5WGHZ3cQUr3K8G3bBjp0rt5JbqoTDLHUhXjcapekK0ijZMbaIi\nQ0IIIYQQQogmsVGpTa2gaYaEme0H8EE0etQ6Gr1w/zCdlPdRAAcBHAfwZnefXPC6l6CRr/UWd789\n/dtvA/geAAGAOwC8Mw3NbErYUzYrMdcrPI0zVL2tFEUw436lzpS4nV9OF7leQ/t37rE8/GSJqsc+\nxU1WuOkvZ6u8aOJkxPWAs+c+hOTraB85ZOLEScoA0EeO/rHZked25XpxYT9V78kar2PdTPFRmpYQ\n64F9Pm0nmhmRqAP4eXe/z8z6ANxrZncAuBnAZ939fWb2HgDvAXALAJhZCOD9AD4zL2Jm3w7gZQCu\nTv90F4BXYoncsGZh5AtMIeB+7AM5rt6zys+n6p3AZ6l67YyTBx8dq3OHDd41xq1/eWMX10gcvor7\n+Rm5E4/dz22NXKzzDInxGrf2imzDIkse7DpBHiB3aR/ZKM5zOwUxjxUAmCHXX7FTrzIBORdOiBbh\nG1hs3QqaNrPb3U+7+33pz7MAjgLYC+ANAG5Ln3YbgO9b8LKfBvD3ABY2kHYABQA5NCrIswDONmvd\nQgghhBBCsEhgq/q3GdmQGgkzO4hGL9q7Aexy99PpQ2fQSH2Cme1FY2z3jQBeMv9ad/+ymX0ewGkA\nBuBP3P3oMu/zDgDvBkCd0JbJDDHl0JXh2m/kCDWCTXowdyI9zvUyD+XIBZtDs1Q96+YezEZ2g4fs\nYnXi7siRPcJzZI/1JDmCcLbCnREylOV6wEPy/mB3fWGf5SvkiX6hcdMImeSy26l67tz05Kg+QdUT\nF4bDEJObYbQTTTckzKwXjSjDu9x9xuyZD9Pd3Z6ZkPUHaIz4ThY+x8yeDeBKAPMJk3eY2cvd/V8X\nv5e7HwJwKH0d7awWkEOsfeSbm15yrnDeyDdzxEsWu76k3dmW6abqZckD6R4a4bZr3fkv3FSpwR6u\n3qkp7hT5uTrvu1Zktwouc4+VKrmWq04+F7AH+g2R0/4KGW6u2ZPkydZsQ6JY4022zmV30rQAfgpr\nvc5NmRTtx2aNNqyGphoSZpZFw4j4kLt/LP3zWTPb7e6nzWw3nkljuhbAR1IjYjuA15lZHcBlAL7i\n7nOp5j8DeCmAbzEkhBBCCCGEaCc2b3ug89PMrk0G4K8AHHX331vw0D8BeBuA96X//0cAcPdLFrz2\nVgCfcPd/MLMfAPCfzey30IjGvhKN6MWGUY+56Rv8EDDX0p1ybuccI0Z0POEWMLY743XuvpiJBqh6\nx4rcZvR9We7UMm78AOgme4WZnYLKMdvjxT1PlWJuWljFuQPpJmvcY3muyvX4FyPu5bpYJ8+5MG5E\npys7TNOaio7RtAAgjmeoembkctVOvmvdhGiOxPp5GYAfAfCAmc1PJPtlNAyIvzOztwN4CsCbz6Nz\nOxqjvB9AY398yt3/T3OWLIQQQgghBA92e+l2ommGhLvfheXruV59ntfevODnGMBP8Fa2duKYO213\nskpucQluW0A2yRaLIjAZCU9Q9c5UuBGJgz1cL2R/nutljiJuvc9UlVsvNVHjrS/i7grMRNzzVHfI\n3Re1hOvxH6tyvcKnitz6pmlyRGKEPN7naI07fyggzuRhagEAe0KIsQvLnduIYKvVJtJxzZEQZKrO\nPQ05eTf2GrdTELPDRS3izlVod7LOvcCw+z3vIN/4Z0LudyOKuTevIblYndknoSvkro2dHUG2c+h6\n/VnuBg+Rvxtd5O/GwR7ud+M1toeq95XxPprWF8FNbWLT7kNb2bO0tpph4jDE3rRpCy1nVVtmZi9P\nh8Ut/NuLmrMkIYQQQgghOoPEV/dvM7JaV/anAXzVzN7k7vNdlv4SQBsbEwYjhTPDsJeiM8/OPNfL\n3E9uJztaIU8DJoeVtxL9zi0X7s5wPUsnStwC1VrCPfbYXuGRMnd7mQyQPeqFkLsvJiLuvhgDu2Um\nd+p7hlx8nAm4+6NAjmD1ks8tAyHvOskuZu7pOkjVi8hNNdjtZBPnNsHYiqhGAngUwO8A+IKZvd3d\nvwT+PBsyBiPNQ8hnuTdzNbLZSb7eoy/LrbnY3fMCmtYIeXAPuyNXGHDTwnYF3JqG3gz32GOnl2TJ\nQ7zqZMOkjzyQrpeol1S49RvbC+RJl+Cub7gxy5TGeI17SRspcWskInLXF7b3k+zPonJF73dT9QJy\nkmjJuF2gKs7VK8ftPedibObfWr2EFWl0bWr1KprHag0Jd/dPmNmjAD5qZh8AuzegEEIIIYQQHYba\nv6bRB3d/3MxeAeADAK5u2qoIhEEe/d3PpmhlyF7mkRrXC769wI2YbMtzPZEHa1fStIpd4zQtAHCy\nT93JEZP+LLmQnpz+sj3H7eyzkzwNuEKcHA0A3W2cpTeQ5W7rri6u1/Uct5EMejLc9c2SszeqCffG\ngT15uzvkrm+OnHrFPPddXOOmrSXkTgSlZIiqN2vce4yZTD9Vr5hwr+ObgU72vK/qm+ru1yz4eQ7A\nm83sQNNWRSBOypice6jVy1iSwhA3XWVmbh9Vjx1mPVW9l6ZVqjxN09oMfD7kGolnTr6YqveKndz0\njb4stwYhF3BTkU6UuE6Fp0q8m//HprmXqpNlbtvrkWCEqjcZcVsjX1q65vxPWgMTVe53l3zfj9EK\n1wlwus69OTwZPErTmqg8QdMCgHyG11EKAEo17mcXx1yrPSanAGOL1U26A/WtGpEwsz/GyobUz3CX\nI4QQQgghROfgW9WQAHDPgp//bwC/2sS1bBnq5GExpYAbQThTeZCqV65yPZFbiakqNwIT5K+l6lVi\n7snx0VnyDBNy8fZ4jZu+8TTR6f9A5TRPDMBcOE3VY89EMXBTuQJy/5Aucgihlzl0BECRPKxxOpng\n6tVO0rTqZA99T45b6J8JuNG/au3c+Z+0BoKA/N01bkOXJCFHTMg4+I1J2okVDQl3v23+ZzN718Lf\nxfoZtL1UvX3JfqqeF7iH/GzpcareVqJITuWayXHbDJ4sci8wiXNvbnYUyB3SyE6lHmKEf8C5ecxh\n0t7pB7FdRNWrkweFRuSkaHa3vx6yYdJV66HqhcT0l74Cd1hef8g99qJwjqqXCbnngnrMdVaacdPq\nNgPq2tSggz8GIYQQQggh2JjmSAguU36Kqjfo26l6CdkzxyQgd9BKEm6XID7c6NBEyA15R0l7z1g5\nznX0oZ/sxa3EvO09ER6naQH8zio9wTaqXtW5O3cQ3PNouc797lbq3GNvssZtU1Uz7sBBZrfEKOam\nDp2tP0zVmytzi8HZ10mQ7wkS31oRiS09R8LMZvFMJKLb7BvtfAyN2RLc+JkQQgghhBAdRLxVi63d\nndvjTAAATk99kapX7Bnj6lV4RW4AEIa83Nlshts6t1Jt74gEey5F1blFaZcPcIOazx/gejUrMbc4\nmtmuFQDGqjwt9jTbuhMXB6A/GabqVY3bevhswK1Hmq1yi9WHyJ9fZNyIxJRxm2pMFB+jaTn5WHay\nh559ng/JEYkucnH5XPkYVa/dcd/CEQmxOahG3AtWPstNQRgo8IrByxG3M0ilzTtK5bLcdIv9yeVU\nvQPd3AvqHvJAOiM7gXLkC/TJIq97yU4/SNMCgFHyjfXjtbuoeoUMN62OfXNYD7k3r93WS9WbDsiz\nCxKuYZLN8LY3E3DPoxXydahen6Lq5bPcAXeZIEfV24qoRkIIIYQQQgixZhSREG1NtXaGqheQezyH\nxB7U1WiSprUZqEXctLVu47ZrPVPhpg5FSXtnU47XuF6lmRovpSExbnrEALgtLjM57rE3GT1J1WOn\nl/SRW4TG4Baohs69/FfJLUILxDRW9uTohFy8zaZEjrSzvxtbjS09R0JsTRLnhqg1R2L95LI7qXpZ\n4974swfSFckNw7hbC3RzSySwvYu3wjM1Xk45APSQ00FCcB0U7JvDoa5LqHpOvnUYwSNUvXrMTb2K\nyB3wanVePVd/1z6aFgAUq9xUn3KVW5cYt7mhsxXZypOthRBCCCGEEOvAAdSV2iTE+jGiF1wh1vYi\nNO7ZcVuOG5LoDslzOGrcU2Yu4HmpLvcX07QA4KE6tzg6Sy5UZ0cQcgG3mHku5s5sCckpp10huyCX\nm7oWJ7yICVMLAIJAt05ibXSwHSFDohWwh8VkQm5eeT3mtghlZgey2/i1O3FSafUSVmSAe2+DvgzX\nkNhZ4LaTzQbcy8HX53gpEn1BgaYFALvxXKpe3rnrO4kHqXrF+ihVLzC20ck1dNipVzVyOk2S8GpC\nSjG5RiLhnldEZ9MYSKfUJiGEEEIIIcQaUURCUDFyiDpO2IVV7PQhdsnr1qGnsJuq98Jhrlf4NXu4\nXtyLdnI7v0RVbnX05AluJ55txJrNkDw0owZu8WwMbhOHgFy8Xa2fpeoN5g9Q9XrBne8zlnC7XlXI\nsxCYERN2lJ098E10OBpIJ9jE5DZ57Hat7K5NYv0EZCPseQPcfbtjeI6ql+vhpjaViQPfAH7njaEc\n7+qyLU/uilThtloexB6qXp48oK2Q5Q64Y3O2zu3aNFk8StXLhLx2rQBQyPFqOOKYmyKaOLcVb0Cu\nL0nINSHiwuj09q9yFQshhBBCCNEUDLGv7t8Fv5PZsJndYWaPp/8/r0VuZofN7Nr05+Nm9oCZHUn/\n/4bzvV4RiRbQ7sXWtYjbbUSsn/4ct/95lHA96sfPcDu/VE5xU5FqCXnwA5nZOm9/FOvcaM50jdvb\nvi+3g6qXs26qHruLETsVyTLcY9l6ucX07Mi4O+94DgvcW53pIndmi7IAOh8npzaZ2Q0Abnb3mxc9\n9B4An3X395nZe9Lfb1mj/I3uPmZmVwD4DIB/XOnJikgIIYQQQgjRBOZTm1bzj8AbANyW/nwbgO9b\n/AQz6zKzj5jZUTP7OIDlvNv9AM6b46qIRAtgtzANAq4nLZ/jFpTWojGalhnXK9fuuaQD4O6Lp0rc\nz6+acKNhA1muV304x/X0ReQaiTliROKpaIKmBQAHctdS9aZxhqpXQD9Vrw/ciMm2hKtXD7nHctW4\n9U2lGrfxArNmJRty62m6C/upenPlY1Q90X6sodh6u5nds+D3Q+5+aA1vtcvdT6c/nwGwa4nn/CSA\nkrtfaWZXA7hv0eOfNzMDcCmAN5/vDWVItAD2ULWoPk3VSxJutxbm9jq5yK3dear2b1S9R6a4nWQy\nQ9yg5rN7ucdeH9mQmI64p8wCcS7FEMgD1cDtdLPLL6XqReD28h91bhejKOQ6KdhzH3oCburVRPQo\nVS+KSzStfJZbCF4sc48V0fms4Uw/5u7LenHM7G4AeQC9AIbN7Ej60C3u/ulvek93N1tyauwrAPxR\n+pyvmdnXFj0+n9r0LACfNbPD7r6s50GGhBBCCCGEEE2gMZCOpOV+PbBijcRZM9vt7qfNbDeAdRe9\nuvsTZnYWwFUAlvVqypDoAGLyRFHRPvRml4pKrp9MwE3NuajA9ZJmA67ebI1bAEr++NCb5UUkBrPE\noRQAHiVHOovGbScbOc9jDfBbLVeXd+Ctix3OjSZmyXM46n3cCMzJqS/QtNhzJAr5vVS9Wp3bEj5J\nuMceu1LYO3o82xI4EG/cJv8TgLcBeF/6/6UKpe8E8EMAPmdmzwNw9VJCZrYTwCUAnlrpDZtmSJjZ\nBwC8HsA5d39e+rdhAB8FcBDAcQBvdvdJM3srGlXlBmAWwE+6+78v0AoB3APglLu/vllr3qwYuHc3\nYYYbBq6TBxVtJTLG7S9++QD3WHnZRdwOX3uv4F5Q63Pc7f3qUe4sBBBv5nIhd1v76tzUl3P+dape\nlty1iW2YZMD97k4HvFozAKgkXEORfW8YEusa2NegfIFs1GV6qHqV2jhVr052KvgW61LFjEisgvcB\n+DszezsaBsBSNQ5/BuCvzewogKMA7l30+OfNLEbjAvUed19xWmczIxK3AvgTAB9c8Lfl2lI9CeCV\nqVHx3QAOAbh+weveicbGcqvrhBBCCCGEaCJsO8LdDwM4vMTfxwG8+jyvLQN4yzKPHVzrWppmSLj7\nnWZ2cNGf3wDghvTn29D4EG5x9y8teM5XAHyjeb6Z7QPwPQB+A8DPNWe1mxt2mJDdyYgZMdlqIVF2\nOsMLBrj7dvel3AhCdge5V37A7QI1mOcW+M5N8LzqoxXuvq0E3JTJPLkYfJtzZ6zMBeRUroQbQWBv\nbxhwU5tmktPnf9IaMGKqWV/3ZTQtAEjIhe/ViHvsBQE3zVFzLi6cDYxIbDgbXSOxmrZUbwfwzwt+\n/wMAvwjgvH0mzewdAN4NgNc3TlwwQci7Wdpq9SAHAm4LyYP95A5f5OvL1IPc9JzpaW572iK5a1Mp\n5m3vGLhG3faY23p4IuDWIEzaitH2NdPt3JTOPttJ1ZsA90ad7ZPJWIGqF4Y8vVzITR2KyJ0Nu7Lc\nFu5BwD1PVaojVL1MhnuLVq9zjXY2DsDJKejtRMsG0rl747NdgJndiIYhcUv6+3yNxeL8reU0D7n7\nZe7OvfsSQgghhBBiHSS+un+bkY2OSCzbliodivGXAL47zfECgJcB+A9m9joABQD9ZvY37v7DG7lo\nM669ZSAPVWOHHcmzGlzNwdbNtgL3s5utcUPep5/menFrMfe7MVPlbu+D01zP5mSVd+Xo9uWGk66P\nXnJ6xFkjD38kX3R7nRu9YlM1bjF4H7jF9CXjRsR6cryITkxO1w2Ne17uzXKjf8U6twkGm63WgMWx\noV2bNpyNjkjMt6UCFrSlMrMDAD4G4Efc/bH5J7v7L7n7vrT44y0APrfRRoQQQgghhBDrYpXRCEUk\nFmFmH0ajsHq7mZ0E8KtYvi3VrwDYBuBPG1O5UV9pst9Gw55EzZ5QyoYf4VCh1nqp1LlnlhNFrtd6\ntMJtcdkVkqf3ZrjHXnfI3R8DOZ4vZzjDzVGfqFeoemHILe5NwC2knyS3Vz2QXEzVi5w7U6Zq3P1b\ncfIsBOJ1o0aurWPfE7DXV6lNUPXEhdPJjWKa2bXpB5d56FvaUrn7jwP48fPoHcYSra6EEEIIIYRo\nRzZ4jsSGo+R1sakIQ+4okTjmetHYjFS5edGPzHJbcD6rh1tPc92BM1S9rn5uRKLnOHd7SzGvW8uJ\nIre+ZLzO9boGzs2knY25x0po3IhJCdy893rAPfamwf38Zmvczj7FKq8rVz3m5uRnQnJjSG5wTbQh\nHWxHyJAQQgghhBCiWSgiscVhd21iw87XzOe4nrR8ltfZJ6pzc0nLbR6RCMnHXuzcXtYv2Mb19A1d\nyfW6IuBub/85bl75JT28bjJZcpelq2rcjlxfHeV2vBpNuOepcwHXQ59PuBGOroTbZSkJuNeN7Tnu\n8My5PO/cUk4maVoAcHr2PqpeNsONFA/m91D1JoqPnf9Ja6DdMwHYdHrXJhkSQgghhBBCNAlFJLY4\nbI9/u+PkORLl6ihNK6pvrW4UfeRON8/tr1H1Bnq4E16twI3ABH1cL/3wTm7NytAMzxM5USPn+HNL\nLnBpP/lyM8OOcHCPPXYXrWLMrfepGfe7O+3TVL3QecdzQK5/CcnRv1yGO8Ok3TtDbjkccBkSQggh\nhBBCiLXgQEebdjIkxLdQi7j91MX6qcTcdh57urhezb4h7sRYr3FrGuJJ7vqq5M5ITC/VdMT1qI9X\nuftiJuK65EbJ9VK1kOuhJ5fn0OulxqKvU/Um5x6i6oG4vT0Fbv1GGHCjTfWYe+zNFB+l6okLRxEJ\nIYQQQgghxJpRREII0RJOGzc69Ogst5tH9Nhuql7hGDcCUyfnvT85x50MPl7jRThK5F70UzWuC22s\nzK29mjNyRMK59S91sgsyIeuVI269WT7HnbwdJ9x6LiZRPEfVi+vc+hJ2TQhzyvhWxOGIO7jaWoaE\nEEIIIYQQTaJzzQgZEh0Be9pzknA9c+wuUFuJWeN6DWeifVS92Tr3FDJe43ZDKdW5EYlpsl6ljSfa\nVsmNz7cVuPUllRJ3rsKIP0LVq7e5B3Jb12VUvUrM9aozZxfMls7RtDYDrghCW+FQ+1chhBBCCCHE\nWnEZEkIIIYQQQoh14B2c3CRDogMwclvAIMhT9eJYqU3r5UXBVVS9nXnuvsgH3F4UUcLtmTmY425v\naOz2rzy9yLmf3XCenBZGLt7uDrn7Ymf8bKpe0uY3Dn3YQdUrY5Kql83whjXG8QxNS1w4Ztxbz3ZP\nn1ZqkxBCCCGEEGJdxB08SEKGRAdQr0+1egmiSXRluF7hQsj13GQD7snxucPcY7mnizuQbmK2h6p3\n17khmtZxbjdU+oWP3Uf9XMQd4rUD3OLtK4a4kd0i2en6cJEbQWC3CE2S9vYyi/XT7hGEZtDBdoQM\nCSGEEEIIIZqBQwPphBAtYnuBG5F4/jauF3LvFdzc4/xVfVQ9n+GevpN/5XrBrxninYILIS+nHABO\nlbnH3miFKof9BW50aLLG9agPcjsZoy/LrYG5Yu75VL17cJaqlwkLNC3yrhVizXgHhyRkSAghhBBC\nCNEM1P5ViAuD2QUqSbg572wK+T1UvYu6qHLY/3xuBCH3/S+k6vnO7VS94NwYVW9vgTckCwCGj/K8\nuLvIRRKTZe7Bd6bE1Xt4lluDcPc5bheoUp0bQQi4chjOckMm3catMckGvOMlqs/RtAAgIef4B+Qu\nRlGdO8iU3RnSjPvdTZJZqh6bRmpT51oSMiSEEEIIIYRoAg5X1ybR3mQyg1Q9dheodo8iMIljbiL4\nYJab45+9lJtHjy5eHjMA2CzXc4inTlPl4nHu/p08xfO6Tpe5+2KyyvUaTkdcj/90jeuiz5Jd/nu7\nud/dvgz5XBBwL//3jPdT9UrJOE2LPbcgJOuxS3EDy1L12B253Lm1ZpuBDrYjZEgIIYQQQgjRLJTa\nJNqawMjtQcS6Geg+SNUjp0Wj/jQ3jz6sHKHqeZHr+So9wfX03f3IXqreo7O8KMJ0xD1aCtwAAiox\nV+9smbtvT9dKVL0zZa6Hvpzj7t9inXtj0+vcyPhcyOvKlc1007QAvse/VOV2vNqKcxramcZkaxkS\nQgghhBBCiDXiikgIJt2FA1S9ntxOqt5cwI1wVGojNK1clrut7G4UbE8Vu1PLk/cOUPXCgOsVrtS5\nnsNqnexWJ3NxNy8CU024B4uT42GnK9zLzVCeu75dEffYG8hxbxy6Qq5ekRzBKhm3c06v8c71lZBb\n99ebvYiqt6fwAqpeybnzgs7M3k/VixNuZHwzTEHXQDohhBBCCCHEmmh0bepcU0KGRAuoRrxuFM3Q\ny2a4XmsjeumrtTM0rWZQq41S9Urxa6h6T85wJ0dXE25E57IB7pyL3UNcL6lxHZs4PsfLA6/E3H3R\nk+EWNeQDrkf9ZJF7Yd7VxY1e7chx1zcXcyMIUzXu/s05d07IaPJ1mtZU8ShNCwBmguNUva7cLqpe\nhpXX50AAABbLSURBVDi7CQByWW69T7HMnXOxGejkYmvulUcIIYQQQggB4JmBdKv5d6GY2bCZ3WFm\nj6f/H1rFaw6b2bXpz8fN7AEzO5L+/w3ne70iEi0gjrn5gWzafX3tzMWD3AjCJd01qt4L956j6mVz\nXK9mEJJ75ZN7+bNhNvI4NsedEXK8xK2VmuU25MJkjfvd2NfD9aiTmyLROdDLjcA8OkE+Xmq8GTDs\nukR2Tv5c+RhVj002M0zVC0NuhCOO2z/C4eQqCTO7AcDN7n7zoofeA+Cz7v4+M3tP+vsta5S/0d3H\nzOwKAJ8B8I8rPVkRCSGEEEIIIZrC6qIRpPSnNwC4Lf35NgDft/gJZtZlZh8xs6Nm9nEAy3lR+gGc\nt3JfEYkW0JXfR9XrK3B720+VnqDq1aIxql47c0lyBVVvTze3t/3O7+J6DW2Yl+MPANjG9VQhz93e\n7q8dp+p1fZFX35Q7zfV47e3idiB7eIZ7rHSF3MtXjtwirZscXXOy3y8id/maC6apelGdd+4LyFO8\nA/Jka3a3vy7yPUFIrrmYKT5K1Wt35lObVsl2M7tnwe+H3P3QGt5ul7vPh/POAFiqAOcnAZTc/Uoz\nuxrAfYse/7yZGYBLAbz5fG/YEkPCzI4DmAUQA6i7+7Vm9iYA7wVwJYDr3P2e9LmvAfA+ADkANQDv\ndvfPtWLdQgghhBBCrB5HbKtOhxtz92uXe9DM7gaQB9ALYNjM5qfC3uLun/6md3V3M1vKgnkFgD9K\nn/M1M/vaosfnU5ueBeCzZnbY3eeWW1MrIxI3uvtCV/WDAN4I4M8XPW8MwPe6+4iZPQ/ApwFwze0N\nph5zvcxxUiXrlal6W4mBkOu5GerldjGyqw5S9ZKrn0vVQx+3qxRmuV2bgoSb+N5f4UX/tt1Jrm0i\nRxB25bl55Tu7uNEmNr0ZbkRiIMutR5qOuOeqvqnz1nSuiVyGV/NTqpygaQGAg7svnNwatFg+TtUT\nF8YaIxIra7lfD6xYI3HWzHa7+2kz2w1g3YWR7v6EmZ0FcBWAf1vueW1TI+HuR939W+Jd7n6/u89P\nNHsIQJeZcc+AQgghhBBCNIFklf8R+CcAb0t/fhuWLpS+E8APAUDqoL96KSEz2wngEgBPrfSGrYpI\nOIDPpCGXP19D/td/BHCfuy/pgjezdwB4N4BBzjKFEEIIIYRYL07v2rQC7wPwd2b2djQMgKVqHP4M\nwF+b2VEARwHcu+jxz5tZDCAL4D3ufnalN2yVIfEd7n4qtXbuMLNH3P3OlV5gZs8F8H4A37Xcc1KD\n5FD6/LZtvlevc8fXj8+2f+uzrcLubm7RXLVG/oqe5ha+Bz3ktoUZ8vae4Q4IjO7kbm/paV7B6+nJ\nbTQtAJiucVOH5urcAPgMeaBaTB6u+Ogc9/MLubXReJqcCXfOjlP1yjXeuWpb3wtoWgB/4NuZ6bup\neu7cNEJxYTiAxMjNF9wPAzi8xN/HAbz6PK8tA3jLMo8dXOtaWpLa5O6n0v+fA/BxANet9Hwz25c+\n70fdndtSSAghhBBCiCaxgalNG86GRyTMrAdA4O6z6c/fBeB/rvD8QQCfRCO88sUNWmZTCUJuQWkY\ncD1fW6ldK5td3JlWCAPuiaX0uRUjlGsm+CJXrzbN9W3MTXG/Gw+c4bZuLsW8oWCzEXfA2P93invs\ndWe465uJuAPp5iKuy39XF