chatbot_notebook.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import os\n",
    "import yaml\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.style.use('fivethirtyeight')\n",
    "plt.rcParams['figure.figsize'] = [12, 6]\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "%config InlineBackend.figure_format = 'retina'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "filenames = (os.listdir(\"data/\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.DataFrame(columns={\"question\",\"answer\",\"category\"})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "for file in filenames:\n",
    "    \n",
    "    data_curr = pd.DataFrame(columns={\"question\",\"answer\",\"category\"})\n",
    "    \n",
    "    file = open(\"data/\"+file)\n",
    "    file = yaml.load(file)\n",
    "    category = file[\"categories\"]\n",
    "    conversation = file[\"conversations\"]\n",
    "    \n",
    "    question = []\n",
    "    answer = []\n",
    "\n",
    "    for i in range(len(conversation)):\n",
    "        st = str(conversation[i]).split(\",\")\n",
    "        ques = st[0]\n",
    "        ans = st[1]\n",
    "        question.append(ques)\n",
    "        answer.append(ans)\n",
    "    \n",
    "    data_curr[\"question\"]=question\n",
    "    data_curr[\"answer\"]=answer\n",
    "    data_curr[\"category\"]=str(category[0])\n",
    "    data_curr = data_curr[['question','answer','category']]\n",
    "    \n",
    "    data = data.append(data_curr,ignore_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data[['question','answer','category']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(584, 3)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>question</th>\n",
       "      <th>answer</th>\n",
       "      <th>category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>['What is AI?'</td>\n",
       "      <td>'Artificial Intelligence is the branch of eng...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>['What is AI?'</td>\n",
       "      <td>'AI is the field of science which concerns it...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>['Are you sentient?'</td>\n",
       "      <td>'Sort of.']</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>['Are you sentient?'</td>\n",
       "      <td>\"By the strictest dictionary definition of th...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>['Are you sentient?'</td>\n",
       "      <td>\"Even though I'm a construct I do have a subj...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               question                                             answer  \\\n",
       "0        ['What is AI?'   'Artificial Intelligence is the branch of eng...   \n",
       "1        ['What is AI?'   'AI is the field of science which concerns it...   \n",
       "2  ['Are you sentient?'                                        'Sort of.']   \n",
       "3  ['Are you sentient?'   \"By the strictest dictionary definition of th...   \n",
       "4  ['Are you sentient?'   \"Even though I'm a construct I do have a subj...   \n",
       "\n",
       "  category  \n",
       "0       AI  \n",
       "1       AI  \n",
       "2       AI  \n",
       "3       AI  \n",
       "4       AI  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2cb46c8b668>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 305,
       "width": 412
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data[\"category\"].value_counts().plot.bar()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "data.to_csv(\"dataset_not_preprocessed.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>question</th>\n",
       "      <th>answer</th>\n",
       "      <th>category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>['What is AI?'</td>\n",
       "      <td>'Artificial Intelligence is the branch of eng...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>['What is AI?'</td>\n",
       "      <td>'AI is the field of science which concerns it...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>['Are you sentient?'</td>\n",
       "      <td>'Sort of.']</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>['Are you sentient?'</td>\n",
       "      <td>\"By the strictest dictionary definition of th...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>['Are you sentient?'</td>\n",
       "      <td>\"Even though I'm a construct I do have a subj...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               question                                             answer  \\\n",
       "0        ['What is AI?'   'Artificial Intelligence is the branch of eng...   \n",
       "1        ['What is AI?'   'AI is the field of science which concerns it...   \n",
       "2  ['Are you sentient?'                                        'Sort of.']   \n",
       "3  ['Are you sentient?'   \"By the strictest dictionary definition of th...   \n",
       "4  ['Are you sentient?'   \"Even though I'm a construct I do have a subj...   \n",
       "\n",
       "  category  \n",
       "0       AI  \n",
       "1       AI  \n",
       "2       AI  \n",
