MILNET Attention-lr_0.001-Variant.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Which weights to choose? 1:normal 2:lle 3:iso3\n",
      "[0 0 0 ... 1 1 1]\n",
      "data loaded\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/owner/anaconda3/lib/python3.6/site-packages/h5py/__init__.py:34: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n",
      "/home/owner/anaconda3/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: compiletime version 3.5 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.6\n",
      "  return f(*args, **kwds)\n"
     ]
    }
   ],
   "source": [
    "import sys, os\n",
    "sys.path.append(os.pardir)\n",
    "\n",
    "os.environ[\"KERAS_BACKEND\"]='tensorflow'\n",
    "#import glob\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "weight_choice = input('Which weights to choose? 1:normal 2:lle 3:iso')\n",
    "\n",
    "\n",
    "\n",
    "#from multiprocessing import Pool\n",
    "#import multiprocessing as multi\n",
    "#from data.func import load_npy, padding_mat\n",
    "#sys.path.append('C:\\\\ProgramData\\\\Anaconda3\\\\pkgs\\\\pydot-1.2.3-py36hd4f83f9_0\\\\Lib\\\\site-packages')\n",
    "#sys.path\n",
    "\n",
    "\n",
    "path_to_15_40 = 'milnet+edu/data/Preprocessed/15words+40sent/'\n",
    "path_to_20_50 = 'milnet+edu/data/Preprocessed/20words+50sent/'\n",
    "\n",
    "path =path_to_15_40\n",
    "\n",
    "x_train = np.load('/home/owner/デスクトップ/'+path+'train_data_features_sorted.npy')#np.load('/home/owner/デスクトップ/PythonFile/imdb/x_train_sort.npy')\n",
    "y_train = np.load('/home/owner/デスクトップ/'+path+'train_data_scores_binary.npy')#np.load('/home/owner/デスクトップ/PythonFile/imdb/x_test_sort.npy')\n",
    "#print(x_test)\n",
    "\n",
    "x_test = np.load('/home/owner/デスクトップ/'+path+'test_data_features_sorted.npy')#np.load('/home/owner/デスクトップ/PythonFile/imdb/t_train.npy')\n",
    "y_test = np.load('/home/owner/デスクトップ/'+path+'test_data_scores_binary.npy')#np.load('/home/owner/デスクトップ/PythonFile/imdb/t_test.npy')\n",
    "\n",
    "x_valid = np.load('/home/owner/デスクトップ/'+path+'validation_data_features_sorted.npy')\n",
    "y_valid = np.load('/home/owner/デスクトップ/'+path+'validation_data_scores_binary.npy')\n",
    "\n",
    "#train_idx = np.load('/home/owner/デスクトップ/'+path+'train_data_idx_binary_only.npy')\n",
    "#test_idx = np.load('/home/owner/デスクトップ/'+path+'test_data_idx_binary_only.npy')\n",
    "#valid_idx = np.load('/home/owner/デスクトップ/'+path+'validation_data_idx_binary_only.npy')\n",
    "\n",
    "#print(len(train_idx))\n",
    "#x_train = x_train[train_idx]\n",
    "#y_train = y_train[train_idx]\n",
    "#print(len(x_train))\n",
    "#print(len(y_train))\n",
    "\n",
    "#x_test = x_test[test_idx]\n",
    "#y_test = y_test[test_idx]\n",
    "\n",
    "#x_valid = x_valid[valid_idx]\n",
    "#y_valid = y_valid[valid_idx]\n",
    "\n",
    "#print(x_valid)\n",
    "\n",
    "#word_idx=np.load( '/home/owner/デスクトップ/'+ path +'vocab_idx.npy')\n",
    "#print(idx)\n",
    "#mani_choice = input('Which weights to choose? 1:normal 2:lle 3:iso ')\n",
    "if (int(weight_choice)==1):\n",
    "    embWeights=np.load( '/home/owner/デスクトップ/'+ path +'weights.npy')#np.load('/home/owner/デスクトップ/PythonFile/imdb/weights.npy')\n",
    "elif (int(weight_choice)==2):\n",
    "    embWeights=np.load( '/home/owner/デスクトップ/milnet+edu/data/manifold_weights/' + 'weights_LLE_(15,40)_(1000,20,300).npy')#np.load('/home/owner/デスクトップ/PythonFile/imdb/weights.npy')\n",
    "elif (int(weight_choice)==3):\n",
    "     embWeights=np.load( '/home/owner/デスクトップ/milnet+edu/data/manifold_weights/' + 'weights_Isomap_(15,40)_(1000,20,300).npy')\n",
    "\n",
    "#print(len(word_idx))\n",
    "print(y_test)\n",
    "\n",
    "print('data loaded')\n",
    "\n",
    "\n",
    "\n",
    "import keras\n",
    "from keras.layers import Input, merge\n",
    "from keras.models import Model\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Activation, Dropout\n",
    "from keras.layers import Embedding\n",
    "from keras.layers import LSTM\n",
    "import keras.backend as K\n",
    "from keras.layers import Lambda, regularizers, Average\n",
    "from keras.layers import RepeatVector, Permute, Multiply\n",
    "\n",
    "from keras.models import Model\n",
    "from keras.layers import Input, Conv2D, Conv1D, MaxPooling2D, GlobalMaxPooling2D, GlobalMaxPooling1D, MaxPooling1D\n",
    "from keras.layers.embeddings import Embedding\n",
    "from keras.layers.recurrent import GRU\n",
    "from keras.layers.wrappers import Bidirectional, TimeDistributed\n",
    "from keras.layers.core import Dropout, Dense, Lambda, Masking\n",
    "from keras.layers import merge, Layer, Activation, Dot, Concatenate, Flatten, Lambda\n",
    "\n",
    "from keras.initializers import Identity,glorot_normal\n",
    "from keras import regularizers\n",
    "\n",
    "from keras import metrics\n",
    "\n",
    "from keras.utils import plot_model\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "li = [x_train.tolist(), x_test.tolist(), x_valid.tolist()]\n",
    "for idx in range(3):\n",
    "    features = li[idx]\n",
    "    print(len(features))\n",
    "    for i in range(len(features)):\n",
    "        #print(len(features[i]))\n",
    "        if len(features[i])!=40:\n",
    "                print('i:'+str(i)+'   ;;;;;;;;;;;;;;;;;;;;;')\n",
    "                print(len(features[i]))\n",
    "                li[idx][i].append(li[idx][i][-1])\n",
    "        for j in range(len(features[i])):\n",
    "\n",
    "            #print(len(features[i][j]))\n",
    "            if len(features[i][j])!= 15:\n",
    "                print('i:'+str(i)+'j:'+str(j)+'   *************************')\n",
    "\n",
    "x_train=np.asarray(li[0])\n",
    "x_test=np.asarray(li[1])\n",
    "x_valid=np.asarray(li[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(210642, 40, 15) (25510, 40, 15) (25206, 40, 15)\n",
      "(210642,) (25510,) (25206,)\n"
     ]
    }
   ],
   "source": [
    "print(x_train.shape, x_test.shape, x_valid.shape)\n",
    "print(y_train.shape, y_test.shape, y_valid.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "40 15\n",
      "61560 300\n",
      "Attention Model Build Complete\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/owner/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:126: UserWarning: Update your `Model` call to the Keras 2 API: `Model(inputs=[<tf.Tenso..., outputs=[<tf.Tenso...)`\n"
     ]
    }
   ],
   "source": [
    "numSentencesPerDoc, numWordsPerSentence = x_train[0].shape[0], x_train[0].shape[1]\n",
    "print(numSentencesPerDoc, numWordsPerSentence)\n",
    "#print(x_train[0])\n",
    "\n",
    "vocabSize, embeddingSize = embWeights.shape[0], embWeights.shape[1]\n",
    "print(vocabSize, embeddingSize)\n",
    "\n",
    "dropWordEmb = 0.25\n",
    "recursiveClass = GRU\n",
    "\n",
    "filters = 100 #embeddingSize*2\n",
    "windowMin = 3\n",
    "windowMax = 7# dimOfSentimentMetrics = 5\n",
    "batch_size = 200\n",
    "#epochs = 25\n",
    "dimGRU = 50\n",
    "numDensePool=10\n",
    "eta = 1e-4\n",
    "dr = 0.5\n",
    "\n",
    "##\n",
    "\n",
    "#wordsInputs = Input(shape=(numWordsPerSentence,1), batch_shape=(numSentencesPerDoc,numWordsPerSentence,), dtype='int32', name='words_input')\n",
    "\n",
    "x_in = Input( shape = ( numSentencesPerDoc, numWordsPerSentence ) , name='Input' )\n",
    "#x_pop = Lambda( lambda x: x, output_shape=(numWordsPerSentence, ) , name='convert_shape' )( x_in )\n",
    "    \n",
    "#Layer functionの定義\n",
    "embLayer = Embedding( input_dim=embWeights.shape[0], output_dim=embWeights.shape[1], weights=[embWeights]\n",
    "                      ,mask_zero=False , trainable=False, embeddings_regularizer=regularizers.l2(0.0000001)\n",
    "                      , input_length=numWordsPerSentence, name='Embedding' )\n",
    "\n",
    "maxPooledPerDoc = []\n",
    "convNets = []\n",
    "maxPools = []\n",
    "scoresForSentences = []\n",
    "\n",
    "\n",
    "extraDimLayer = Lambda(lambda x: K.expand_dims(x), name='extraDimForConvo')\n",
    "squeezeThirdLayer = Lambda(lambda x: K.squeeze(x, 3), name='squeezeThirdLayer')\n",
    "\n",
    "for windowSize in range(windowMin,windowMax):\n",
    "    name='word_mat_convo_win_size_'+str(windowSize)\n",
    "    #convNet = Conv2D(filters, kernel_size=(windowSize,embeddingSize), padding='valid', activation='relu'\n",
    "    #                 ,strides=1, use_bias=True, input_shape=(numWordsPerSentence, embeddingSize, 1), data_format=\"channels_last\",kernel_initializer=glorot_normal()\n",
    "    #                 ,bias_regularizer=regularizers.l2(eta), kernel_regularizer=regularizers.l2(eta),name=name)\n",
    "    convNet = Conv1D(filters=filters, kernel_size=windowSize, padding='valid', activation='relu', strides=1,use_bias=False,\n",
    "                    kernel_regularizer=regularizers.l2(eta), name = name)\n",
    "    convNets.append(convNet)\n",
    "    name='word_mat_max_pool_win_size_'+str(windowSize)\n",
    "    maxPool = MaxPooling1D(pool_size = int(numWordsPerSentence-windowSize-1), padding='valid')\n",
    "    maxPools.append(maxPool)\n",
    "    \n",
    "    \n",
    "for i in range(numSentencesPerDoc):\n",
    "    maxPooledPerSentence = []\n",
    "    x_pop = Lambda(lambda x: x[:,i], output_shape=(numWordsPerSentence, ) , name='convert_shape_'+'sentence'+str(i+1))( x_in )\n",
    "\n",
    "    for j in range(windowMax-windowMin):   \n",
    "        emb = embLayer(x_pop)\n",
    "        #emb = Dropout(dr,name='DropEmb'+str(i)+str(j))(emb)\n",
    "        #reshaped = extraDimLayer(emb)#Lambda(lambda x: K.expand_dims(x), name='extraDimForConvo_'+str(j)+'_sentence_'+str(i))(emb)\n",
    "        #name='word_mat_convo_win_size_'+str(j)+'_sentence_'+str(i)\n",
    "        #wordsCNN = Conv2D(filters, kernel_size=(windowSize,embeddingSize), padding='valid', \n",
    "        #                    activation='relu', strides=1, use_bias=True, input_shape=(numWordsPerSentence, embeddingSize, 1), data_format=\"channels_last\",\n",
    "        #                    kernel_initializer=glorot_normal(),kernel_regularizer=regularizers.l2(),name=name)(reshaped)\n",
    "        wordsCNN  = convNets[j](emb)\n",
    "        wordsCNN = Dropout(dr,name='DropCNN'+str(i)+str(j))(wordsCNN)\n",
    "        #squeezed = squeezeThirdLayer(wordsCNN)#Lambda(lambda x: K.squeeze(x, 3), name='squeezeThirdLayer_'+str(j)+'_sentence_'+str(i))(wordsCNN)\n",
    "        # newShape = (-1, int(squeezed.shape[1])*int(squeezed.shape[2]))\n",
    "        # squeezed = Lambda(lambda x: K.reshape(x,shape=newShape), name ='squeezeDimForMaxPool'+str(i)+str(j))(squeezed)\n",
    "        #wordsCNNPooled=GlobalMaxPooling1D()(squeezed)\n",
    "        #wordsCNNPooled= MaxPooling1D(pool_size = int(squeezed.shape[1]), padding='valid')(squeezed)\n",
    "        wordsCNNPooled = MaxPooling1D(pool_size=(numWordsPerSentence-(j+windowMin)+1))(wordsCNN)\n",
    "        flattened = Lambda(lambda x: K.squeeze(x, 1))(wordsCNNPooled)\n",
    "        maxPooledPerSentence.append(flattened)\n",
    "        \n",
    "    mergedPoolForSentence = Concatenate(axis = 1)(maxPooledPerSentence)\n",
    "    newShape=(-1,1,int(mergedPoolForSentence.shape[1]))\n",
    "    reshapedPoolForSentence = Lambda(lambda x: K.reshape(x,shape=newShape), name ='switch_axis_'+'sentence'+str(i+1)+'winSize'+str(j+windowMin))(mergedPoolForSentence)\n",
    "    maxPooledPerDoc.append(reshapedPoolForSentence)\n",
    "    \n",
    "    densePoolForSentence = Dense(numDensePool, bias_regularizer=regularizers.l2(eta),\n",
    "                                 kernel_regularizer=regularizers.l2(eta), activation='softmax', use_bias=True)(reshapedPoolForSentence)\n",
    "\n",
    "    densePoolForSentence = Dropout(dr,name='DropDense'+str(i))(densePoolForSentence)\n",
    "    scoresForSentences.append(densePoolForSentence)\n",
    "    \n",
    "#Naive Approach\n",
    "averaged = Average()(scoresForSentences) \n",
    "averaged = Lambda(lambda x:K.reshape(x,shape=(-1,int(averaged.shape[1])*int(averaged.shape[2]))), name ='attend_output')(averaged)\n",
    "out_avg = Dense(1, activation='sigmoid', use_bias=True)(averaged) \n",
    "    \n",
    "#Apply Attention \n",
    "mergedPoolPerDoc = Concatenate(axis = 1)(maxPooledPerDoc)\n",
    "biRnn_ = Bidirectional(GRU(dimGRU,  return_sequences=True), merge_mode='concat')(mergedPoolPerDoc)\n",
    "newShape = (-1, int(mergedPoolPerDoc.shape[1]), int(biRnn_.shape[2]))\n",
    "biRnn = Lambda(lambda x: K.reshape(x,shape=newShape), name ='biRnn_TF_Reminder1')(biRnn_)\n",
    "#biRnn2 = Lambda(lambda x: K.reshape(x,shape=newShape), name ='biRnn_TF_Reminder2')(biRnn_[1])\n",
    "\n",
    "#biRnn_cat = Concatenate(axis = 2)([biRnn1, biRnn2])\n",
    "\n",
    "CONTEXT_DIM = 100\n",
    "\n",
    "#eij = Dense(CONTEXT_DIM, use_bias=True, activation='tanh')(biRnn)\n",
    "#eij = Dense(1, use_bias=False, activation='softmax')(eij)\n",
    "\n",
    "#weighted_input = Dot(axes = 1)([eij, mergedPoolPerDoc])\n",
    "#weighted_input = Lambda(lambda x: K.reshape(x,shape=(-1,int(weighted_input_.shape[1])*int(weighted_input_.shape[2]))), name ='attend_output')(weighted_input_)\n",
    "eij_ = Dense(CONTEXT_DIM, use_bias=True, activation='tanh')(biRnn)\n",
    "eij = Dense(1, use_bias=False, name = 'attention_weights',bias_regularizer=regularizers.l2(eta),\n",
    "                                 kernel_regularizer=regularizers.l2(eta))(eij_)\n",
    "eij_permuted = Permute((2,1))(eij)\n",
    "eij_normalized = Activation('softmax',name = 'attention_softmax')(eij_permuted)\n",
    "eij_normalized = Permute((2,1))(eij_normalized)\n",
    "\n",
    "scoresPerDoc = Concatenate(axis = 1)(scoresForSentences)\n",
    "weighted_input = Dot(axes = 1)([eij, mergedPoolPerDoc])\n",
    "#weighted_input = Lambda(lambda x: K.reshape(x,shape=(-1,int(weighted_input_.shape[1])*int(weighted_input_.shape[2]))), name ='attend_output')(weighted_input_)\n",
    "weighted_input = Lambda(lambda x: K.squeeze(x, 1), name='squeezeOutput')(weighted_input)\n",
    "\n",
    "out = Dense(1, activation='sigmoid', use_bias=True)(weighted_input)\n",
    "\n",
    "##\n",
    "\n",
    "model = Model(input=[x_in], output=[out])#,weighted_input,eij_normalized])\n",
    "#adadelta = keras.optimizers.Adadelta(lr=1.0, rho=0.95, epsilon=None, decay=0.0)\n",
    "#model.compile(loss='binary_crossentropy',\n",
    "#              optimizer=adadelta,\n",
    "#              metrics=['accuracy'])\n",
    "         \n",
    "model.compile(loss='binary_crossentropy',\n",
    "              optimizer=keras.optimizers.Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0),\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "print(\"Attention Model Build Complete\")\n",
    "##\n",
    "#model_avg = Model(inputs=[x_in], outputs=[out_avg])\n",
    "#model_avg.compile(loss='binary_crossentropy',\n",
    "#              optimizer=keras.optimizers.Adam(lr=0.0001, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0),\n",
    "#              metrics=['accuracy'])\n",
    "\n",
    "#print(\"Average Model Build Complete\")\n",
    "##\n",
    "#save model to png file\n",
    "#from keras.utils import plot_model\n",
    "#plot_model( model, to_file='model.png' )\n",
    "\n",
    "#モデルを保存せず直接可視化\n",
    "#from IPython.display import SVG\n",
    "#from keras.utils.vis_utils import model_to_dot\n",
    "#SVG( model_to_dot( model ).create( prog='dot', format='svg' ) )\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<tf.Tensor 'lambda_237/Squeeze:0' shape=(?, 100) dtype=float32>,\n",
       " <tf.Tensor 'lambda_238/Squeeze:0' shape=(?, 100) dtype=float32>,\n",
       " <tf.Tensor 'lambda_239/Squeeze:0' shape=(?, 100) dtype=float32>,\n",
       " <tf.Tensor 'lambda_240/Squeeze:0' shape=(?, 100) dtype=float32>]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "maxPooledPerSentence"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "save_model_path = '/home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_iso_docvec.hdf5'\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#name = './params_milnet_adam_0720_3_adjCNN_fixed_weights.hdf5'\n",