9drXSMX+h+d4Tb9yAXdVL0kWbZJzJqZrY6c/0lrgN3+VRGEzsbhiNG5\n+7gVEYldAO4ys39Howr8k+7+KTO7ycxOAngpgE+a2Xwbq58C8GwAv5KO7D6SpkQJIYQQQgjRxjgS\nxKv6txnZ8IiEux8D8C0Ji+7+cTTSlxb//dcB/PoGLG3DyBLb2gFAf2E/VW+K66jaUhEOdh5zTzfX\n6xqVuL6Dyii3JiSOuR/gU+PcvgvlmPv5hcRSrip5wFg+5G7r4+Upqt7p4DhVrwfc9qXlca6/K2/c\niM4pO0XVKyXc2r/uwgGaFnugWkD2wdaz3GMlqnOvuUYemGfkiE4cc4chNoMNLLbecDTZWgghhBBC\niCbgcHqxdTshQ6IFsIfPRDEvlxQAorq6QK2XLDlZMCQPtcp2c/WmJ7mnkJNT/VS9yRrXk1YhRyQq\nxE5BpyvkrkgRd9BlhuzF3e7cuaSjxh1adpp84zCQkCMmAXfYZX3pruzrpju7naZVjbmDKXuzu6h6\nZuwIxwBVb7b0OFUP4B4rm4HNmra0GmRICCGEEEII0RQ2dI7EhiNDogWwIxKZgNsqqLfrWVQ9vjeD\nR0D+7MrkHP8g4HZqyQ5x13dRH7cZ/c4aV4+cioszj3Drmx4Z53mZKzE3DzxKuOepsnG9kNPBOFUv\nSrjFYRG5i1GWfLkeBNer7gHX42rg1YQMZw7StAAg8gpVLyFHnkNyTUMUc8/LFXIXrXbH4Yg9avUy\nmoYMCSGEEEIIIZqEK7VpaxOQOz5kM2RPFdnz5WTvSJHo9WfnkhZyO6h6l/Rwe0V3b+fqZV90EVUP\nF3GnKWMbt8sSxrmdgvZ94RGq3uBjPC9435NcD/PxIjf68iC3qQ/yMXdGyP785VQ95wYTERE7fAFA\nXOfe2JTI85EuivfQtKrG7X4XOvfWqUqO1mUTcje9Lq43vZbnXofGZv6NqsfHN+2wudUgQ0IIIYQQ\nQogm4FD71y1PknC9BcMFbg3CALjWfTHkug4zfbyITo047RTg16uwJzzmDnKjYbjyEqqcb+d1VgEA\nVLm5x3ZmlKrnda5XuF7jHTHkwce4qp9cI1EvUPXGKlyva3+OO6fBySGJmLx/e7LcDmkDVW69WdV4\nEZMaeXI027vc49zPLgH3YHmOv5iqFwbcK+UdaP+IhLtSm4QQQgghhBBrRKlNgkqpzu02wt6Lc9EZ\nql6xyvMK18kTLNldmx6Y5no1fzAmn3ye5E6ztSe4vffjo2epepVj3Nzeygz3y/bYKV5Ep5Zwjz3m\n1G0AKJNd6kVyjn+F/F0rJVwveEyOnubIk7Ij8o3SNHjR5xP2IE0L4HaUAoCcca9DMbjH3jD2UfUG\nE+6ci3bH4UjUtUkIIYQQQgixNlRsvUkxBKReyk7OryxH3MnRWbJXvVrnTgEF8QvUld9N0wK401MB\n4Mad3Bx/r3FPPj7CrX9JRrm996sjXC/z5BluR7ORaW5nmhMl3nd3osb1krKZjbjH8lPgRk57Y+6+\nnQjOUfX6wJ1snXdu16uScc8FJeNdh3bi2TQtgJ+mMgdulkKpzo3sRgF3306QO022PQ7VSAghhBBC\nCCHWiiZbb0rMQmQynDy83gLXC74ty/WObI+566sVrqLqzQS8CEw54XrUewJuRGJ3N9dzY13cr6hX\nuNG1aJQcQTjFja7Nlrldr/rz5A5uNW7nISZnKtxj75E6tz4nQ+7lH4A79f2gH6DqGXl9hQw3glUm\nT1avJryuUnPkaAn7WOlBD1VvzrjRsOkKtxYum+Fub7vj4HeIbCc61pAQQgghhBCitbgmW29ODKyu\n/rV6kaIzz0j9fqreWPgYVa+ecKeAVmq8/M+ozq0vyee4MzjOlX+YqufRDFUv6OXmRWcGuP3AB7Zz\na0wKRW6njFwX92JQPd6+p+ARckTimvx+ql6J3LWJPYeD3cs/S+69nzWuV32qzo3WxcQbr27ynIY5\n494TVI173kvIdZ295EnUsXOPlc1AQu7i1k6071VMCCGEEEKITYyra5MQQgghhBBiPahGYhMSWAY9\n+R0UrW05cus4chuwBNz0jbmYN0AOAGrEdrJBwC3oY7f2nauTCxgf564vyHJb+5anuKeQwRdy0y16\nL+OcA77BbJkqd+XOMZrWJWe5qS87nhqm6j2vv0DV+8QIN13l/ukpqt4OckFpjTwwzwPud20g5J6b\nmalhUzH3e8tORZoFtzi6WOPqZUPusZzPcFsttz3uav8qhBBCCCGEWDtq/7oJCSyDngzHGzlV57Y+\n6wq5g4XYlu5M+SRVLwh4h1kht4umBQDViDsI6FiRW8ycxGQP/cs4LZHn6T5A9vhv5343QG5xiQmu\n1zo3wBvMFD7EHdC2r8Ld1nCUe+xN17ge8KmAu73bnevF7SEfy1MRN5Idk1M3IqLeRMBtG16xOa5e\nzG2qEdWnqXqV6ghVjz3ytv1xpTYJIYQQQggh1oaD30mrnehYQ6KAbjzHX0TR6stwvcxsz1KR3AZx\ntpvzuc3DHAY0FpymaQHAyeJXqXqPz3BbPua3kb0Yu7heYTonuF51dk1DcpLrta6P87zCk1/neuhn\n5rg1DcWIe7mJyP1arwz3UfV2kYdJ8v2Z3GGIFXINxyPJ0zStXufm5O93bjvUbHg5VW+q9yVUvapx\nW8Kzr+Nfn/h7qh4fRSSEEEIIIYQQ60CGxCYkRoxJ4+QJlmNud5DuhOs5nAY3X7Mc8CIIADAHXh1C\nJeHmfibk4XvFiOslPfsYN896F56i6hEbcgEA7n94