       "3       AI  \n",
       "4       AI  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "preprocessed_text = []\n",
    "for sent in data[\"question\"]:\n",
    "    sentence=[]\n",
    "    for word in sent.split():\n",
    "        word=word.lower()\n",
    "        word = (\"\".join(x for x in word if x.isalnum()))\n",
    "        sentence.append(word)\n",
    "    \n",
    "    preprocessed_text.append(\" \".join(str(x) for x in sentence))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "data[\"question\"] = preprocessed_text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "preprocessed_text = []\n",
    "for sent in data[\"answer\"]:\n",
    "    sentence=[]\n",
    "    for word in sent.split():\n",
    "        word=word.lower()\n",
    "        word = (\"\".join(x for x in word if x.isalnum()))\n",
    "        sentence.append(word)\n",
    "    \n",
    "    preprocessed_text.append(\" \".join(str(x) for x in sentence))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "data[\"answer\"] = preprocessed_text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>question</th>\n",
       "      <th>answer</th>\n",
       "      <th>category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>what is ai</td>\n",
       "      <td>artificial intelligence is the branch of engin...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>what is ai</td>\n",
       "      <td>ai is the field of science which concerns itse...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>are you sentient</td>\n",
       "      <td>sort of</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>are you sentient</td>\n",
       "      <td>by the strictest dictionary definition of the ...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>are you sentient</td>\n",
       "      <td>even though im a construct i do have a subject...</td>\n",
       "      <td>AI</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           question                                             answer  \\\n",
       "0        what is ai  artificial intelligence is the branch of engin...   \n",
       "1        what is ai  ai is the field of science which concerns itse...   \n",
       "2  are you sentient                                            sort of   \n",
       "3  are you sentient  by the strictest dictionary definition of the ...   \n",
       "4  are you sentient  even though im a construct i do have a subject...   \n",
       "\n",
       "  category  \n",
       "0       AI  \n",
       "1       AI  \n",
       "2       AI  \n",
       "3       AI  \n",
       "4       AI  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "data.to_csv(\"dataset_preprocessed.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<B> Preparing word2vec using own text corpus</B>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data=pd.read_csv(\"dataset_preprocessed.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'what is ai artificial intelligence is the branch of engineering and science devoted to constructing machines that think'"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sent = data[\"question\"][0]+\" \"+data[\"answer\"][0]\n",
    "sent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "list_of_sent=[]\n",
    "\n",
    "for i in range(len(data[\"question\"])):\n",
    "    sent = data[\"question\"][i]+\" \"+data[\"answer\"][i]\n",
    "    list_of_sent.append(sent.split())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "584"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(list_of_sent)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "import gensim\n",
    "model_word2vec = gensim.models.Word2Vec(list_of_sent,min_count=5,size=300, workers=4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<B> Preparing word2vec using own Google news data </B>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "pretrained_embeddings_path = \"GoogleNews-vectors-negative300.bin\"\n",
    "\n",
    "model_word2vec = gensim.models.KeyedVectors.load_word2vec_format(pretrained_embeddings_path, binary=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "question_vector=[]\n",
    "for sent in data[\"question\"]:\n",
    "    sent_vec = [model_word2vec[w] for w in sent if w in model_word2vec.wv.vocab]\n",
    "    question_vector.append(sent_vec)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "answer_vector=[]\n",
    "for sent in data[\"answer\"]:\n",
    "    sent_vec = [model_word2vec[w] for w in sent if w in model_word2vec.wv.vocab]\n",
    "    answer_vector.append(sent_vec)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'ai is the field of science which concerns itself with building hardware and software that replicates the functions of the human mind'"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[\"answer\"][1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "sentend=np.ones((300,),dtype=np.float32) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "for sent in question_vector:\n",
    "    sent[19:]=[]\n",
    "    sent.append(sentend)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "for sent in question_vector:\n",