    "save_model = keras.callbacks.ModelCheckpoint(save_model_path, monitor='val_loss', verbose=1\n",
    "                                             , save_best_only=True, save_weights_only=True, mode='min', period=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 210642 samples, validate on 25206 samples\n",
      "Epoch 1/10\n",
      "210642/210642 [==============================] - 305s 1ms/step - loss: 0.4958 - acc: 0.8415 - val_loss: 0.4289 - val_acc: 0.8739\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 0.42892, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_normal_docvec.hdf5\n",
      "Epoch 2/10\n",
      "210642/210642 [==============================] - 299s 1ms/step - loss: 0.4041 - acc: 0.8733 - val_loss: 0.3880 - val_acc: 0.8813\n",
      "\n",
      "Epoch 00002: val_loss improved from 0.42892 to 0.38801, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_normal_docvec.hdf5\n",
      "Epoch 3/10\n",
      "210642/210642 [==============================] - 296s 1ms/step - loss: 0.3853 - acc: 0.8810 - val_loss: 0.3930 - val_acc: 0.8798\n",
      "\n",
      "Epoch 00003: val_loss did not improve\n",
      "Epoch 4/10\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3776 - acc: 0.8852 - val_loss: 0.3738 - val_acc: 0.8844\n",
      "\n",
      "Epoch 00004: val_loss improved from 0.38801 to 0.37384, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_normal_docvec.hdf5\n",
      "Epoch 5/10\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3692 - acc: 0.8904 - val_loss: 0.4015 - val_acc: 0.8757\n",
      "\n",
      "Epoch 00005: val_loss did not improve\n",
      "Epoch 6/10\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3688 - acc: 0.8922 - val_loss: 0.3717 - val_acc: 0.8890\n",
      "\n",
      "Epoch 00006: val_loss improved from 0.37384 to 0.37174, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_normal_docvec.hdf5\n",
      "Epoch 7/10\n",
      "210642/210642 [==============================] - 296s 1ms/step - loss: 0.3646 - acc: 0.8968 - val_loss: 0.3869 - val_acc: 0.8849\n",
      "\n",
      "Epoch 00007: val_loss did not improve\n",
      "Epoch 8/10\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3637 - acc: 0.8976 - val_loss: 0.3803 - val_acc: 0.8892\n",
      "\n",
      "Epoch 00008: val_loss did not improve\n",
      "Epoch 9/10\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3627 - acc: 0.8996 - val_loss: 0.3816 - val_acc: 0.8905\n",
      "\n",
      "Epoch 00009: val_loss did not improve\n",
      "Epoch 10/10\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3608 - acc: 0.9012 - val_loss: 0.4286 - val_acc: 0.8686\n",
      "\n",
      "Epoch 00010: val_loss did not improve\n"
     ]
    }
   ],
   "source": [
    "history_normal = model.fit(x_train, y_train, batch_size = batch_size, verbose=1, epochs=10 #epochs\n",
    "                        ,validation_data=(x_valid,y_valid), shuffle=True, callbacks=[save_model])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 210642 samples, validate on 25206 samples\n",
      "Epoch 1/15\n",
      "210642/210642 [==============================] - 307s 1ms/step - loss: 0.3853 - acc: 0.8551 - val_loss: 0.3813 - val_acc: 0.8752\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 0.38126, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_lle_docvec.hdf5\n",
      "Epoch 2/15\n",
      "210642/210642 [==============================] - 299s 1ms/step - loss: 0.3384 - acc: 0.8747 - val_loss: 0.3433 - val_acc: 0.8777\n",
      "\n",
      "Epoch 00002: val_loss improved from 0.38126 to 0.34329, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_lle_docvec.hdf5\n",
      "Epoch 3/15\n",
      "210642/210642 [==============================] - 297s 1ms/step - loss: 0.3286 - acc: 0.8800 - val_loss: 0.3365 - val_acc: 0.8834\n",
      "\n",
      "Epoch 00003: val_loss improved from 0.34329 to 0.33645, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_lle_docvec.hdf5\n",
      "Epoch 4/15\n",
      "210642/210642 [==============================] - 296s 1ms/step - loss: 0.3246 - acc: 0.8831 - val_loss: 0.3425 - val_acc: 0.8817\n",
      "\n",
      "Epoch 00004: val_loss did not improve\n",
      "Epoch 5/15\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3220 - acc: 0.8859 - val_loss: 0.3457 - val_acc: 0.8780\n",
      "\n",
      "Epoch 00005: val_loss did not improve\n",
      "Epoch 6/15\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3203 - acc: 0.8887 - val_loss: 0.3376 - val_acc: 0.8842\n",
      "\n",
      "Epoch 00006: val_loss did not improve\n",
      "Epoch 7/15\n",
      "210642/210642 [==============================] - 296s 1ms/step - loss: 0.3181 - acc: 0.8910 - val_loss: 0.3359 - val_acc: 0.8847\n",
      "\n",
      "Epoch 00007: val_loss improved from 0.33645 to 0.33593, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_lle_docvec.hdf5\n",
      "Epoch 8/15\n",
      "210642/210642 [==============================] - 296s 1ms/step - loss: 0.3164 - acc: 0.8927 - val_loss: 0.3514 - val_acc: 0.8852\n",
      "\n",
      "Epoch 00008: val_loss did not improve\n",
      "Epoch 9/15\n",
      "210642/210642 [==============================] - 296s 1ms/step - loss: 0.3151 - acc: 0.8943 - val_loss: 0.3388 - val_acc: 0.8840\n",
      "\n",
      "Epoch 00009: val_loss did not improve\n",
      "Epoch 10/15\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3154 - acc: 0.8944 - val_loss: 0.3396 - val_acc: 0.8854\n",
      "\n",
      "Epoch 00010: val_loss did not improve\n",
      "Epoch 11/15\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3141 - acc: 0.8961 - val_loss: 0.3392 - val_acc: 0.8884\n",
      "\n",
      "Epoch 00011: val_loss did not improve\n",
      "Epoch 12/15\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3123 - acc: 0.8978 - val_loss: 0.3383 - val_acc: 0.8834\n",
      "\n",
      "Epoch 00012: val_loss did not improve\n",
      "Epoch 13/15\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3124 - acc: 0.8982 - val_loss: 0.3397 - val_acc: 0.8870\n",
      "\n",
      "Epoch 00013: val_loss did not improve\n",
      "Epoch 14/15\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3106 - acc: 0.8998 - val_loss: 0.3443 - val_acc: 0.8851\n",
      "\n",
      "Epoch 00014: val_loss did not improve\n",
      "Epoch 15/15\n",
      "210642/210642 [==============================] - 295s 1ms/step - loss: 0.3094 - acc: 0.9011 - val_loss: 0.3385 - val_acc: 0.8870\n",
      "\n",
      "Epoch 00015: val_loss did not improve\n"
     ]
    }
   ],
   "source": [
    "history_LLE = model.fit(x_train, y_train, batch_size = batch_size, verbose=1, epochs=15 #epochs\n",
    "                        ,validation_data=(x_valid,y_valid), shuffle=True, callbacks=[save_model])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.load_weights(save_model_path)#('./params_milnet_adam_0718_2_fixed_weights.hdf5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 210642 samples, validate on 25206 samples\n",
      "Epoch 1/15\n",
      "210642/210642 [==============================] - 307s 1ms/step - loss: 4.3865 - acc: 0.7322 - val_loss: 4.3508 - val_acc: 0.7336\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 4.35075, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_iso_docvec.hdf5\n",
      "Epoch 2/15\n",
      "210642/210642 [==============================] - 304s 1ms/step - loss: 4.3701 - acc: 0.7322 - val_loss: 4.3463 - val_acc: 0.7336\n",
      "\n",
      "Epoch 00002: val_loss improved from 4.35075 to 4.34627, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_iso_docvec.hdf5\n",
      "Epoch 3/15\n",
      "210642/210642 [==============================] - 303s 1ms/step - loss: 4.3682 - acc: 0.7322 - val_loss: 2.3144 - val_acc: 0.7336\n",
      "\n",
      "Epoch 00003: val_loss improved from 4.34627 to 2.31436, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_iso_docvec.hdf5\n",
      "Epoch 4/15\n",
      "210642/210642 [==============================] - 304s 1ms/step - loss: 4.4360 - acc: 0.7287 - val_loss: 4.3498 - val_acc: 0.7336\n",
      "\n",
      "Epoch 00004: val_loss did not improve\n",
      "Epoch 5/15\n",
      "210642/210642 [==============================] - 304s 1ms/step - loss: 4.3702 - acc: 0.7322 - val_loss: 4.3467 - val_acc: 0.7336\n",
      "\n",
      "Epoch 00005: val_loss did not improve\n",
      "Epoch 6/15\n",
      "210642/210642 [==============================] - 303s 1ms/step - loss: 4.3685 - acc: 0.7322 - val_loss: 4.3458 - val_acc: 0.7336\n",
      "\n",
      "Epoch 00006: val_loss did not improve\n",
      "Epoch 7/15\n",
      "210642/210642 [==============================] - 303s 1ms/step - loss: 4.3681 - acc: 0.7322 - val_loss: 4.3456 - val_acc: 0.7336\n",
      "\n",
      "Epoch 00007: val_loss did not improve\n",
      "Epoch 8/15\n",
      "210642/210642 [==============================] - 302s 1ms/step - loss: 4.3679 - acc: 0.7322 - val_loss: 3.0663 - val_acc: 0.7336\n",
      "\n",
      "Epoch 00008: val_loss did not improve\n",
      "Epoch 9/15\n",
      " 93800/210642 [============>.................] - ETA: 2:42 - loss: 7.1176 - acc: 0.5629"
     ]
    },
    {
     "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-6-8fa3fa044f81>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m history_ISO = model.fit(x_train, y_train, batch_size = batch_size, verbose=1, epochs=15 #epochs\n\u001b[0;32m----> 2\u001b[0;31m                         ,validation_data=(x_valid,y_valid), shuffle=True, callbacks=[save_model])\n\u001b[0m",
      "\u001b[0;32m~/anaconda3/lib/python3.6/site-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, **kwargs)\u001b[0m\n\u001b[1;32m   1710\u001b[0m                               \u001b[0minitial_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minitial_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1711\u001b[0m                               \u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1712\u001b[0;31m                               validation_steps=validation_steps)\n\u001b[0m\u001b[1;32m   1713\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1714\u001b[0m     def evaluate(self, x=None, y=None,\n",
      "\u001b[0;32m~/anaconda3/lib/python3.6/site-packages/keras/engine/training.py\u001b[0m in \u001b[0;36m_fit_loop\u001b[0;34m(self, f, ins, out_labels, batch_size, epochs, verbose, callbacks, val_f, val_ins, shuffle, callback_metrics, initial_epoch, steps_per_epoch, validation_steps)\u001b[0m\n\u001b[1;32m   1233\u001b[0m                         \u001b[0mins_batch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mins_batch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtoarray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1234\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1235\u001b[0;31m                     \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins_batch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1236\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[1;32m   1237\u001b[0m                         \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2473\u001b[0m         \u001b[0msession\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2474\u001b[0m         updated = session.run(fetches=fetches, feed_dict=feed_dict,\n\u001b[0;32m-> 2475\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2476\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   2477\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\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~/anaconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\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~/anaconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\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[0;32mNone\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~/anaconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\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~/anaconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\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_ISO = model.fit(x_train, y_train, batch_size = batch_size, verbose=1, epochs=15 #epochs\n",
    "                        ,validation_data=(x_valid,y_valid), shuffle=True, callbacks=[save_model])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "history = model_avg.fit(x_train, y_train, batch_size = batch_size, verbose=1, epochs=300 #epochs\n",
    "                        , validation_data=(x_valid,y_valid), shuffle=True, callbacks=[save_model])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#plot history\n",
    "import matplotlib.pyplot as plt\n",
    "def plot_history(history):\n",
    "    # 精度の履歴をプロット\n",
    "    plt.plot(history.history['acc'],label=\"accuracy\")\n",
    "    plt.plot(history.history['val_acc'],\"o-\",label=\"val_acc\")\n",
    "    plt.title('model accuracy')\n",
    "    plt.xlabel('epoch')\n",
    "    plt.ylabel('accuracy')\n",
    "    plt.legend(loc=\"lower right\")\n",
    "    plt.show()\n",
    "\n",
    "    # 損失の履歴をプロット\n",
    "    plt.plot(history.history['loss'],label=\"loss\",)\n",
    "    plt.plot(history.history['val_loss'],\"o-\",label=\"val_loss\")\n",
    "    plt.title('model loss')\n",
    "    plt.xlabel('epoch')\n",
    "    plt.ylabel('loss')\n",
    "    plt.legend(loc='lower right')\n",
    "    plt.show()\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSty9pwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f7ba5397fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f7ba41347b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_history(history_LLE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "save_model_path = '/home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001/params_adadelta_normal_0.3775_0.8758 .hdf5'\n",
    "model.load_weights(save_model_path)#('./params_milnet_adam_0718_2_fixed_weights.hdf5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25510/25510 [==============================] - 41s 2ms/step\n"
     ]
    }
   ],
   "source": [
    "score=model.evaluate(x_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.3352030573201759 0.8841630733045864\n"
     ]
    }
   ],
   "source": [
    "print(score[0],score[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.55685514]], dtype=float32)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.predict_on_batch(x_test[0:1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/owner/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:24: UserWarning: Update your `Model` call to the Keras 2 API: `Model(inputs=[<tf.Tenso..., outputs=[<tf.Tenso...)`\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmwAAAN3CAYAAACP8wbJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7ef51890f7f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "##\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.gridspec as gridspec\n",
    "\n",
    "def plot_weights(w):\n",
    "    l = len(w)\n",
    "    fig = plt.figure(figsize=(10,16))\n",
    "    gs = gridspec.GridSpec(int(l/2+1),2)\n",
    "    \n",
    "    for i in range(l):\n",
    "        j = int( i/2 )\n",
    "        k = i%2\n",
    "        if k==0:\n",
    "            ax = plt.subplot(gs[j,k])\n",
    "            ax.hist( w[i].flatten() ,bins=100 ,range=(-0.5,0.5) )\n",
    "            #ax.set_title(title[i])\n",
    "        else:\n",
    "            ax = plt.subplot(gs[j,k])\n",
    "            ax.hist( w[i].flatten() ,bins=100 ,range=(-0.5,0.5) )\n",
    "            #ax.set_title(title[i])\n",
    "    plt.show()\n",
    "\n",
    "save_model_path = '/home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/params_adadelta_normal 0.489.hdf5'\n",
    "model_test = Model(input=[x_in], output=[out,eij_normalized])\n",
    "model_test.load_weights(save_model_path)\n",
    "plot_weights(model_test.get_weights())\n",
    "##"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([[ 0.00613822, -0.01595823, -0.01638355, ..., -0.09339832,\n",
       "          0.23449233, -0.0163934 ],\n",
       "        [ 0.00570766, -0.01440853, -0.01028535, ...,  0.06878978,\n",
       "          0.02321846,  0.01221276],\n",
