Named Entity Recognition using LSTM wihout padded values.ipynb

{"cells":[{"metadata":{"id":"q4gSqDnOaJ7W"},"cell_type":"markdown","source":"<h2 align=center> Named Entity Recognition (NER) using LSTMs with Keras</h2>"},{"metadata":{"trusted":true},"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n\n# You can write up to 5GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","execution_count":1,"outputs":[{"output_type":"stream","text":"/kaggle/input/ner_dataset.csv\n","name":"stdout"}]},{"metadata":{"id":"oLK7Y1jiNXDa","outputId":"0f319464-ee49-4035-f6fb-8396f488c41f","trusted":true},"cell_type":"code","source":"%matplotlib inline\nimport matplotlib.pyplot as plt\nnp.random.seed(0)\nplt.style.use(\"ggplot\")\n\nimport tensorflow as tf\nprint('Tensorflow version:', tf.__version__)","execution_count":2,"outputs":[{"output_type":"stream","text":"Tensorflow version: 2.1.0\n","name":"stdout"}]},{"metadata":{"id":"4N_AW6lMbB5N"},"cell_type":"markdown","source":"# LOADING AND EXPLORING THE NER DATASET-->"},{"metadata":{},"cell_type":"markdown","source":"*Essential info about the tagged entities*:\n- geo = Geographical Entity\n- org = Organization\n- per = Person\n- gpe = Geopolitical Entity\n- tim = Time indicator\n- art = Artifact\n- eve = Event\n- nat = Natural Phenomenon"},{"metadata":{"id":"mCKmz4SAbI_m","outputId":"a03b1aed-dc60-4f03-cb06-19e440fa6367","trusted":true},"cell_type":"code","source":"df=pd.read_csv(\"/kaggle/input/ner_dataset.csv\",encoding='latin1')\ndf.head(30)","execution_count":3,"outputs":[{"output_type":"execute_result","execution_count":3,"data":{"text/plain":"     Sentence #           Word  POS    Tag\n0   Sentence: 1      Thousands  NNS      O\n1           NaN             of   IN      O\n2           NaN  demonstrators  NNS      O\n3           NaN           have  VBP      O\n4           NaN        marched  VBN      O\n5           NaN        through   IN      O\n6           NaN         London  NNP  B-geo\n7           NaN             to   TO      O\n8           NaN        protest   VB      O\n9           NaN            the   DT      O\n10          NaN            war   NN      O\n11          NaN             in   IN      O\n12          NaN           Iraq  NNP  B-geo\n13          NaN            and   CC      O\n14          NaN         demand   VB      O\n15          NaN            the   DT      O\n16          NaN     withdrawal   NN      O\n17          NaN             of   IN      O\n18          NaN        British   JJ  B-gpe\n19          NaN         troops  NNS      O\n20          NaN           from   IN      O\n21          NaN           that   DT      O\n22          NaN        country   NN      O\n23          NaN              .    .      O\n24  Sentence: 2       Families  NNS      O\n25          NaN             of   IN      O\n26          NaN       soldiers  NNS      O\n27          NaN         killed  VBN      O\n28          NaN             in   IN      O\n29          NaN            the   DT      O","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>Sentence #</th>\n      <th>Word</th>\n      <th>POS</th>\n      <th>Tag</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>Sentence: 1</td>\n      <td>Thousands</td>\n      <td>NNS</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>NaN</td>\n      <td>of</td>\n      <td>IN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>NaN</td>\n      <td>demonstrators</td>\n      <td>NNS</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>NaN</td>\n      <td>have</td>\n      <td>VBP</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>NaN</td>\n      <td>marched</td>\n      <td>VBN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>NaN</td>\n      <td>through</td>\n      <td>IN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>NaN</td>\n      <td>London</td>\n      <td>NNP</td>\n      <td>B-geo</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>NaN</td>\n      <td>to</td>\n      <td>TO</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td>NaN</td>\n      <td>protest</td>\n      <td>VB</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>9</th>\n      <td>NaN</td>\n      <td>the</td>\n      <td>DT</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>10</th>\n      <td>NaN</td>\n      <td>war</td>\n      <td>NN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>11</th>\n      <td>NaN</td>\n      <td>in</td>\n      <td>IN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>12</th>\n      <td>NaN</td>\n      <td>Iraq</td>\n      <td>NNP</td>\n      <td>B-geo</td>\n    </tr>\n    <tr>\n      <th>13</th>\n      <td>NaN</td>\n      <td>and</td>\n      <td>CC</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>14</th>\n      <td>NaN</td>\n      <td>demand</td>\n      <td>VB</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>15</th>\n      <td>NaN</td>\n      <td>the</td>\n      <td>DT</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>16</th>\n      <td>NaN</td>\n      <td>withdrawal</td>\n      <td>NN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>17</th>\n      <td>NaN</td>\n      <td>of</td>\n      <td>IN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>18</th>\n      <td>NaN</td>\n      <td>British</td>\n      <td>JJ</td>\n      <td>B-gpe</td>\n    </tr>\n    <tr>\n      <th>19</th>\n      <td>NaN</td>\n      <td>troops</td>\n      <td>NNS</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>20</th>\n      <td>NaN</td>\n      <td>from</td>\n      <td>IN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>21</th>\n      <td>NaN</td>\n      <td>that</td>\n      <td>DT</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>22</th>\n      <td>NaN</td>\n      <td>country</td>\n      <td>NN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>23</th>\n      <td>NaN</td>\n      <td>.