    "    if(len(sent)<20):\n",
    "        for i in range(20-len(sent)):\n",
    "            sent.append(sentend)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "for sent in answer_vector:\n",
    "    sent[19:]=[]\n",
    "    sent.append(sentend)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "for sent in answer_vector:\n",
    "    if(len(sent)<20):\n",
    "        for i in range(20-len(sent)):\n",
    "            sent.append(sentend)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "with open('conversation.pickle','wb') as f:\n",
    "    pickle.dump([question_vector,answer_vector],f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pickle\n",
    "import numpy as np\n",
    "from keras.models import Sequential\n",
    "import gensim\n",
    "from keras.layers.recurrent import LSTM,SimpleRNN\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.layers import Dense, Dropout, Activation\n",
    "from keras.layers.normalization import BatchNormalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('conversation.pickle','rb') as f:\n",
    "    question_vector , answer_vector = pickle.load(f)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "vec_x=np.array(question_vector,dtype=np.float64)\n",
    "vec_y=np.array(answer_vector,dtype=np.float64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3504000"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vec_x.size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train,x_test, y_train,y_test = train_test_split(vec_x, vec_y, test_size=0.2, random_state=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(467, 20, 300) (467, 20, 300) (117, 20, 300) (117, 20, 300)\n"
     ]
    }
   ],
   "source": [
    "print(x_train.shape, y_train.shape, x_test.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "lstm_2 (LSTM)                (None, 20, 300)           721200    \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 20, 300)           0         \n",
      "_________________________________________________________________\n",
      "lstm_3 (LSTM)                (None, 20, 300)           721200    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 20, 300)           0         \n",
      "_________________________________________________________________\n",
      "lstm_4 (LSTM)                (None, 20, 300)           721200    \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 20, 300)           0         \n",
      "_________________________________________________________________\n",
      "lstm_5 (LSTM)                (None, 20, 300)           721200    \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 20, 300)           0         \n",
      "=================================================================\n",
      "Total params: 2,884,800\n",
      "Trainable params: 2,884,800\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model=Sequential()\n",
    "\n",
    "model.add(LSTM(output_dim=300,input_shape=x_train.shape[1:],return_sequences=True, init='glorot_normal', inner_init='glorot_normal', activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "\n",
    "model.add(LSTM(output_dim=300,input_shape=x_train.shape[1:],return_sequences=True, init='glorot_normal', inner_init='glorot_normal', activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "\n",
    "model.add(LSTM(output_dim=300,input_shape=x_train.shape[1:],return_sequences=True, init='glorot_normal', inner_init='glorot_normal', activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "\n",
    "model.add(LSTM(output_dim=300,input_shape=x_train.shape[1:],return_sequences=True, init='glorot_normal', inner_init='glorot_normal', activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "\n",
    "model.compile(loss='cosine_proximity', optimizer='adam', metrics=['accuracy'])\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 467 samples, validate on 117 samples\n",
      "Epoch 1/200\n",
      "467/467 [==============================] - 10s 21ms/step - loss: -0.3021 - acc: 0.0407 - val_loss: -0.4905 - val_acc: 0.0573\n",
      "Epoch 2/200\n",
      "467/467 [==============================] - 6s 13ms/step - loss: -0.3459 - acc: 0.0582 - val_loss: -0.4968 - val_acc: 0.0483\n",
      "Epoch 3/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3485 - acc: 0.0534 - val_loss: -0.4992 - val_acc: 0.0654\n",
      "Epoch 4/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3492 - acc: 0.0572 - val_loss: -0.4991 - val_acc: 0.0611\n",
      "Epoch 5/200\n",
      "467/467 [==============================] - 6s 13ms/step - loss: -0.3507 - acc: 0.0613 - val_loss: -0.5015 - val_acc: 0.0620\n",
      "Epoch 6/200\n",
      "467/467 [==============================] - 6s 14ms/step - loss: -0.3518 - acc: 0.0617 - val_loss: -0.5012 - val_acc: 0.0637\n",
      "Epoch 7/200\n",
      "467/467 [==============================] - 6s 14ms/step - loss: -0.3521 - acc: 0.0605 - val_loss: -0.5008 - val_acc: 0.0628\n",
      "Epoch 8/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3518 - acc: 0.0589 - val_loss: -0.5008 - val_acc: 0.0624\n",