       "        [-0.00955735,  0.07102745,  0.05873981, ..., -0.04907799,\n",
       "         -0.11747277,  0.03790778],\n",
       "        ...,\n",
       "        [-0.01039937, -0.03331398, -0.03901637, ...,  0.07031082,\n",
       "          0.09033212, -0.02839166],\n",
       "        [-0.01678702,  0.02661383, -0.0132541 , ...,  0.0219114 ,\n",
       "         -0.02555005, -0.00124587],\n",
       "        [-0.00696072, -0.00591729, -0.00442338, ...,  0.14777248,\n",
       "         -0.08882821,  0.00289663]], dtype=float32),\n",
       " array([-0.08097516,  0.00817164, -0.05212835,  0.00857434,  0.01582111,\n",
       "         0.00218245, -0.12096858,  0.06406359,  0.17368901, -0.06494537],\n",
       "       dtype=float32)]"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.get_layer('dense_43').get_weights()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "Input (InputLayer)              (None, 40, 15)       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence1 (Lambda (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence2 (Lambda (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence3 (Lambda (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence4 (Lambda (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence5 (Lambda (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence6 (Lambda (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence7 (Lambda (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence8 (Lambda (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence9 (Lambda (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence10 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence11 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence12 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence13 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence14 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence15 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence16 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence17 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence18 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence19 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence20 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence21 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence22 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence23 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence24 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence25 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence26 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence27 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence28 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence29 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence30 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence31 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence32 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence33 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence34 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence35 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence36 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence37 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence38 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence39 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "convert_shape_sentence40 (Lambd (None, 15)           0           Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "Embedding (Embedding)           (None, 15, 300)      18468000    convert_shape_sentence1[0][0]    \n",
      "                                                                 convert_shape_sentence1[0][0]    \n",
      "                                                                 convert_shape_sentence1[0][0]    \n",
      "                                                                 convert_shape_sentence2[0][0]    \n",
      "                                                                 convert_shape_sentence2[0][0]    \n",
      "                                                                 convert_shape_sentence2[0][0]    \n",
      "                                                                 convert_shape_sentence3[0][0]    \n",
      "                                                                 convert_shape_sentence3[0][0]    \n",
      "                                                                 convert_shape_sentence3[0][0]    \n",
      "                                                                 convert_shape_sentence4[0][0]    \n",
      "                                                                 convert_shape_sentence4[0][0]    \n",
      "                                                                 convert_shape_sentence4[0][0]    \n",
      "                                                                 convert_shape_sentence5[0][0]    \n",
      "                                                                 convert_shape_sentence5[0][0]    \n",
      "                                                                 convert_shape_sentence5[0][0]    \n",
      "                                                                 convert_shape_sentence6[0][0]    \n",
      "                                                                 convert_shape_sentence6[0][0]    \n",
      "                                                                 convert_shape_sentence6[0][0]    \n",
      "                                                                 convert_shape_sentence7[0][0]    \n",
      "                                                                 convert_shape_sentence7[0][0]    \n",
      "                                                                 convert_shape_sentence7[0][0]    \n",
      "                                                                 convert_shape_sentence8[0][0]    \n",
      "                                                                 convert_shape_sentence8[0][0]    \n",
      "                                                                 convert_shape_sentence8[0][0]    \n",
      "                                                                 convert_shape_sentence9[0][0]    \n",
      "                                                                 convert_shape_sentence9[0][0]    \n",
      "                                                                 convert_shape_sentence9[0][0]    \n",
      "                                                                 convert_shape_sentence10[0][0]   \n",
      "                                                                 convert_shape_sentence10[0][0]   \n",
      "                                                                 convert_shape_sentence10[0][0]   \n",
      "                                                                 convert_shape_sentence11[0][0]   \n",
      "                                                                 convert_shape_sentence11[0][0]   \n",
      "                                                                 convert_shape_sentence11[0][0]   \n",
      "                                                                 convert_shape_sentence12[0][0]   \n",
      "                                                                 convert_shape_sentence12[0][0]   \n",
      "                                                                 convert_shape_sentence12[0][0]   \n",
      "                                                                 convert_shape_sentence13[0][0]   \n",
      "                                                                 convert_shape_sentence13[0][0]   \n",
      "                                                                 convert_shape_sentence13[0][0]   \n",
      "                                                                 convert_shape_sentence14[0][0]   \n",
      "                                                                 convert_shape_sentence14[0][0]   \n",
      "                                                                 convert_shape_sentence14[0][0]   \n",
      "                                                                 convert_shape_sentence15[0][0]   \n",
      "                                                                 convert_shape_sentence15[0][0]   \n",
      "                                                                 convert_shape_sentence15[0][0]   \n",
      "                                                                 convert_shape_sentence16[0][0]   \n",
      "                                                                 convert_shape_sentence16[0][0]   \n",
      "                                                                 convert_shape_sentence16[0][0]   \n",
      "                                                                 convert_shape_sentence17[0][0]   \n",
      "                                                                 convert_shape_sentence17[0][0]   \n",
      "                                                                 convert_shape_sentence17[0][0]   \n",
      "                                                                 convert_shape_sentence18[0][0]   \n",
      "                                                                 convert_shape_sentence18[0][0]   \n",
      "                                                                 convert_shape_sentence18[0][0]   \n",
      "                                                                 convert_shape_sentence19[0][0]   \n",
      "                                                                 convert_shape_sentence19[0][0]   \n",
      "                                                                 convert_shape_sentence19[0][0]   \n",
      "                                                                 convert_shape_sentence20[0][0]   \n",
      "                                                                 convert_shape_sentence20[0][0]   \n",
      "                                                                 convert_shape_sentence20[0][0]   \n",
      "                                                                 convert_shape_sentence21[0][0]   \n",
      "                                                                 convert_shape_sentence21[0][0]   \n",
      "                                                                 convert_shape_sentence21[0][0]   \n",
      "                                                                 convert_shape_sentence22[0][0]   \n",
      "                                                                 convert_shape_sentence22[0][0]   \n",
      "                                                                 convert_shape_sentence22[0][0]   \n",
      "                                                                 convert_shape_sentence23[0][0]   \n",
      "                                                                 convert_shape_sentence23[0][0]   \n",
      "                                                                 convert_shape_sentence23[0][0]   \n",
      "                                                                 convert_shape_sentence24[0][0]   \n",
      "                                                                 convert_shape_sentence24[0][0]   \n",
      "                                                                 convert_shape_sentence24[0][0]   \n",
      "                                                                 convert_shape_sentence25[0][0]   \n",
      "                                                                 convert_shape_sentence25[0][0]   \n",
      "                                                                 convert_shape_sentence25[0][0]   \n",
      "                                                                 convert_shape_sentence26[0][0]   \n",
      "                                                                 convert_shape_sentence26[0][0]   \n",
      "                                                                 convert_shape_sentence26[0][0]   \n",
      "                                                                 convert_shape_sentence27[0][0]   \n",
      "                                                                 convert_shape_sentence27[0][0]   \n",
      "                                                                 convert_shape_sentence27[0][0]   \n",
      "                                                                 convert_shape_sentence28[0][0]   \n",
      "                                                                 convert_shape_sentence28[0][0]   \n",
      "                                                                 convert_shape_sentence28[0][0]   \n",
      "                                                                 convert_shape_sentence29[0][0]   \n",
      "                                                                 convert_shape_sentence29[0][0]   \n",
      "                                                                 convert_shape_sentence29[0][0]   \n",
      "                                                                 convert_shape_sentence30[0][0]   \n",
      "                                                                 convert_shape_sentence30[0][0]   \n",
      "                                                                 convert_shape_sentence30[0][0]   \n",
      "                                                                 convert_shape_sentence31[0][0]   \n",
      "                                                                 convert_shape_sentence31[0][0]   \n",
      "                                                                 convert_shape_sentence31[0][0]   \n",
      "                                                                 convert_shape_sentence32[0][0]   \n",
      "                                                                 convert_shape_sentence32[0][0]   \n",
      "                                                                 convert_shape_sentence32[0][0]   \n",
      "                                                                 convert_shape_sentence33[0][0]   \n",
      "                                                                 convert_shape_sentence33[0][0]   \n",
      "                                                                 convert_shape_sentence33[0][0]   \n",
      "                                                                 convert_shape_sentence34[0][0]   \n",
      "                                                                 convert_shape_sentence34[0][0]   \n",
      "                                                                 convert_shape_sentence34[0][0]   \n",
      "                                                                 convert_shape_sentence35[0][0]   \n",
      "                                                                 convert_shape_sentence35[0][0]   \n",
      "                                                                 convert_shape_sentence35[0][0]   \n",
      "                                                                 convert_shape_sentence36[0][0]   \n",
      "                                                                 convert_shape_sentence36[0][0]   \n",
      "                                                                 convert_shape_sentence36[0][0]   \n",
      "                                                                 convert_shape_sentence37[0][0]   \n",
      "                                                                 convert_shape_sentence37[0][0]   \n",
      "                                                                 convert_shape_sentence37[0][0]   \n",
      "                                                                 convert_shape_sentence38[0][0]   \n",
      "                                                                 convert_shape_sentence38[0][0]   \n",
      "                                                                 convert_shape_sentence38[0][0]   \n",
      "                                                                 convert_shape_sentence39[0][0]   \n",