N1Vvezc3D7we91P1zpR455bhPHfI03SNG4k9\nW+HqheSBapN17rmgl9zBLcttyoXxGvd4MXD3B1NuPDjLEwOQj7nHMnvo8WjA7dpUM25kdyI6RtVr\nfxzNiCm2Cx1rSAghhBBCCNFSXBGJTUk5mcaDpU9StPoKeyg688Qx2XNY4lr3ScL1Poj186mnuB76\ni85yPfSzEdfreqbK1auM9lL1yD5XbM/zIlhHZ7mRUzZFcq3huSr3PPW4PULVO1nhdiBje4WTgOsG\nLzj3uzblp2haI9N307QAYKLnCqpeYNxbsZkit/MiO3Lfk+d2X2x3HGr/KoQQQgghhFgzDneuE6+d\n6FhDwj1BPeYkcI/P3E/RmccCbn6lk70FYv3cFz1B1RuauoyqN9bFPfa6yWeQiSrX57+3mzwNmFsC\ngzHi9tYS7md3tsz97KZqXL27qrdT9Yrl41S9bIY7GTwT8maOAEC5yvVas2sknDjZms347JFWL2FT\nM1WfaPUSNhh1bRJCCCGEEEKsCxkSm5AYcVxs9SKWxBNujYRoH05U76XqHUm4k6OvCwepegNZrtdw\nfzdXry/DPXnP1rmtc5ideNge3JjsEJ6KyB3SyBEENjG51ixqcy9uO0cQhGgtDnRwRILcUE4IIYQQ\nQggxj///7d1/rFd1Hcfx58tUqIBNxV8LA3RQEpkWGsvF0NXm1qam5dTmpFobFfpHs2mrP1qtjVrL\n1WKVazn+KUa0lOUf0EB0tUzI8DJhoJAtCDUJcnf+KPDVH99zx7fb5X7PufdwvtxzXw929z2/vt/z\nvu9z+Nx9vp/z+XxK/hsvSWdL+q2kZ4vXnqM+SNoiaVGx/LykHZK2F6839Hp/i1skIpo3Z8riWj/v\n8mn1tiBMP6Pe55jPn1rvUDynq95vbV4/Vu93JRdMrbfD3NtqnO15+5F6f9cXXqu3BeEl1TefDJz6\nz+Rn9LuIOK7ev22SlgLLbC8btus+YJPtlZLuK9bvrfjx19h+WdK7gI3Aw6MdnBaJiIiIiIiTotPZ\nusxPDW4AVhfLq4Ebhx8g6a2S1kjaJenXwInGDp8BHO51wja3SAzC0d39DmKSmQm83O8g+mnH4Qdr\n/ryeh0z6nPdBct4fyXvzkvPmJefVze53AD1sgKMzSx47VdK2rvUHbD9Q4Vzn2z5YLL8AjDRpx+eB\nV21fKuky4Klh+x+VJOBi4JZeJ2xzRWK37UX9DmIykbQtOW9Wct685Lw/kvfmJefNS87bx/Z1dX2W\npD8CU4BpwNmShsYivtf2hmHntaSRntlcAvygOGZA0sCw/UOPNl0CbJK0xfbgiWJqc0UiIiIiIqIV\nbH8QRu0j8aKkC20flHQh8NI4zrVX0ovAAuDJEx2XPhIRERERERPfeuDOYvlORu4o/ThwO4CkhcBl\nI32QpPOAucBfRzthm1skqjxTFvVIzpuXnDcvOe+P5L15yXnzkvMYj5XAWkmfpVMBGKmPw4+AByXt\nAnYBwyfAelTSMeAM4D7bL452QtmZRCYiIiIiIqrJo00REREREVFZKhIREREREVFZ6yoSkq6TtFvS\nc8WsfjFOvXIqaYmkpyQdlfSJYfuOFVOtb5e0vrmo26XENVjeNa397yQt6EecE13Z8kPSzZIsaVGx\nPkfSa133+o+bi7o9yuRf0i2Sdkp6RtLPm46xDUqUJ/d33ct7JB3p2pcyvQYlrsFsSZskDUjaImlW\nP+KM6KVVfSQkvQXYA3wU2A9sBW6zvbOvgU1gZXIqaQ6dGRDvAdbbXte1b9D2tCZjbpuS12CG7VeK\n5euBL9Q5dvVkULb8kDQdeAQ4E1hhe1vxf+A3thc2GnSLlLzP5wFrgWttH5Z0nu0xD284GVX9Oynp\nLuAK258p1lOmj1PJe/2XdMqU1ZKuBT5t+46+BBwxira1SFwFPGd7n+1/A2voTBceY9czp7aftz0A\n1DK/e/yfMtfgla7VtwPt+YagOWXLj28C3wZebzK4SaBM/j8HrLJ9GCCViDGp+nfyNuAXjUQ2eZS5\nBguAzcXyoyPsjzgltK0i8Q7gb13r+4ttMXbjzelUSdskPSHpxnpDmzRKXQNJX5S0F/gOcHdDsbVJ\nzzxLej9wke1HRnj/XEl/lvSYpA+fxDjbqsx9Ph+YL+n3RZmSVrfqSpfpkmbTGUd+c9fmlOnjV+Ya\nPA3cVCx/HJgu6ZwGYouopM3zSMSpYbbtA5IuBjZL2mF7b7+DaiPbq4BVkm4HvsbxSWmiBpJOA74H\nLBth90HgnbYPSfoA8JCk9wxrKYrxOx2YBywFZgGPS3qv7SOjvivG6lZgne1jXdtSpjfjHuCHkpbR\nmUDsAHBs1HdE9EHbWiQOABd1rc8qtsXYjSuntg8Ur/uALcAVdQY3SVS9BmuAfFNYXa88TwcWAlsk\nPQ8sBtZLWmT7DduHAGz/CdhL59vzKK/Mfb6fTj+s/9j+C53nzOc1FF9bVClPbmXYY00p02vR8xrY\n/rvtm2xfAXy12JYKc5xy2laR2ArMkzRX0pl0CsGMKjE+Y86ppLMkTSmWZwJXA+n4Xl3Pa1B0Qh3y\nMeDZBuNri1HzbPtftmfanmN7DvAEcH3R2frcogMlxTe184B9zf8KE1qZsuYhOq0RQ2XKfJLnqkqV\n6ZLeDZwF/KFrW8r0epQp02cWraAAXwF+1nCMEaW0qiJh+yiwAthAZ9rvtbaf6W9UE9uJcirpG8Xo\nQEi6UtJ+4JPATyQN5fxSYJukp+l0FluZEbSqK3MNgBXFcJjbgS+Rx5oqK5