      "Epoch 9/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3513 - acc: 0.0599 - val_loss: -0.4966 - val_acc: 0.0697\n",
      "Epoch 10/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3514 - acc: 0.0601 - val_loss: -0.5006 - val_acc: 0.0658\n",
      "Epoch 11/200\n",
      "467/467 [==============================] - 6s 14ms/step - loss: -0.3524 - acc: 0.0601 - val_loss: -0.5030 - val_acc: 0.0662\n",
      "Epoch 12/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3530 - acc: 0.0592 - val_loss: -0.5028 - val_acc: 0.0632\n",
      "Epoch 13/200\n",
      "467/467 [==============================] - 6s 14ms/step - loss: -0.3529 - acc: 0.0623 - val_loss: -0.5003 - val_acc: 0.0530\n",
      "Epoch 14/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3536 - acc: 0.0595 - val_loss: -0.5017 - val_acc: 0.0594\n",
      "Epoch 15/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3540 - acc: 0.0596 - val_loss: -0.5035 - val_acc: 0.0628\n",
      "Epoch 16/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3540 - acc: 0.0577 - val_loss: -0.5019 - val_acc: 0.0598\n",
      "Epoch 17/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3529 - acc: 0.0646 - val_loss: -0.5034 - val_acc: 0.0637\n",
      "Epoch 18/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3541 - acc: 0.0604 - val_loss: -0.5037 - val_acc: 0.0577\n",
      "Epoch 19/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3538 - acc: 0.0587 - val_loss: -0.5027 - val_acc: 0.0632\n",
      "Epoch 20/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3549 - acc: 0.0655 - val_loss: -0.5021 - val_acc: 0.0654\n",
      "Epoch 21/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3545 - acc: 0.0589 - val_loss: -0.5045 - val_acc: 0.0641\n",
      "Epoch 22/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3551 - acc: 0.0600 - val_loss: -0.5050 - val_acc: 0.0662\n",
      "Epoch 23/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3567 - acc: 0.0576 - val_loss: -0.5046 - val_acc: 0.0637\n",
      "Epoch 24/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3564 - acc: 0.0591 - val_loss: -0.4977 - val_acc: 0.0573\n",
      "Epoch 25/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3561 - acc: 0.0610 - val_loss: -0.5007 - val_acc: 0.0628\n",
      "Epoch 26/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3560 - acc: 0.0610 - val_loss: -0.5054 - val_acc: 0.0662\n",
      "Epoch 27/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3563 - acc: 0.0616 - val_loss: -0.5042 - val_acc: 0.0632\n",
      "Epoch 28/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3570 - acc: 0.0616 - val_loss: -0.5052 - val_acc: 0.0688\n",
      "Epoch 29/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3587 - acc: 0.0649 - val_loss: -0.5062 - val_acc: 0.0650\n",
      "Epoch 30/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3577 - acc: 0.0586 - val_loss: -0.5002 - val_acc: 0.0594\n",
      "Epoch 31/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3586 - acc: 0.0608 - val_loss: -0.5063 - val_acc: 0.0658\n",
      "Epoch 32/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3589 - acc: 0.0604 - val_loss: -0.5071 - val_acc: 0.0637\n",
      "Epoch 33/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3603 - acc: 0.0604 - val_loss: -0.5072 - val_acc: 0.0667\n",
      "Epoch 34/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3598 - acc: 0.0613 - val_loss: -0.5006 - val_acc: 0.0628\n",
      "Epoch 35/200\n",
      "467/467 [==============================] - 7s 14ms/step - loss: -0.3616 - acc: 0.0616 - val_loss: -0.5071 - val_acc: 0.0684\n",
      "Epoch 36/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3617 - acc: 0.0663 - val_loss: -0.5041 - val_acc: 0.0658\n",
      "Epoch 37/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3626 - acc: 0.0624 - val_loss: -0.5081 - val_acc: 0.0564\n",
      "Epoch 38/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3626 - acc: 0.0617 - val_loss: -0.5049 - val_acc: 0.0556\n",
      "Epoch 39/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3643 - acc: 0.0630 - val_loss: -0.5076 - val_acc: 0.0667\n",
      "Epoch 40/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3628 - acc: 0.0641 - val_loss: -0.5076 - val_acc: 0.0637\n",
      "Epoch 41/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3635 - acc: 0.0618 - val_loss: -0.5031 - val_acc: 0.0650\n",
      "Epoch 42/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3646 - acc: 0.0618 - val_loss: -0.5076 - val_acc: 0.0598\n",
      "Epoch 43/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3657 - acc: 0.0627 - val_loss: -0.5067 - val_acc: 0.0671\n",
      "Epoch 44/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3670 - acc: 0.0633 - val_loss: -0.5061 - val_acc: 0.0628\n",
      "Epoch 45/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3666 - acc: 0.0625 - val_loss: -0.5064 - val_acc: 0.0603\n",
      "Epoch 46/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3668 - acc: 0.0627 - val_loss: -0.5091 - val_acc: 0.0675\n",
      "Epoch 47/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3698 - acc: 0.0632 - val_loss: -0.5030 - val_acc: 0.0603\n",