      "                                                                 convert_shape_sentence39[0][0]   \n",
      "                                                                 convert_shape_sentence39[0][0]   \n",
      "                                                                 convert_shape_sentence40[0][0]   \n",
      "                                                                 convert_shape_sentence40[0][0]   \n",
      "                                                                 convert_shape_sentence40[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_4 (Conv1D)               (None, 13, 100)      90100       Embedding[0][0]                  \n",
      "                                                                 Embedding[3][0]                  \n",
      "                                                                 Embedding[6][0]                  \n",
      "                                                                 Embedding[9][0]                  \n",
      "                                                                 Embedding[12][0]                 \n",
      "                                                                 Embedding[15][0]                 \n",
      "                                                                 Embedding[18][0]                 \n",
      "                                                                 Embedding[21][0]                 \n",
      "                                                                 Embedding[24][0]                 \n",
      "                                                                 Embedding[27][0]                 \n",
      "                                                                 Embedding[30][0]                 \n",
      "                                                                 Embedding[33][0]                 \n",
      "                                                                 Embedding[36][0]                 \n",
      "                                                                 Embedding[39][0]                 \n",
      "                                                                 Embedding[42][0]                 \n",
      "                                                                 Embedding[45][0]                 \n",
      "                                                                 Embedding[48][0]                 \n",
      "                                                                 Embedding[51][0]                 \n",
      "                                                                 Embedding[54][0]                 \n",
      "                                                                 Embedding[57][0]                 \n",
      "                                                                 Embedding[60][0]                 \n",
      "                                                                 Embedding[63][0]                 \n",
      "                                                                 Embedding[66][0]                 \n",
      "                                                                 Embedding[69][0]                 \n",
      "                                                                 Embedding[72][0]                 \n",
      "                                                                 Embedding[75][0]                 \n",
      "                                                                 Embedding[78][0]                 \n",
      "                                                                 Embedding[81][0]                 \n",
      "                                                                 Embedding[84][0]                 \n",
      "                                                                 Embedding[87][0]                 \n",
      "                                                                 Embedding[90][0]                 \n",
      "                                                                 Embedding[93][0]                 \n",
      "                                                                 Embedding[96][0]                 \n",
      "                                                                 Embedding[99][0]                 \n",
      "                                                                 Embedding[102][0]                \n",
      "                                                                 Embedding[105][0]                \n",
      "                                                                 Embedding[108][0]                \n",
      "                                                                 Embedding[111][0]                \n",
      "                                                                 Embedding[114][0]                \n",
      "                                                                 Embedding[117][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_5 (Conv1D)               (None, 12, 100)      120100      Embedding[1][0]                  \n",
      "                                                                 Embedding[4][0]                  \n",
      "                                                                 Embedding[7][0]                  \n",
      "                                                                 Embedding[10][0]                 \n",
      "                                                                 Embedding[13][0]                 \n",
      "                                                                 Embedding[16][0]                 \n",
      "                                                                 Embedding[19][0]                 \n",
      "                                                                 Embedding[22][0]                 \n",
      "                                                                 Embedding[25][0]                 \n",
      "                                                                 Embedding[28][0]                 \n",
      "                                                                 Embedding[31][0]                 \n",
      "                                                                 Embedding[34][0]                 \n",
      "                                                                 Embedding[37][0]                 \n",
      "                                                                 Embedding[40][0]                 \n",
      "                                                                 Embedding[43][0]                 \n",
      "                                                                 Embedding[46][0]                 \n",
      "                                                                 Embedding[49][0]                 \n",
      "                                                                 Embedding[52][0]                 \n",
      "                                                                 Embedding[55][0]                 \n",
      "                                                                 Embedding[58][0]                 \n",
      "                                                                 Embedding[61][0]                 \n",
      "                                                                 Embedding[64][0]                 \n",
      "                                                                 Embedding[67][0]                 \n",
      "                                                                 Embedding[70][0]                 \n",
      "                                                                 Embedding[73][0]                 \n",
      "                                                                 Embedding[76][0]                 \n",
      "                                                                 Embedding[79][0]                 \n",
      "                                                                 Embedding[82][0]                 \n",
      "                                                                 Embedding[85][0]                 \n",
      "                                                                 Embedding[88][0]                 \n",
      "                                                                 Embedding[91][0]                 \n",
      "                                                                 Embedding[94][0]                 \n",
      "                                                                 Embedding[97][0]                 \n",
      "                                                                 Embedding[100][0]                \n",
      "                                                                 Embedding[103][0]                \n",
      "                                                                 Embedding[106][0]                \n",
      "                                                                 Embedding[109][0]                \n",
      "                                                                 Embedding[112][0]                \n",
      "                                                                 Embedding[115][0]                \n",
      "                                                                 Embedding[118][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_6 (Conv1D)               (None, 11, 100)      150100      Embedding[2][0]                  \n",
      "                                                                 Embedding[5][0]                  \n",
      "                                                                 Embedding[8][0]                  \n",
      "                                                                 Embedding[11][0]                 \n",
      "                                                                 Embedding[14][0]                 \n",
      "                                                                 Embedding[17][0]                 \n",
      "                                                                 Embedding[20][0]                 \n",
      "                                                                 Embedding[23][0]                 \n",
      "                                                                 Embedding[26][0]                 \n",
      "                                                                 Embedding[29][0]                 \n",
      "                                                                 Embedding[32][0]                 \n",
      "                                                                 Embedding[35][0]                 \n",
      "                                                                 Embedding[38][0]                 \n",
      "                                                                 Embedding[41][0]                 \n",
      "                                                                 Embedding[44][0]                 \n",
      "                                                                 Embedding[47][0]                 \n",
      "                                                                 Embedding[50][0]                 \n",
      "                                                                 Embedding[53][0]                 \n",
      "                                                                 Embedding[56][0]                 \n",
      "                                                                 Embedding[59][0]                 \n",
      "                                                                 Embedding[62][0]                 \n",
      "                                                                 Embedding[65][0]                 \n",
      "                                                                 Embedding[68][0]                 \n",
      "                                                                 Embedding[71][0]                 \n",
      "                                                                 Embedding[74][0]                 \n",
      "                                                                 Embedding[77][0]                 \n",
      "                                                                 Embedding[80][0]                 \n",
      "                                                                 Embedding[83][0]                 \n",
      "                                                                 Embedding[86][0]                 \n",
      "                                                                 Embedding[89][0]                 \n",
      "                                                                 Embedding[92][0]                 \n",
      "                                                                 Embedding[95][0]                 \n",
      "                                                                 Embedding[98][0]                 \n",
      "                                                                 Embedding[101][0]                \n",
      "                                                                 Embedding[104][0]                \n",
      "                                                                 Embedding[107][0]                \n",
      "                                                                 Embedding[110][0]                \n",
      "                                                                 Embedding[113][0]                \n",
      "                                                                 Embedding[116][0]                \n",
      "                                                                 Embedding[119][0]                \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_127 (MaxPooling1D (None, 1, 100)       0           conv1d_4[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_128 (MaxPooling1D (None, 1, 100)       0           conv1d_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_129 (MaxPooling1D (None, 1, 100)       0           conv1d_6[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_130 (MaxPooling1D (None, 1, 100)       0           conv1d_4[1][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_131 (MaxPooling1D (None, 1, 100)       0           conv1d_5[1][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_132 (MaxPooling1D (None, 1, 100)       0           conv1d_6[1][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_133 (MaxPooling1D (None, 1, 100)       0           conv1d_4[2][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_134 (MaxPooling1D (None, 1, 100)       0           conv1d_5[2][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_135 (MaxPooling1D (None, 1, 100)       0           conv1d_6[2][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_136 (MaxPooling1D (None, 1, 100)       0           conv1d_4[3][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_137 (MaxPooling1D (None, 1, 100)       0           conv1d_5[3][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_138 (MaxPooling1D (None, 1, 100)       0           conv1d_6[3][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_139 (MaxPooling1D (None, 1, 100)       0           conv1d_4[4][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_140 (MaxPooling1D (None, 1, 100)       0           conv1d_5[4][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_141 (MaxPooling1D (None, 1, 100)       0           conv1d_6[4][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_142 (MaxPooling1D (None, 1, 100)       0           conv1d_4[5][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_143 (MaxPooling1D (None, 1, 100)       0           conv1d_5[5][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_144 (MaxPooling1D (None, 1, 100)       0           conv1d_6[5][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_145 (MaxPooling1D (None, 1, 100)       0           conv1d_4[6][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_146 (MaxPooling1D (None, 1, 100)       0           