</td>\n      <td>.</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>24</th>\n      <td>Sentence: 2</td>\n      <td>Families</td>\n      <td>NNS</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>25</th>\n      <td>NaN</td>\n      <td>of</td>\n      <td>IN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>26</th>\n      <td>NaN</td>\n      <td>soldiers</td>\n      <td>NNS</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>27</th>\n      <td>NaN</td>\n      <td>killed</td>\n      <td>VBN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>28</th>\n      <td>NaN</td>\n      <td>in</td>\n      <td>IN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>29</th>\n      <td>NaN</td>\n      <td>the</td>\n      <td>DT</td>\n      <td>O</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"metadata":{"id":"riOztP-8NXHT","outputId":"200d251b-17c9-451b-9e96-523f458be47e","trusted":true},"cell_type":"code","source":"df.isnull().sum()","execution_count":4,"outputs":[{"output_type":"execute_result","execution_count":4,"data":{"text/plain":"Sentence #    1000616\nWord                0\nPOS                 0\nTag                 0\ndtype: int64"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"Looking at the dataset here we will use ffill feature to fill the null values."},{"metadata":{"trusted":true},"cell_type":"code","source":"df=df.fillna(method='ffill')\ndf.head()","execution_count":5,"outputs":[{"output_type":"execute_result","execution_count":5,"data":{"text/plain":"    Sentence #           Word  POS Tag\n0  Sentence: 1      Thousands  NNS   O\n1  Sentence: 1             of   IN   O\n2  Sentence: 1  demonstrators  NNS   O\n3  Sentence: 1           have  VBP   O\n4  Sentence: 1        marched  VBN   O","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>Sentence #</th>\n      <th>Word</th>\n      <th>POS</th>\n      <th>Tag</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>Sentence: 1</td>\n      <td>Thousands</td>\n      <td>NNS</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>Sentence: 1</td>\n      <td>of</td>\n      <td>IN</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>Sentence: 1</td>\n      <td>demonstrators</td>\n      <td>NNS</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>Sentence: 1</td>\n      <td>have</td>\n      <td>VBP</td>\n      <td>O</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>Sentence: 1</td>\n      <td>marched</td>\n      <td>VBN</td>\n      <td>O</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df.isnull().sum()","execution_count":6,"outputs":[{"output_type":"execute_result","execution_count":6,"data":{"text/plain":"Sentence #    0\nWord          0\nPOS           0\nTag           0\ndtype: int64"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df['Tag'].value_counts()","execution_count":7,"outputs":[{"output_type":"execute_result","execution_count":7,"data":{"text/plain":"O        887908\nB-geo     37644\nB-tim     20333\nB-org     20143\nI-per     17251\nB-per     16990\nI-org     16784\nB-gpe     15870\nI-geo      7414\nI-tim      6528\nB-art       402\nB-eve       308\nI-art       297\nI-eve       253\nB-nat       201\nI-gpe       198\nI-nat        51\nName: Tag, dtype: int64"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"n_tags=df['Tag'].nunique()\nn_tags","execution_count":8,"outputs":[{"output_type":"execute_result","execution_count":8,"data":{"text/plain":"17"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"n_words=df['Word'].nunique()\nn_words","execution_count":9,"outputs":[{"output_type":"execute_result","execution_count":9,"data":{"text/plain":"35178"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"words=list(set(df['Word']))","execution_count":10,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# words","execution_count":11,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"n_words=len(words)","execution_count":12,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"tags=list(set(df['Tag']))","execution_count":13,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"tags","execution_count":14,"outputs":[{"output_type":"execute_result","execution_count":14,"data":{"text/plain":"['B-art',\n 'I-art',\n 'I-eve',\n 'I-geo',\n 'O',\n 'I-gpe',\n 'B-nat',\n 'I-nat',\n 'B-gpe',\n 'I-tim',\n 'B-per',\n 'I-org',\n 'B-org',\n 'B-tim',\n 'B-eve',\n 'B-geo',\n 'I-per']"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"n_tags=len(tags)","execution_count":15,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"class