      "Epoch 48/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3690 - acc: 0.0646 - val_loss: -0.5057 - val_acc: 0.0654\n",
      "Epoch 49/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3696 - acc: 0.0657 - val_loss: -0.5109 - val_acc: 0.0637\n",
      "Epoch 50/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3700 - acc: 0.0630 - val_loss: -0.5100 - val_acc: 0.0671\n",
      "Epoch 51/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3693 - acc: 0.0584 - val_loss: -0.5092 - val_acc: 0.0620\n",
      "Epoch 52/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3707 - acc: 0.0697 - val_loss: -0.5079 - val_acc: 0.0671\n",
      "Epoch 53/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3718 - acc: 0.0631 - val_loss: -0.5066 - val_acc: 0.0615\n",
      "Epoch 54/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3727 - acc: 0.0623 - val_loss: -0.5079 - val_acc: 0.0658\n",
      "Epoch 55/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3729 - acc: 0.0633 - val_loss: -0.5106 - val_acc: 0.0662\n",
      "Epoch 56/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3748 - acc: 0.0631 - val_loss: -0.5089 - val_acc: 0.0697\n",
      "Epoch 57/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3732 - acc: 0.0632 - val_loss: -0.5086 - val_acc: 0.0739\n",
      "Epoch 58/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3736 - acc: 0.0668 - val_loss: -0.5110 - val_acc: 0.0650\n",
      "Epoch 59/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3750 - acc: 0.0649 - val_loss: -0.5053 - val_acc: 0.0667\n",
      "Epoch 60/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3741 - acc: 0.0621 - val_loss: -0.5079 - val_acc: 0.0688\n",
      "Epoch 61/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3760 - acc: 0.0745 - val_loss: -0.5068 - val_acc: 0.0714\n",
      "Epoch 62/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3761 - acc: 0.0686 - val_loss: -0.5101 - val_acc: 0.0688\n",
      "Epoch 63/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3773 - acc: 0.0635 - val_loss: -0.5073 - val_acc: 0.0769\n",
      "Epoch 64/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3774 - acc: 0.0645 - val_loss: -0.5137 - val_acc: 0.0692\n",
      "Epoch 65/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3767 - acc: 0.0690 - val_loss: -0.5098 - val_acc: 0.0714\n",
      "Epoch 66/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3762 - acc: 0.0623 - val_loss: -0.5128 - val_acc: 0.0667\n",
      "Epoch 67/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3749 - acc: 0.0647 - val_loss: -0.5117 - val_acc: 0.0726\n",
      "Epoch 68/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3776 - acc: 0.0690 - val_loss: -0.5088 - val_acc: 0.0697\n",
      "Epoch 69/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3783 - acc: 0.0623 - val_loss: -0.5081 - val_acc: 0.0739\n",
      "Epoch 70/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3784 - acc: 0.0643 - val_loss: -0.5053 - val_acc: 0.0744\n",
      "Epoch 71/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3805 - acc: 0.0652 - val_loss: -0.5056 - val_acc: 0.0709\n",
      "Epoch 72/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3798 - acc: 0.0643 - val_loss: -0.5092 - val_acc: 0.0709\n",
      "Epoch 73/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3792 - acc: 0.0706 - val_loss: -0.5096 - val_acc: 0.0739\n",
      "Epoch 74/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3806 - acc: 0.0652 - val_loss: -0.5106 - val_acc: 0.0697\n",
      "Epoch 75/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3800 - acc: 0.0618 - val_loss: -0.5111 - val_acc: 0.0718\n",
      "Epoch 76/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3816 - acc: 0.0712 - val_loss: -0.5116 - val_acc: 0.0675\n",
      "Epoch 77/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3814 - acc: 0.0713 - val_loss: -0.5095 - val_acc: 0.0705\n",
      "Epoch 78/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3822 - acc: 0.0694 - val_loss: -0.5104 - val_acc: 0.0697\n",
      "Epoch 79/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3837 - acc: 0.0651 - val_loss: -0.5097 - val_acc: 0.0774\n",
      "Epoch 80/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3844 - acc: 0.0650 - val_loss: -0.5085 - val_acc: 0.0769\n",
      "Epoch 81/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3834 - acc: 0.0679 - val_loss: -0.5075 - val_acc: 0.0722\n",
      "Epoch 82/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3837 - acc: 0.0653 - val_loss: -0.5089 - val_acc: 0.0684\n",
      "Epoch 83/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3834 - acc: 0.0696 - val_loss: -0.5067 - val_acc: 0.0769\n",
      "Epoch 84/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3829 - acc: 0.0697 - val_loss: -0.5080 - val_acc: 0.0705\n",
      "Epoch 85/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3836 - acc: 0.0656 - val_loss: -0.5058 - val_acc: 0.0671\n",
      "Epoch 86/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3836 - acc: 0.0698 - val_loss: -0.5030 - val_acc: 0.0658\n",