conv1d_5[6][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_147 (MaxPooling1D (None, 1, 100)       0           conv1d_6[6][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_148 (MaxPooling1D (None, 1, 100)       0           conv1d_4[7][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_149 (MaxPooling1D (None, 1, 100)       0           conv1d_5[7][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_150 (MaxPooling1D (None, 1, 100)       0           conv1d_6[7][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_151 (MaxPooling1D (None, 1, 100)       0           conv1d_4[8][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_152 (MaxPooling1D (None, 1, 100)       0           conv1d_5[8][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_153 (MaxPooling1D (None, 1, 100)       0           conv1d_6[8][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_154 (MaxPooling1D (None, 1, 100)       0           conv1d_4[9][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_155 (MaxPooling1D (None, 1, 100)       0           conv1d_5[9][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_156 (MaxPooling1D (None, 1, 100)       0           conv1d_6[9][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_157 (MaxPooling1D (None, 1, 100)       0           conv1d_4[10][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_158 (MaxPooling1D (None, 1, 100)       0           conv1d_5[10][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_159 (MaxPooling1D (None, 1, 100)       0           conv1d_6[10][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_160 (MaxPooling1D (None, 1, 100)       0           conv1d_4[11][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_161 (MaxPooling1D (None, 1, 100)       0           conv1d_5[11][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_162 (MaxPooling1D (None, 1, 100)       0           conv1d_6[11][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_163 (MaxPooling1D (None, 1, 100)       0           conv1d_4[12][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_164 (MaxPooling1D (None, 1, 100)       0           conv1d_5[12][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_165 (MaxPooling1D (None, 1, 100)       0           conv1d_6[12][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_166 (MaxPooling1D (None, 1, 100)       0           conv1d_4[13][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_167 (MaxPooling1D (None, 1, 100)       0           conv1d_5[13][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_168 (MaxPooling1D (None, 1, 100)       0           conv1d_6[13][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_169 (MaxPooling1D (None, 1, 100)       0           conv1d_4[14][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_170 (MaxPooling1D (None, 1, 100)       0           conv1d_5[14][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_171 (MaxPooling1D (None, 1, 100)       0           conv1d_6[14][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_172 (MaxPooling1D (None, 1, 100)       0           conv1d_4[15][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_173 (MaxPooling1D (None, 1, 100)       0           conv1d_5[15][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_174 (MaxPooling1D (None, 1, 100)       0           conv1d_6[15][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_175 (MaxPooling1D (None, 1, 100)       0           conv1d_4[16][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_176 (MaxPooling1D (None, 1, 100)       0           conv1d_5[16][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_177 (MaxPooling1D (None, 1, 100)       0           conv1d_6[16][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_178 (MaxPooling1D (None, 1, 100)       0           conv1d_4[17][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_179 (MaxPooling1D (None, 1, 100)       0           conv1d_5[17][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_180 (MaxPooling1D (None, 1, 100)       0           conv1d_6[17][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_181 (MaxPooling1D (None, 1, 100)       0           conv1d_4[18][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_182 (MaxPooling1D (None, 1, 100)       0           conv1d_5[18][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_183 (MaxPooling1D (None, 1, 100)       0           conv1d_6[18][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_184 (MaxPooling1D (None, 1, 100)       0           conv1d_4[19][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_185 (MaxPooling1D (None, 1, 100)       0           conv1d_5[19][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_186 (MaxPooling1D (None, 1, 100)       0           conv1d_6[19][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_187 (MaxPooling1D (None, 1, 100)       0           conv1d_4[20][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_188 (MaxPooling1D (None, 1, 100)       0           conv1d_5[20][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_189 (MaxPooling1D (None, 1, 100)       0           conv1d_6[20][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_190 (MaxPooling1D (None, 1, 100)       0           conv1d_4[21][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_191 (MaxPooling1D (None, 1, 100)       0           conv1d_5[21][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_192 (MaxPooling1D (None, 1, 100)       0           conv1d_6[21][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_193 (MaxPooling1D (None, 1, 100)       0           conv1d_4[22][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_194 (MaxPooling1D (None, 1, 100)       0           conv1d_5[22][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_195 (MaxPooling1D (None, 1, 100)       0           conv1d_6[22][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_196 (MaxPooling1D (None, 1, 100)       0           conv1d_4[23][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_197 (MaxPooling1D (None, 1, 100)       0           conv1d_5[23][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_198 (MaxPooling1D (None, 1, 100)       0           conv1d_6[23][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_199 (MaxPooling1D (None, 1, 100)       0           conv1d_4[24][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_200 (MaxPooling1D (None, 1, 100)       0           conv1d_5[24][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_201 (MaxPooling1D (None, 1, 100)       0           conv1d_6[24][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_202 (MaxPooling1D (None, 1, 100)       0           conv1d_4[25][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_203 (MaxPooling1D (None, 1, 100)       0           conv1d_5[25][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_204 (MaxPooling1D (None, 1, 100)       0           conv1d_6[25][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_205 (MaxPooling1D (None, 1, 100)       0           conv1d_4[26][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_206 (MaxPooling1D (None, 1, 100)       0           conv1d_5[26][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_207 (MaxPooling1D (None, 1, 100)       0           conv1d_6[26][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_208 (MaxPooling1D (None, 1, 100)       0           conv1d_4[27][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_209 (MaxPooling1D (None, 1, 100)       0           conv1d_5[27][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_210 (MaxPooling1D (None, 1, 100)       0           conv1d_6[27][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_211 (MaxPooling1D (None, 1, 100)       0           conv1d_4[28][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_212 (MaxPooling1D (None, 1, 100)       0           conv1d_5[28][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_213 (MaxPooling1D (None, 1, 100)       0           conv1d_6[28][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_214 (MaxPooling1D (None, 1, 100)       0           conv1d_4[29][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_215 (MaxPooling1D (None, 1, 100)       0           conv1d_5[29][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_216 (MaxPooling1D (None, 1, 100)       0           conv1d_6[29][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_217 (MaxPooling1D (None, 1, 100)       0           conv1d_4[30][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_218 (MaxPooling1D (None, 1, 100)       0           conv1d_5[30][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_219 (MaxPooling1D (None, 1, 100)       0           conv1d_6[30][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_220 (MaxPooling1D (None, 1, 100)       0           conv1d_4[31][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_221 (MaxPooling1D (None, 1, 100)       0           conv1d_5[31][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_222 (MaxPooling1D (None, 1, 100)       0           conv1d_6[31][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_223 (MaxPooling1D (None, 1, 100)       0           conv1d_4[32][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_224 (MaxPooling1D (None, 1, 100)       0           conv1d_5[32][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_225 (MaxPooling1D (None, 1, 100)       0           conv1d_6[32][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_226 (MaxPooling1D (None, 1, 100)       0           conv1d_4[33][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_227 (MaxPooling1D (None, 1, 100)       0           conv1d_5[33][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_228 (MaxPooling1D (None, 1, 100)       0           conv1d_6[33][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_229 (MaxPooling1D (None, 1, 100)       0           conv1d_4[34][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_230 (MaxPooling1D (None, 1, 100)       0           conv1d_5[34][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_231 (MaxPooling1D (None, 1, 100)       0           conv1d_6[34][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_232 (MaxPooling1D (None, 1, 100)       0           conv1d_4[35][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_233 (MaxPooling1D (None, 1, 100)       0           conv1d_5[35][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_234 (MaxPooling1D (None, 1, 100)       0           conv1d_6[35][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_235 (MaxPooling1D (None, 1, 100)       0           conv1d_4[36][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_236 (MaxPooling1D (None, 1, 100)       0           conv1d_5[36][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_237 (MaxPooling1D (None, 1, 100)       0           conv1d_6[36][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_238 (MaxPooling1D (None, 1, 100)       0           conv1d_4[37][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_239 (MaxPooling1D (None, 1, 100)       0           conv1d_5[37][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_240 (MaxPooling1D (None, 1, 100)       0           conv1d_6[37][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_241 (MaxPooling1D (None, 1, 100)       0           conv1d_4[38][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_242 (MaxPooling1D (None, 1, 100)       0           conv1d_5[38][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_243 (MaxPooling1D (None, 1, 100)       0           conv1d_6[38][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_244 (MaxPooling1D (None, 1, 100)       0           conv1d_4[39][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_245 (MaxPooling1D (None, 1, 100)       0           conv1d_5[39][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_246 (MaxPooling1D (None, 1, 100)       0           conv1d_6[39][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_121 (Lambda)             (None, 100)          0           max_pooling1d_127[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_122 (Lambda)             (None, 100)          0           max_pooling1d_128[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_123 (Lambda)             (None, 100)          0           max_pooling1d_129[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_124 (Lambda)             (None, 100)          0           max_pooling1d_130[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_125 (Lambda)             (None, 100)          0           max_pooling1d_131[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_126 (Lambda)             (None, 100)          0           max_pooling1d_132[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_127 (Lambda)             (None, 100)          0           max_pooling1d_133[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_128 (Lambda)             (None, 100)          0           max_pooling1d_134[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_129 (Lambda)             (None, 100)          0           max_pooling1d_135[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_130 (Lambda)             (None, 100)          0           max_pooling1d_136[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_131 (Lambda)             (None, 100)          0           max_pooling1d_137[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_132 (Lambda)             (None, 100)          0           max_pooling1d_138[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_133 (Lambda)             (None, 100)          0           max_pooling1d_139[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_134 (Lambda)             (None, 100)          0           max_pooling1d_140[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_135 (Lambda)             (None, 100)          0           max_pooling1d_141[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_136 (Lambda)             (None, 100)          0           max_pooling1d_142[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_137 (Lambda)             (None, 100)          0           max_pooling1d_143[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_138 (Lambda)             (None, 100)          0           max_pooling1d_144[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_139 (Lambda)             (None, 100)          0           max_pooling1d_145[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_140 (Lambda)             (None, 100)          0           max_pooling1d_146[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_141 (Lambda)             (None, 100)          0           max_pooling1d_147[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_142 (Lambda)             (None, 100)          0           max_pooling1d_148[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_143 (Lambda)             (None, 100)          0           max_pooling1d_149[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_144 (Lambda)             (None, 100)          0           max_pooling1d_150[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_145 (Lambda)             (None, 100)          0           max_pooling1d_151[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_146 (Lambda)             (None, 100)          0           max_pooling1d_152[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_147 (Lambda)             (None, 100)          0           max_pooling1d_153[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_148 (Lambda)             (None, 100)          0           max_pooling1d_154[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_149 (Lambda)             (None, 100)          0           max_pooling1d_155[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_150 (Lambda)             (None, 100)          0           max_pooling1d_156[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_151 (Lambda)             (None, 100)          0           max_pooling1d_157[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_152 (Lambda)             (None, 100)          0           max_pooling1d_158[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_153 (Lambda)             (None, 100)          0           max_pooling1d_159[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_154 (Lambda)             (None, 100)          0           max_pooling1d_160[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_155 (Lambda)             (None, 100)          0           max_pooling1d_161[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_156 (Lambda)             (None, 100)          0           max_pooling1d_162[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_157 (Lambda)             (None, 100)          0           max_pooling1d_163[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_158 (Lambda)             (None, 100)          0           max_pooling1d_164[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_159 (Lambda)             (None, 100)          0           max_pooling1d_165[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_160 (Lambda)             (None, 100)          0           max_pooling1d_166[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_161 (Lambda)             (None, 100)          0           max_pooling1d_167[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_162 (Lambda)             (None, 100)          0           max_pooling1d_168[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_163 (Lambda)             (None, 100)          0           max_pooling1d_169[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_164 (Lambda)             (None, 100)          0           max_pooling1d_170[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_165 (Lambda)             (None, 100)          0           max_pooling1d_171[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_166 (Lambda)             (None, 100)          0           max_pooling1d_172[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_167 (Lambda)             (None, 100)          0           max_pooling1d_173[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_168 (Lambda)             (None, 100)          0           max_pooling1d_174[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_169 (Lambda)             (None, 100)          0           max_pooling1d_175[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_170 (Lambda)             (None, 100)          0           max_pooling1d_176[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_171 (Lambda)             (None, 100)          0           max_pooling1d_177[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_172 (Lambda)             (None, 100)          0           max_pooling1d_178[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_173 (Lambda)             (None, 100)          0           max_pooling1d_179[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_174 (Lambda)             (None, 100)          0           max_pooling1d_180[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_175 (Lambda)             (None, 100)          0           max_pooling1d_181[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_176 (Lambda)             (None, 100)          0           max_pooling1d_182[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_177 (Lambda)             (None, 100)          0           max_pooling1d_183[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_178 (Lambda)             (None, 100)          0           max_pooling1d_184[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_179 (Lambda)             (None, 100)          0           max_pooling1d_185[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_180 (Lambda)             (None, 100)          0           max_pooling1d_186[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_181 (Lambda)             (None, 100)          0           max_pooling1d_187[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_182 (Lambda)             (None, 100)          0           max_pooling1d_188[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_183 (Lambda)             (None, 100)          0           max_pooling1d_189[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_184 (Lambda)             (None, 100)          0           max_pooling1d_190[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_185 (Lambda)             (None, 100)          0           max_pooling1d_191[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_186 (Lambda)             (None, 100)          0           max_pooling1d_192[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_187 (Lambda)             (None, 100)          0           max_pooling1d_193[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_188 (Lambda)             (None, 100)          0           max_pooling1d_194[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_189 (Lambda)             (None, 100)          0           max_pooling1d_195[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_190 (Lambda)             (None, 100)          0           max_pooling1d_196[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_191 (Lambda)             (None, 100)          0           max_pooling1d_197[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_192 (Lambda)             (None, 100)          0           max_pooling1d_198[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_193 (Lambda)             (None, 100)          0           max_pooling1d_199[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_194 (Lambda)             (None, 100)          0           max_pooling1d_200[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_195 (Lambda)             (None, 100)          0           max_pooling1d_201[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_196 (Lambda)             (None, 100)          0           max_pooling1d_202[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_197 (Lambda)             (None, 100)          0           max_pooling1d_203[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_198 (Lambda)             (None, 100)          0           max_pooling1d_204[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_199 (Lambda)             (None, 100)          0           max_pooling1d_205[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_200 (Lambda)             (None, 100)          0           max_pooling1d_206[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_201 (Lambda)             (None, 100)          0           max_pooling1d_207[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_202 (Lambda)             (None, 100)          0           max_pooling1d_208[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_203 (Lambda)             (None, 100)          0           max_pooling1d_209[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_204 (Lambda)             (None, 100)          0           max_pooling1d_210[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_205 (Lambda)             (None, 100)          0           max_pooling1d_211[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_206 (Lambda)             (None, 100)          0           max_pooling1d_212[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_207 (Lambda)             (None, 100)          0           max_pooling1d_213[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_208 (Lambda)             (None, 100)          0           max_pooling1d_214[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_209 (Lambda)             (None, 100)          0           max_pooling1d_215[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_210 (Lambda)             (None, 100)          0           max_pooling1d_216[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_211 (Lambda)             (None, 100)          0           max_pooling1d_217[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_212 (Lambda)             (None, 100)          0           max_pooling1d_218[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_213 (Lambda)             (None, 100)          0           max_pooling1d_219[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_214 (Lambda)             (None, 100)          0           max_pooling1d_220[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_215 (Lambda)             (None, 100)          0           max_pooling1d_221[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_216 (Lambda)             (None, 100)          0           max_pooling1d_222[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_217 (Lambda)             (None, 100)          0           max_pooling1d_223[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_218 (Lambda)             (None, 100)          0           max_pooling1d_224[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_219 (Lambda)             (None, 100)          0           max_pooling1d_225[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_220 (Lambda)             (None, 100)          0           max_pooling1d_226[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_221 (Lambda)             (None, 100)          0           max_pooling1d_227[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_222 (Lambda)             (None, 100)          0           max_pooling1d_228[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_223 (Lambda)             (None, 100)          0           max_pooling1d_229[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_224 (Lambda)             (None, 100)          0           max_pooling1d_230[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_225 (Lambda)             (None, 100)          0           max_pooling1d_231[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_226 (Lambda)             (None, 100)          0           max_pooling1d_232[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_227 (Lambda)             (None, 100)          0           max_pooling1d_233[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_228 (Lambda)             (None, 100)          0           max_pooling1d_234[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_229 (Lambda)             (None, 100)          0           max_pooling1d_235[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_230 (Lambda)             (None, 100)          0           max_pooling1d_236[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_231 (Lambda)             (None, 100)          0           max_pooling1d_237[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_232 (Lambda)             (None, 100)          0           max_pooling1d_238[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_233 (Lambda)             (None, 100)          0           max_pooling1d_239[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_234 (Lambda)             (None, 100)          0           max_pooling1d_240[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_235 (Lambda)             (None, 100)          0           max_pooling1d_241[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_236 (Lambda)             (None, 100)          0           max_pooling1d_242[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_237 (Lambda)             (None, 100)          0           max_pooling1d_243[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_238 (Lambda)             (None, 100)          0           max_pooling1d_244[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_239 (Lambda)             (None, 100)          0           max_pooling1d_245[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "lambda_240 (Lambda)             (None, 100)          0           max_pooling1d_246[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_42 (Concatenate)    (None, 300)          0           lambda_121[0][0]                 \n",
      "                                                                 lambda_122[0][0]                 \n",
      "                                                                 lambda_123[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_43 (Concatenate)    (None, 300)          0           lambda_124[0][0]                 \n",
      "                                                                 lambda_125[0][0]                 \n",
      "                                                                 lambda_126[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_44 (Concatenate)    (None, 300)          0           lambda_127[0][0]                 \n",
      "                                                                 lambda_128[0][0]                 \n",
      "                                                                 lambda_129[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_45 (Concatenate)    (None, 300)          0           lambda_130[0][0]                 \n",
      "                                                                 lambda_131[0][0]                 \n",
      "                                                                 lambda_132[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_46 (Concatenate)    (None, 300)          0           lambda_133[0][0]                 \n",
      "                                                                 lambda_134[0][0]                 \n",
      "                                                                 lambda_135[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_47 (Concatenate)    (None, 300)          0           lambda_136[0][0]                 \n",
      "                                                                 lambda_137[0][0]                 \n",
      "                                                                 lambda_138[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_48 (Concatenate)    (None, 300)          0           lambda_139[0][0]                 \n",
      "                                                                 lambda_140[0][0]                 \n",
      "                                                                 lambda_141[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_49 (Concatenate)    (None, 300)          0           lambda_142[0][0]                 \n",
      "                                                                 lambda_143[0][0]                 \n",
      "                                                                 lambda_144[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_50 (Concatenate)    (None, 300)          0           lambda_145[0][0]                 \n",
      "                                                                 lambda_146[0][0]                 \n",
      "                                                                 lambda_147[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_51 (Concatenate)    (None, 300)          0           lambda_148[0][0]                 \n",
      "                                                                 lambda_149[0][0]                 \n",
      "                                                                 lambda_150[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_52 (Concatenate)    (None, 300)          0           lambda_151[0][0]                 \n",
      "                                                                 lambda_152[0][0]                 \n",
      "                                                                 lambda_153[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_53 (Concatenate)    (None, 300)          0           lambda_154[0][0]                 \n",
      "                                                                 lambda_155[0][0]                 \n",
      "                                                                 lambda_156[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_54 (Concatenate)    (None, 300)          0           lambda_157[0][0]                 \n",
      "                                                                 lambda_158[0][0]                 \n",
      "                                                                 lambda_159[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_55 (Concatenate)    (None, 300)          0           lambda_160[0][0]                 \n",
      "                                                                 lambda_161[0][0]                 \n",
      "                                                                 lambda_162[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_56 (Concatenate)    (None, 300)          0           lambda_163[0][0]                 \n",