nnE1kCDBT5Xwcst/3P\nkxtxu5TM/wbgkKSddMqULw+1BEU5Fe7zW4E1/t+hHVOm16DkNVgK7Ja0Bzgf+FZfgo3ooVXDv0ZE\nRERERDNa1SIRERERERHNSEUiIiIiIiIqS0UiIiIiIiIqS0UiIiIiIiIqS0UiIiIiIiIqy8zWERGn\nKEnnAJuK1QvozGz7j2L9Vdsf6ktgERERZPjXiIgJQdLXgUHb3+13LBEREZBHmyIiJiRJg8XrUkmP\nSXpY0j5JKyV9StKTknZIuqQ47lxJv5K0tfi5ur+/QURETHSpSERETHzvA5bTmXn4DmC+7auAnwJ3\nFcd8H7jf9pXAzcW+iIiIMUsfiYiIiW+r7YMAkvYCG4vtO4BriuWPAAskDb1nhqRptgcbjTQiIloj\nFYmIiInvja7lN7vW3+R4OX8asNj2600GFhER7ZVHmyIiJoeNHH/MCUmX9zGWiIhogVQkIiImh7uB\nRZIGJO2k06ciIiJizDL8a0REREREVJYWiYiIiIiIqCwViYiIiIiIqCwViYiIiIiIqCwViYiIiIiI\nqCwViYiIiIiIqCwViYiIiIiIqCwViYiIiIiIqOy/lMA9ae1SESYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f6d217d9350>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_mel(t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {},
   "outputs": [],
   "source": [
    "b = model.predict(t.T.reshape(1,t.shape[1],t.shape[0]))[0].T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### conv2d - mse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAxIAAAEYCAYAAADIwMCEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f6d232d9250>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_mel(b) #(b-1.)*80"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### conv2d - bix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 331,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAxIAAAEYCAYAAADIwMCEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8dwAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f00ed4e0dd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_mel(b) #(b-1.)*80"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1, 128, 32)"
      ]
     },
     "execution_count": 293,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "t.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T10:24:59.198625Z",
     "start_time": "2017-11-17T10:24:59.081762Z"
    },
    "_cell_guid": "0c99ba3b-e8ca-40cb-8d29-2b0e89a385c7",
    "_uuid": "429139ca4f71487c6cfe3e8dfbb6a659eb9bb9c8"
   },
   "outputs": [],
   "source": [
    "model.load_weights('./weights/starter_pyramid_conv.hdf5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T10:32:14.882322Z",
     "start_time": "2017-11-17T10:32:14.863617Z"
    },
    "_cell_guid": "c6d9b369-9979-4bcd-8540-4653e6544f84",
    "_uuid": "6a0bb3c22b7b5c43db0ec5673333ab3de8f08724"
   },
   "outputs": [],
   "source": [
    "def test_generator(test_batch_size,augment=False):\n",
    "    while True:\n",
    "        for start in range(0, len(test_paths), test_batch_size):\n",
    "            x_batch = []\n",
    "            end = min(start + test_batch_size, len(test_paths))\n",
    "            this_paths = test_paths[start:end]\n",
    "            for x in this_paths:\n",
    "                x_batch.append(process_wav_file(x,reshape=True,augment=augment,pval=0.5))\n",