      "Epoch 87/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3827 - acc: 0.0692 - val_loss: -0.5090 - val_acc: 0.0756\n",
      "Epoch 88/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3841 - acc: 0.0671 - val_loss: -0.5092 - val_acc: 0.0748\n",
      "Epoch 89/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3842 - acc: 0.0712 - val_loss: -0.5086 - val_acc: 0.0731\n",
      "Epoch 90/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3846 - acc: 0.0696 - val_loss: -0.5076 - val_acc: 0.0786\n",
      "Epoch 91/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3861 - acc: 0.0677 - val_loss: -0.5002 - val_acc: 0.0752\n",
      "Epoch 92/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3849 - acc: 0.0655 - val_loss: -0.5019 - val_acc: 0.0842\n",
      "Epoch 93/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3848 - acc: 0.0721 - val_loss: -0.5104 - val_acc: 0.0769\n",
      "Epoch 94/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3860 - acc: 0.0666 - val_loss: -0.5086 - val_acc: 0.0842\n",
      "Epoch 95/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3856 - acc: 0.0690 - val_loss: -0.5085 - val_acc: 0.0688\n",
      "Epoch 96/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3843 - acc: 0.0673 - val_loss: -0.5079 - val_acc: 0.0761\n",
      "Epoch 97/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3840 - acc: 0.0662 - val_loss: -0.5047 - val_acc: 0.0786\n",
      "Epoch 98/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3857 - acc: 0.0704 - val_loss: -0.5078 - val_acc: 0.0722\n",
      "Epoch 99/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3851 - acc: 0.0711 - val_loss: -0.5098 - val_acc: 0.0803\n",
      "Epoch 100/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3868 - acc: 0.0726 - val_loss: -0.5095 - val_acc: 0.0846\n",
      "Epoch 101/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3866 - acc: 0.0713 - val_loss: -0.5108 - val_acc: 0.0842\n",
      "Epoch 102/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3855 - acc: 0.0754 - val_loss: -0.5084 - val_acc: 0.0705\n",
      "Epoch 103/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3854 - acc: 0.0743 - val_loss: -0.5075 - val_acc: 0.0795\n",
      "Epoch 104/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3878 - acc: 0.0736 - val_loss: -0.5077 - val_acc: 0.0838\n",
      "Epoch 105/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3894 - acc: 0.0729 - val_loss: -0.5078 - val_acc: 0.0859\n",
      "Epoch 106/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3890 - acc: 0.0771 - val_loss: -0.5067 - val_acc: 0.0868\n",
      "Epoch 107/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3894 - acc: 0.0690 - val_loss: -0.5088 - val_acc: 0.0885\n",
      "Epoch 108/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3890 - acc: 0.0718 - val_loss: -0.5071 - val_acc: 0.0744\n",
      "Epoch 109/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3851 - acc: 0.0709 - val_loss: -0.5109 - val_acc: 0.0808\n",
      "Epoch 110/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3871 - acc: 0.0704 - val_loss: -0.5111 - val_acc: 0.0726\n",
      "Epoch 111/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3880 - acc: 0.0761 - val_loss: -0.5046 - val_acc: 0.0752\n",
      "Epoch 112/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3886 - acc: 0.0727 - val_loss: -0.5086 - val_acc: 0.0812\n",
      "Epoch 113/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3888 - acc: 0.0745 - val_loss: -0.5116 - val_acc: 0.0846\n",
      "Epoch 114/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3897 - acc: 0.0737 - val_loss: -0.5092 - val_acc: 0.0919\n",
      "Epoch 115/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3901 - acc: 0.0781 - val_loss: -0.5087 - val_acc: 0.0850\n",
      "Epoch 116/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3894 - acc: 0.0743 - val_loss: -0.5077 - val_acc: 0.0825\n",
      "Epoch 117/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3895 - acc: 0.0746 - val_loss: -0.5080 - val_acc: 0.0803\n",
      "Epoch 118/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3910 - acc: 0.0776 - val_loss: -0.5111 - val_acc: 0.0893\n",
      "Epoch 119/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3887 - acc: 0.0779 - val_loss: -0.5089 - val_acc: 0.0838\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 120/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3900 - acc: 0.0746 - val_loss: -0.5094 - val_acc: 0.0808\n",
      "Epoch 121/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3900 - acc: 0.0744 - val_loss: -0.5055 - val_acc: 0.0885\n",
      "Epoch 122/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3912 - acc: 0.0723 - val_loss: -0.5088 - val_acc: 0.0855\n",
      "Epoch 123/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3911 - acc: 0.0770 - val_loss: -0.5123 - val_acc: 0.0855\n",
      "Epoch 124/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3910 - acc: 0.0741 - val_loss: -0.5105 - val_acc: 0.0791\n",
      "Epoch 125/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3918 - acc: 0.0772 - val_loss: -0.5114 - val_acc: 0.0876\n",