      "                                                                 lambda_164[0][0]                 \n",
      "                                                                 lambda_165[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_57 (Concatenate)    (None, 300)          0           lambda_166[0][0]                 \n",
      "                                                                 lambda_167[0][0]                 \n",
      "                                                                 lambda_168[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_58 (Concatenate)    (None, 300)          0           lambda_169[0][0]                 \n",
      "                                                                 lambda_170[0][0]                 \n",
      "                                                                 lambda_171[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_59 (Concatenate)    (None, 300)          0           lambda_172[0][0]                 \n",
      "                                                                 lambda_173[0][0]                 \n",
      "                                                                 lambda_174[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_60 (Concatenate)    (None, 300)          0           lambda_175[0][0]                 \n",
      "                                                                 lambda_176[0][0]                 \n",
      "                                                                 lambda_177[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_61 (Concatenate)    (None, 300)          0           lambda_178[0][0]                 \n",
      "                                                                 lambda_179[0][0]                 \n",
      "                                                                 lambda_180[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_62 (Concatenate)    (None, 300)          0           lambda_181[0][0]                 \n",
      "                                                                 lambda_182[0][0]                 \n",
      "                                                                 lambda_183[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_63 (Concatenate)    (None, 300)          0           lambda_184[0][0]                 \n",
      "                                                                 lambda_185[0][0]                 \n",
      "                                                                 lambda_186[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_64 (Concatenate)    (None, 300)          0           lambda_187[0][0]                 \n",
      "                                                                 lambda_188[0][0]                 \n",
      "                                                                 lambda_189[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_65 (Concatenate)    (None, 300)          0           lambda_190[0][0]                 \n",
      "                                                                 lambda_191[0][0]                 \n",
      "                                                                 lambda_192[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_66 (Concatenate)    (None, 300)          0           lambda_193[0][0]                 \n",
      "                                                                 lambda_194[0][0]                 \n",
      "                                                                 lambda_195[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_67 (Concatenate)    (None, 300)          0           lambda_196[0][0]                 \n",
      "                                                                 lambda_197[0][0]                 \n",
      "                                                                 lambda_198[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_68 (Concatenate)    (None, 300)          0           lambda_199[0][0]                 \n",
      "                                                                 lambda_200[0][0]                 \n",
      "                                                                 lambda_201[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_69 (Concatenate)    (None, 300)          0           lambda_202[0][0]                 \n",
      "                                                                 lambda_203[0][0]                 \n",
      "                                                                 lambda_204[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_70 (Concatenate)    (None, 300)          0           lambda_205[0][0]                 \n",
      "                                                                 lambda_206[0][0]                 \n",
      "                                                                 lambda_207[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_71 (Concatenate)    (None, 300)          0           lambda_208[0][0]                 \n",
      "                                                                 lambda_209[0][0]                 \n",
      "                                                                 lambda_210[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_72 (Concatenate)    (None, 300)          0           lambda_211[0][0]                 \n",
      "                                                                 lambda_212[0][0]                 \n",
      "                                                                 lambda_213[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_73 (Concatenate)    (None, 300)          0           lambda_214[0][0]                 \n",
      "                                                                 lambda_215[0][0]                 \n",
      "                                                                 lambda_216[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_74 (Concatenate)    (None, 300)          0           lambda_217[0][0]                 \n",
      "                                                                 lambda_218[0][0]                 \n",
      "                                                                 lambda_219[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_75 (Concatenate)    (None, 300)          0           lambda_220[0][0]                 \n",
      "                                                                 lambda_221[0][0]                 \n",
      "                                                                 lambda_222[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_76 (Concatenate)    (None, 300)          0           lambda_223[0][0]                 \n",
      "                                                                 lambda_224[0][0]                 \n",
      "                                                                 lambda_225[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_77 (Concatenate)    (None, 300)          0           lambda_226[0][0]                 \n",
      "                                                                 lambda_227[0][0]                 \n",
      "                                                                 lambda_228[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_78 (Concatenate)    (None, 300)          0           lambda_229[0][0]                 \n",
      "                                                                 lambda_230[0][0]                 \n",
      "                                                                 lambda_231[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_79 (Concatenate)    (None, 300)          0           lambda_232[0][0]                 \n",
      "                                                                 lambda_233[0][0]                 \n",
      "                                                                 lambda_234[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_80 (Concatenate)    (None, 300)          0           lambda_235[0][0]                 \n",
      "                                                                 lambda_236[0][0]                 \n",
      "                                                                 lambda_237[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_81 (Concatenate)    (None, 300)          0           lambda_238[0][0]                 \n",
      "                                                                 lambda_239[0][0]                 \n",
      "                                                                 lambda_240[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence1winSize5 ( (None, 1, 300)       0           concatenate_42[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence2winSize5 ( (None, 1, 300)       0           concatenate_43[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence3winSize5 ( (None, 1, 300)       0           concatenate_44[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence4winSize5 ( (None, 1, 300)       0           concatenate_45[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence5winSize5 ( (None, 1, 300)       0           concatenate_46[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence6winSize5 ( (None, 1, 300)       0           concatenate_47[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence7winSize5 ( (None, 1, 300)       0           concatenate_48[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence8winSize5 ( (None, 1, 300)       0           concatenate_49[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence9winSize5 ( (None, 1, 300)       0           concatenate_50[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence10winSize5  (None, 1, 300)       0           concatenate_51[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence11winSize5  (None, 1, 300)       0           concatenate_52[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence12winSize5  (None, 1, 300)       0           concatenate_53[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence13winSize5  (None, 1, 300)       0           concatenate_54[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence14winSize5  (None, 1, 300)       0           concatenate_55[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence15winSize5  (None, 1, 300)       0           concatenate_56[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence16winSize5  (None, 1, 300)       0           concatenate_57[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence17winSize5  (None, 1, 300)       0           concatenate_58[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence18winSize5  (None, 1, 300)       0           concatenate_59[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence19winSize5  (None, 1, 300)       0           concatenate_60[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence20winSize5  (None, 1, 300)       0           concatenate_61[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence21winSize5  (None, 1, 300)       0           concatenate_62[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence22winSize5  (None, 1, 300)       0           concatenate_63[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence23winSize5  (None, 1, 300)       0           concatenate_64[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence24winSize5  (None, 1, 300)       0           concatenate_65[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence25winSize5  (None, 1, 300)       0           concatenate_66[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence26winSize5  (None, 1, 300)       0           concatenate_67[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence27winSize5  (None, 1, 300)       0           concatenate_68[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence28winSize5  (None, 1, 300)       0           concatenate_69[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence29winSize5  (None, 1, 300)       0           concatenate_70[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence30winSize5  (None, 1, 300)       0           concatenate_71[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence31winSize5  (None, 1, 300)       0           concatenate_72[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence32winSize5  (None, 1, 300)       0           concatenate_73[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence33winSize5  (None, 1, 300)       0           concatenate_74[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence34winSize5  (None, 1, 300)       0           concatenate_75[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence35winSize5  (None, 1, 300)       0           concatenate_76[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence36winSize5  (None, 1, 300)       0           concatenate_77[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence37winSize5  (None, 1, 300)       0           concatenate_78[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence38winSize5  (None, 1, 300)       0           concatenate_79[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence39winSize5  (None, 1, 300)       0           concatenate_80[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "switch_axis_sentence40winSize5  (None, 1, 300)       0           concatenate_81[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dense_43 (Dense)                (None, 1, 10)        3010        switch_axis_sentence1winSize5[0][\n",
      "__________________________________________________________________________________________________\n",
      "dense_44 (Dense)                (None, 1, 10)        3010        switch_axis_sentence2winSize5[0][\n",
      "__________________________________________________________________________________________________\n",
      "dense_45 (Dense)                (None, 1, 10)        3010        switch_axis_sentence3winSize5[0][\n",
      "__________________________________________________________________________________________________\n",
      "dense_46 (Dense)                (None, 1, 10)        3010        switch_axis_sentence4winSize5[0][\n",
      "__________________________________________________________________________________________________\n",
      "dense_47 (Dense)                (None, 1, 10)        3010        switch_axis_sentence5winSize5[0][\n",
      "__________________________________________________________________________________________________\n",
      "dense_48 (Dense)                (None, 1, 10)        3010        switch_axis_sentence6winSize5[0][\n",
      "__________________________________________________________________________________________________\n",
      "dense_49 (Dense)                (None, 1, 10)        3010        switch_axis_sentence7winSize5[0][\n",
      "__________________________________________________________________________________________________\n",
      "dense_50 (Dense)                (None, 1, 10)        3010        switch_axis_sentence8winSize5[0][\n",
      "__________________________________________________________________________________________________\n",
      "dense_51 (Dense)                (None, 1, 10)        3010        switch_axis_sentence9winSize5[0][\n",
      "__________________________________________________________________________________________________\n",