    "            x_batch = np.array(x_batch)\n",
    "            yield x_batch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "metadata": {
    "ExecuteTime": {
     "start_time": "2017-11-17T10:32:45.947Z"
    },
    "_cell_guid": "1fb8aed4-de12-43c5-84bf-b803e3d640fa",
    "_uuid": "631a38cb0013e5772f6987854145ad76ecf6c430"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2478/2478 [==============================] - 1230s  \n"
     ]
    }
   ],
   "source": [
    "predictions = model.predict_generator(test_generator(64,augment=False), int(np.ceil(len(test_paths)/64.)), verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save(\"cache/predictions_pyramid_noaug.npy\",predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2477/2478 [============================>.] - ETA: 0s"
     ]
    }
   ],
   "source": [
    "num_aug = 2 \n",
    "for i in range(num_aug):\n",
    "    predictions +=  model.predict_generator(test_generator(64,augment=True), int(np.ceil(len(test_paths)/64.)), verbose=1)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = predictions/(num_aug + 1.0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T11:30:44.236246Z",
     "start_time": "2017-11-17T11:30:44.21858Z"
    },
    "_cell_guid": "b1cdab5c-9816-4690-87d8-de2c97cf0e7d",
    "_uuid": "24eb7e512eace4567494e0a8e356a826f4283c4d"
   },
   "outputs": [],
   "source": [
    "classes = np.argmax(predictions, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T11:31:11.212517Z",
     "start_time": "2017-11-17T11:31:10.786357Z"
    },
    "_cell_guid": "1da523cf-fdbf-4ab1-9300-0147155aa247",
    "_uuid": "f25d4e626202aa115bd0460f4de8d07f9727c83e"
   },
   "outputs": [],
   "source": [
    "# last batch will contain padding, so remove duplicates\n",
    "submission = dict()\n",
    "for i in range(len(test_paths)):\n",
    "    fname, label = os.path.basename(test_paths[i]), id2name[classes[i]]\n",
    "    submission[fname] = label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-11-17T11:32:05.154527Z",
     "start_time": "2017-11-17T11:32:04.983371Z"
    },
    "_cell_guid": "9a95d147-3f4b-4386-8597-5fa60be43542",
    "_uuid": "bdf63bce43a0525a02ac18ca3f90aeba06ce6e99"
   },
   "outputs": [],
   "source": [
    "with open('subm/starter_submission{}_3xtta.csv'.format(exp_name), 'w') as fout:\n",
    "    fout.write('fname,label\\n')\n",
    "    for fname, label in submission.items():\n",
    "        fout.write('{},{}\\n'.format(fname, label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "_cell_guid": "8bea6850-15c6-44e7-bdb4-9555ad196f85",
    "_uuid": "555315ef622793711ff5643928dac874c8cb0ed2",
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.14"
  },
  "toc": {
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": "block",
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}