      "Epoch 126/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3919 - acc: 0.0798 - val_loss: -0.5063 - val_acc: 0.0863\n",
      "Epoch 127/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3914 - acc: 0.0807 - val_loss: -0.5107 - val_acc: 0.0833\n",
      "Epoch 128/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3909 - acc: 0.0755 - val_loss: -0.5127 - val_acc: 0.0855\n",
      "Epoch 129/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3925 - acc: 0.0793 - val_loss: -0.5088 - val_acc: 0.0863\n",
      "Epoch 130/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3910 - acc: 0.0759 - val_loss: -0.5123 - val_acc: 0.0842\n",
      "Epoch 131/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3887 - acc: 0.0792 - val_loss: -0.5166 - val_acc: 0.0791\n",
      "Epoch 132/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3900 - acc: 0.0709 - val_loss: -0.5147 - val_acc: 0.0876\n",
      "Epoch 133/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3912 - acc: 0.0713 - val_loss: -0.5123 - val_acc: 0.0868\n",
      "Epoch 134/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3924 - acc: 0.0779 - val_loss: -0.5119 - val_acc: 0.0906\n",
      "Epoch 135/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3914 - acc: 0.0796 - val_loss: -0.5122 - val_acc: 0.0889\n",
      "Epoch 136/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3927 - acc: 0.0726 - val_loss: -0.5131 - val_acc: 0.0850\n",
      "Epoch 137/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3927 - acc: 0.0827 - val_loss: -0.5100 - val_acc: 0.0906\n",
      "Epoch 138/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3925 - acc: 0.0810 - val_loss: -0.5133 - val_acc: 0.0846\n",
      "Epoch 139/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3927 - acc: 0.0768 - val_loss: -0.5129 - val_acc: 0.0778\n",
      "Epoch 140/200\n",
      "467/467 [==============================] - 7s 15ms/step - loss: -0.3934 - acc: 0.0813 - val_loss: -0.5108 - val_acc: 0.0872\n",
      "Epoch 141/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3937 - acc: 0.0848 - val_loss: -0.5111 - val_acc: 0.0812\n",
      "Epoch 142/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3926 - acc: 0.0812 - val_loss: -0.5084 - val_acc: 0.0889\n",
      "Epoch 143/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3928 - acc: 0.0814 - val_loss: -0.5108 - val_acc: 0.0863\n",
      "Epoch 144/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3943 - acc: 0.0742 - val_loss: -0.5090 - val_acc: 0.0855\n",
      "Epoch 145/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3930 - acc: 0.0775 - val_loss: -0.5079 - val_acc: 0.0812\n",
      "Epoch 146/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3934 - acc: 0.0792 - val_loss: -0.5140 - val_acc: 0.0821\n",
      "Epoch 147/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3933 - acc: 0.0772 - val_loss: -0.5087 - val_acc: 0.0829\n",
      "Epoch 148/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3918 - acc: 0.0788 - val_loss: -0.5032 - val_acc: 0.0859\n",
      "Epoch 149/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3915 - acc: 0.0785 - val_loss: -0.5104 - val_acc: 0.0744\n",
      "Epoch 150/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3887 - acc: 0.0822 - val_loss: -0.5071 - val_acc: 0.0821\n",
      "Epoch 151/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3919 - acc: 0.0797 - val_loss: -0.5091 - val_acc: 0.0821\n",
      "Epoch 152/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3929 - acc: 0.0749 - val_loss: -0.5077 - val_acc: 0.0765\n",
      "Epoch 153/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3942 - acc: 0.0805 - val_loss: -0.5094 - val_acc: 0.0885\n",
      "Epoch 154/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3935 - acc: 0.0790 - val_loss: -0.5095 - val_acc: 0.0791\n",
      "Epoch 155/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3944 - acc: 0.0799 - val_loss: -0.5092 - val_acc: 0.0855\n",
      "Epoch 156/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3944 - acc: 0.0885 - val_loss: -0.5078 - val_acc: 0.0752\n",
      "Epoch 157/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3924 - acc: 0.0785 - val_loss: -0.5105 - val_acc: 0.0778\n",
      "Epoch 158/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3916 - acc: 0.0749 - val_loss: -0.5093 - val_acc: 0.0778\n",
      "Epoch 159/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3938 - acc: 0.0805 - val_loss: -0.5095 - val_acc: 0.0765\n",
      "Epoch 160/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3940 - acc: 0.0753 - val_loss: -0.5105 - val_acc: 0.0786\n",
      "Epoch 161/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3938 - acc: 0.0840 - val_loss: -0.5111 - val_acc: 0.0885\n",
      "Epoch 162/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3927 - acc: 0.0847 - val_loss: -0.5126 - val_acc: 0.0927\n",
      "Epoch 163/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3928 - acc: 0.0823 - val_loss: -0.5080 - val_acc: 0.0850\n",
      "Epoch 164/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3936 - acc: 0.0815 - val_loss: -0.5111 - val_acc: 0.0846\n",