      "dense_52 (Dense)                (None, 1, 10)        3010        switch_axis_sentence10winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_53 (Dense)                (None, 1, 10)        3010        switch_axis_sentence11winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_54 (Dense)                (None, 1, 10)        3010        switch_axis_sentence12winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_55 (Dense)                (None, 1, 10)        3010        switch_axis_sentence13winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_56 (Dense)                (None, 1, 10)        3010        switch_axis_sentence14winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_57 (Dense)                (None, 1, 10)        3010        switch_axis_sentence15winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_58 (Dense)                (None, 1, 10)        3010        switch_axis_sentence16winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_59 (Dense)                (None, 1, 10)        3010        switch_axis_sentence17winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_60 (Dense)                (None, 1, 10)        3010        switch_axis_sentence18winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_61 (Dense)                (None, 1, 10)        3010        switch_axis_sentence19winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_62 (Dense)                (None, 1, 10)        3010        switch_axis_sentence20winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_63 (Dense)                (None, 1, 10)        3010        switch_axis_sentence21winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_64 (Dense)                (None, 1, 10)        3010        switch_axis_sentence22winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_65 (Dense)                (None, 1, 10)        3010        switch_axis_sentence23winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_66 (Dense)                (None, 1, 10)        3010        switch_axis_sentence24winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_67 (Dense)                (None, 1, 10)        3010        switch_axis_sentence25winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_68 (Dense)                (None, 1, 10)        3010        switch_axis_sentence26winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_69 (Dense)                (None, 1, 10)        3010        switch_axis_sentence27winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_70 (Dense)                (None, 1, 10)        3010        switch_axis_sentence28winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_71 (Dense)                (None, 1, 10)        3010        switch_axis_sentence29winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_72 (Dense)                (None, 1, 10)        3010        switch_axis_sentence30winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_73 (Dense)                (None, 1, 10)        3010        switch_axis_sentence31winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_74 (Dense)                (None, 1, 10)        3010        switch_axis_sentence32winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_75 (Dense)                (None, 1, 10)        3010        switch_axis_sentence33winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_76 (Dense)                (None, 1, 10)        3010        switch_axis_sentence34winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_77 (Dense)                (None, 1, 10)        3010        switch_axis_sentence35winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_78 (Dense)                (None, 1, 10)        3010        switch_axis_sentence36winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_79 (Dense)                (None, 1, 10)        3010        switch_axis_sentence37winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_80 (Dense)                (None, 1, 10)        3010        switch_axis_sentence38winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_81 (Dense)                (None, 1, 10)        3010        switch_axis_sentence39winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "dense_82 (Dense)                (None, 1, 10)        3010        switch_axis_sentence40winSize5[0]\n",
      "__________________________________________________________________________________________________\n",
      "DropDense0 (Dropout)            (None, 1, 10)        0           dense_43[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense1 (Dropout)            (None, 1, 10)        0           dense_44[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense2 (Dropout)            (None, 1, 10)        0           dense_45[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense3 (Dropout)            (None, 1, 10)        0           dense_46[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense4 (Dropout)            (None, 1, 10)        0           dense_47[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense5 (Dropout)            (None, 1, 10)        0           dense_48[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense6 (Dropout)            (None, 1, 10)        0           dense_49[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense7 (Dropout)            (None, 1, 10)        0           dense_50[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense8 (Dropout)            (None, 1, 10)        0           dense_51[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense9 (Dropout)            (None, 1, 10)        0           dense_52[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense10 (Dropout)           (None, 1, 10)        0           dense_53[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense11 (Dropout)           (None, 1, 10)        0           dense_54[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense12 (Dropout)           (None, 1, 10)        0           dense_55[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense13 (Dropout)           (None, 1, 10)        0           dense_56[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense14 (Dropout)           (None, 1, 10)        0           dense_57[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense15 (Dropout)           (None, 1, 10)        0           dense_58[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense16 (Dropout)           (None, 1, 10)        0           dense_59[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense17 (Dropout)           (None, 1, 10)        0           dense_60[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense18 (Dropout)           (None, 1, 10)        0           dense_61[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense19 (Dropout)           (None, 1, 10)        0           dense_62[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense20 (Dropout)           (None, 1, 10)        0           dense_63[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense21 (Dropout)           (None, 1, 10)        0           dense_64[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense22 (Dropout)           (None, 1, 10)        0           dense_65[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense23 (Dropout)           (None, 1, 10)        0           dense_66[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense24 (Dropout)           (None, 1, 10)        0           dense_67[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense25 (Dropout)           (None, 1, 10)        0           dense_68[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense26 (Dropout)           (None, 1, 10)        0           dense_69[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense27 (Dropout)           (None, 1, 10)        0           dense_70[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense28 (Dropout)           (None, 1, 10)        0           dense_71[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense29 (Dropout)           (None, 1, 10)        0           dense_72[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense30 (Dropout)           (None, 1, 10)        0           dense_73[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense31 (Dropout)           (None, 1, 10)        0           dense_74[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense32 (Dropout)           (None, 1, 10)        0           dense_75[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense33 (Dropout)           (None, 1, 10)        0           dense_76[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense34 (Dropout)           (None, 1, 10)        0           dense_77[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense35 (Dropout)           (None, 1, 10)        0           dense_78[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense36 (Dropout)           (None, 1, 10)        0           dense_79[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense37 (Dropout)           (None, 1, 10)        0           dense_80[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense38 (Dropout)           (None, 1, 10)        0           dense_81[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "DropDense39 (Dropout)           (None, 1, 10)        0           dense_82[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_82 (Concatenate)    (None, 40, 10)       0           DropDense0[0][0]                 \n",
      "                                                                 DropDense1[0][0]                 \n",
      "                                                                 DropDense2[0][0]                 \n",
      "                                                                 DropDense3[0][0]                 \n",
      "                                                                 DropDense4[0][0]                 \n",
      "                                                                 DropDense5[0][0]                 \n",
      "                                                                 DropDense6[0][0]                 \n",
      "                                                                 DropDense7[0][0]                 \n",
      "                                                                 DropDense8[0][0]                 \n",
      "                                                                 DropDense9[0][0]                 \n",
      "                                                                 DropDense10[0][0]                \n",
      "                                                                 DropDense11[0][0]                \n",
      "                                                                 DropDense12[0][0]                \n",
      "                                                                 DropDense13[0][0]                \n",
      "                                                                 DropDense14[0][0]                \n",
      "                                                                 DropDense15[0][0]                \n",
      "                                                                 DropDense16[0][0]                \n",
      "                                                                 DropDense17[0][0]                \n",
      "                                                                 DropDense18[0][0]                \n",
      "                                                                 DropDense19[0][0]                \n",
      "                                                                 DropDense20[0][0]                \n",
      "                                                                 DropDense21[0][0]                \n",
      "                                                                 DropDense22[0][0]                \n",
      "                                                                 DropDense23[0][0]                \n",
      "                                                                 DropDense24[0][0]                \n",
      "                                                                 DropDense25[0][0]                \n",
      "                                                                 DropDense26[0][0]                \n",
      "                                                                 DropDense27[0][0]                \n",
      "                                                                 DropDense28[0][0]                \n",
      "                                                                 DropDense29[0][0]                \n",
      "                                                                 DropDense30[0][0]                \n",
      "                                                                 DropDense31[0][0]                \n",
      "                                                                 DropDense32[0][0]                \n",
      "                                                                 DropDense33[0][0]                \n",
      "                                                                 DropDense34[0][0]                \n",
      "                                                                 DropDense35[0][0]                \n",
      "                                                                 DropDense36[0][0]                \n",
      "                                                                 DropDense37[0][0]                \n",
      "                                                                 DropDense38[0][0]                \n",
      "                                                                 DropDense39[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "bidirectional_2 (Bidirectional) (None, 40, 100)      18300       concatenate_82[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "biRnn_TF_Reminder1 (Lambda)     (None, 40, 100)      0           bidirectional_2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "dense_84 (Dense)                (None, 40, 100)      10100       biRnn_TF_Reminder1[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "attention_weights (Dense)       (None, 40, 1)        100         dense_84[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "permute_3 (Permute)             (None, 1, 40)        0           attention_weights[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "attention_softmax (Activation)  (None, 1, 40)        0           permute_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "permute_4 (Permute)             (None, 40, 1)        0           attention_softmax[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "attention_mult (Dot)            (None, 10, 1)        0           concatenate_82[0][0]             \n",
      "                                                                 permute_4[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "squeezeOutput (Lambda)          (None, 10)           0           attention_mult[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dense_85 (Dense)                (None, 1)            11          squeezeOutput[0][0]              \n",
      "==================================================================================================\n",
      "Total params: 18,977,211\n",
      "Trainable params: 509,211\n",
      "Non-trainable params: 18,468,000\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 210642 samples, validate on 25206 samples\n",
      "Epoch 11/15\n",
      "210642/210642 [==============================] - 358s 2ms/step - loss: 0.3641 - acc: 0.8926 - val_loss: 0.3600 - val_acc: 0.8914\n",
      "\n",
      "Epoch 00011: val_loss improved from 0.36792 to 0.35997, saving model to /home/owner/デスクトップ/milnet+edu/results/0724 (15,40)/lr = 0.001 with 10 units/params_adam_normal.hdf5\n",
      "Epoch 12/15\n",
      "210642/210642 [==============================] - 357s 2ms/step - loss: 0.3576 - acc: 0.8961 - val_loss: 0.3766 - val_acc: 0.8849\n",
      "\n",
      "Epoch 00012: val_loss did not improve\n",
      "Epoch 13/15\n",
      "210642/210642 [==============================] - 358s 2ms/step - loss: 0.3558 - acc: 0.8977 - val_loss: 0.3654 - val_acc: 0.8908\n",
      "\n",
      "Epoch 00013: val_loss did not improve\n",
      "Epoch 14/15\n",
      "210642/210642 [==============================] - 357s 2ms/step - loss: 0.3534 - acc: 0.9001 - val_loss: 0.3656 - val_acc: 0.8932\n",
      "\n",
      "Epoch 00014: val_loss did not improve\n",
      "Epoch 15/15\n",
      "210642/210642 [==============================] - 358s 2ms/step - loss: 0.3519 - acc: 0.9016 - val_loss: 0.4007 - val_acc: 0.8752\n",
      "\n",
      "Epoch 00015: val_loss did not improve\n"
     ]
    }
   ],
   "source": [
    "history_normal2 = model.fit(x_train, y_train, batch_size = batch_size, verbose=1, initial_epoch=10,epochs=15 #epochs\n",
    "                        ,validation_data=(x_valid,y_valid), shuffle=True, callbacks=[save_model])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}