      "Epoch 165/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3946 - acc: 0.0785 - val_loss: -0.5109 - val_acc: 0.0838\n",
      "Epoch 166/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3957 - acc: 0.0813 - val_loss: -0.5113 - val_acc: 0.0821\n",
      "Epoch 167/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3947 - acc: 0.0827 - val_loss: -0.5053 - val_acc: 0.0833\n",
      "Epoch 168/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3940 - acc: 0.0834 - val_loss: -0.5105 - val_acc: 0.0833\n",
      "Epoch 169/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3948 - acc: 0.0810 - val_loss: -0.5097 - val_acc: 0.0829\n",
      "Epoch 170/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3950 - acc: 0.0868 - val_loss: -0.5129 - val_acc: 0.0838\n",
      "Epoch 171/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3950 - acc: 0.0781 - val_loss: -0.5103 - val_acc: 0.0859\n",
      "Epoch 172/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3956 - acc: 0.0794 - val_loss: -0.5122 - val_acc: 0.0940\n",
      "Epoch 173/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3953 - acc: 0.0855 - val_loss: -0.5145 - val_acc: 0.0846\n",
      "Epoch 174/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3953 - acc: 0.0861 - val_loss: -0.5099 - val_acc: 0.0893\n",
      "Epoch 175/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3966 - acc: 0.0860 - val_loss: -0.5113 - val_acc: 0.0902\n",
      "Epoch 176/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3972 - acc: 0.0864 - val_loss: -0.5117 - val_acc: 0.0846\n",
      "Epoch 177/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3966 - acc: 0.0862 - val_loss: -0.5134 - val_acc: 0.0808\n",
      "Epoch 178/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3974 - acc: 0.0831 - val_loss: -0.5129 - val_acc: 0.0927\n",
      "Epoch 179/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3958 - acc: 0.0846 - val_loss: -0.5130 - val_acc: 0.0855\n",
      "Epoch 180/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3965 - acc: 0.0850 - val_loss: -0.5099 - val_acc: 0.0940\n",
      "Epoch 181/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3971 - acc: 0.0849 - val_loss: -0.5096 - val_acc: 0.0885\n",
      "Epoch 182/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3966 - acc: 0.0834 - val_loss: -0.5080 - val_acc: 0.0829\n",
      "Epoch 183/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3963 - acc: 0.0850 - val_loss: -0.5096 - val_acc: 0.0842\n",
      "Epoch 184/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3971 - acc: 0.0840 - val_loss: -0.5125 - val_acc: 0.0872\n",
      "Epoch 185/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3969 - acc: 0.0850 - val_loss: -0.5138 - val_acc: 0.0932\n",
      "Epoch 186/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3957 - acc: 0.0797 - val_loss: -0.5132 - val_acc: 0.0880\n",
      "Epoch 187/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3955 - acc: 0.0859 - val_loss: -0.5117 - val_acc: 0.0906\n",
      "Epoch 188/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3950 - acc: 0.0824 - val_loss: -0.5102 - val_acc: 0.0855\n",
      "Epoch 189/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3954 - acc: 0.0832 - val_loss: -0.5108 - val_acc: 0.0863\n",
      "Epoch 190/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3962 - acc: 0.0837 - val_loss: -0.5070 - val_acc: 0.0910\n",
      "Epoch 191/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3976 - acc: 0.0866 - val_loss: -0.5122 - val_acc: 0.0910\n",
      "Epoch 192/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3975 - acc: 0.0855 - val_loss: -0.5108 - val_acc: 0.0919\n",
      "Epoch 193/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3971 - acc: 0.0799 - val_loss: -0.5127 - val_acc: 0.0944\n",
      "Epoch 194/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3969 - acc: 0.0870 - val_loss: -0.5112 - val_acc: 0.0953\n",
      "Epoch 195/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3970 - acc: 0.0908 - val_loss: -0.5115 - val_acc: 0.0949\n",
      "Epoch 196/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3978 - acc: 0.0865 - val_loss: -0.5092 - val_acc: 0.0949\n",
      "Epoch 197/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3981 - acc: 0.0829 - val_loss: -0.5077 - val_acc: 0.0949\n",
      "Epoch 198/200\n",
      "467/467 [==============================] - 8s 17ms/step - loss: -0.3977 - acc: 0.0899 - val_loss: -0.5095 - val_acc: 0.0855\n",
      "Epoch 199/200\n",
      "467/467 [==============================] - 7s 16ms/step - loss: -0.3957 - acc: 0.0862 - val_loss: -0.5069 - val_acc: 0.0842\n",
      "Epoch 200/200\n",
      "467/467 [==============================] - 8s 16ms/step - loss: -0.3966 - acc: 0.0845 - val_loss: -0.5119 - val_acc: 0.0859\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x2cc6c4d95f8>"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(x_train, y_train, nb_epoch=200,validation_data=(x_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('LSTM_chatbot.h5')   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}