GetSentence(object):\n    def __init__(self, data):\n        self.n_sentence=1\n        self.data=data\n        self.empty = False\n        function=lambda d:[(w, p, t) for w, p, t in zip(d[\"Word\"].values.tolist(),\n                                                        d[\"POS\"].values.tolist(),\n                                                        d[\"Tag\"].values.tolist())]\n        \n        self.group_sent = self.data.groupby(\"Sentence #\").apply(function)\n        self.all_sentences = [d for d in self.group_sent] ","execution_count":16,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"get=GetSentence(df)\nsentences=get.all_sentences","execution_count":17,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"sentences[0]","execution_count":18,"outputs":[{"output_type":"execute_result","execution_count":18,"data":{"text/plain":"[('Thousands', 'NNS', 'O'),\n ('of', 'IN', 'O'),\n ('demonstrators', 'NNS', 'O'),\n ('have', 'VBP', 'O'),\n ('marched', 'VBN', 'O'),\n ('through', 'IN', 'O'),\n ('London', 'NNP', 'B-geo'),\n ('to', 'TO', 'O'),\n ('protest', 'VB', 'O'),\n ('the', 'DT', 'O'),\n ('war', 'NN', 'O'),\n ('in', 'IN', 'O'),\n ('Iraq', 'NNP', 'B-geo'),\n ('and', 'CC', 'O'),\n ('demand', 'VB', 'O'),\n ('the', 'DT', 'O'),\n ('withdrawal', 'NN', 'O'),\n ('of', 'IN', 'O'),\n ('British', 'JJ', 'B-gpe'),\n ('troops', 'NNS', 'O'),\n ('from', 'IN', 'O'),\n ('that', 'DT', 'O'),\n ('country', 'NN', 'O'),\n ('.', '.', 'O')]"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"sentences[6]","execution_count":19,"outputs":[{"output_type":"execute_result","execution_count":19,"data":{"text/plain":"[('He', 'PRP', 'O'),\n ('said', 'VBD', 'O'),\n ('last', 'JJ', 'O'),\n ('week', 'NN', 'O'),\n (\"'s\", 'POS', 'O'),\n ('tsunami', 'NN', 'O'),\n ('and', 'CC', 'O'),\n ('the', 'DT', 'O'),\n ('massive', 'JJ', 'O'),\n ('underwater', 'NN', 'O'),\n ('earthquake', 'NN', 'O'),\n ('that', 'WDT', 'O'),\n ('triggered', 'VBD', 'O'),\n ('it', 'PRP', 'O'),\n ('has', 'VBZ', 'O'),\n ('affected', 'VBN', 'O'),\n ('millions', 'NNS', 'O'),\n ('in', 'IN', 'O'),\n ('Asia', 'NNP', 'B-geo'),\n ('and', 'CC', 'O'),\n ('Africa', 'NNP', 'B-geo'),\n ('.', '.', 'O')]"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"len(sentences)","execution_count":20,"outputs":[{"output_type":"execute_result","execution_count":20,"data":{"text/plain":"47959"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"# For the input of LSTM model all the sentences must be padded to same length,for that we must know the maximum length of the sequence in the list of sentences."},{"metadata":{"trusted":true},"cell_type":"code","source":"maxl = max([len(s) for s in sentences])\nprint ('Maximum sequence length in the list of sentences:', maxl)","execution_count":21,"outputs":[{"output_type":"stream","text":"Maximum sequence length in the list of sentences: 104\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"# We had splitted each sentences as a list of tuples of the word,POS and tags."},{"metadata":{"trusted":true},"cell_type":"code","source":"w_index={w:i for i,w in enumerate(words)}\nt_index={t:j for j,t in enumerate(tags)}","execution_count":22,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# w_index","execution_count":23,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# t_index","execution_count":24,"outputs":[]},{"metadata":{"id":"VdJst_g5NYY_","trusted":true},"cell_type":"code","source":"from tensorflow.keras.preprocessing.sequence import pad_sequences\nfrom tensorflow.keras.utils import to_categorical","execution_count":25,"outputs":[]},{"metadata":{"id":"nMUQLppspkPj","trusted":true},"cell_type":"code","source":"maxl","execution_count":26,"outputs":[{"output_type":"execute_result","execution_count":26,"data":{"text/plain":"104"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"This is the length of longest sequence in the list os sentences.\nSo this should be our size of max_padding"},{"metadata":{"id":"GhiSTt2UdzYC","outputId":"95d275f6-f386-46d8-a583-2861d64c6b72","trusted":true},"cell_type":"code","source":"X=[[w_index[w[0]]for w in s]for s in sentences]","execution_count":27,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X=pad_sequences(maxlen=maxl,padding='post',sequences=X)","execution_count":28,"outputs":[]},{"metadata":{"id":"SvENHO18pkaQ","trusted":true},"cell_type":"code","source":"X","execution_count":29,"outputs":[{"output_type":"execute_result","execution_count":29,"data":{"text/plain":"array([[31084, 23162, 21337, ...,     0,     0,     0],\n       [17300, 31112, 15817, ...,     0,     0,     0],\n       [17232, 18959, 18840, ...,     0,     0,     0],\n       ...,\n       [24727, 20700,  4587, ...,     0,     0,     0],\n       [24008,  8387, 20860, ...,     0,     0,     0],\n       [ 3167, 19032, 16643, ...,     0,     0,     0]], dtype=int32)"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"Y = [[t_index[w[2]] for w in s] for s in sentences]","execution_count":30,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"Y=pad_sequences(maxlen=maxl,padding='post',sequences=Y)","execution_count":31,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"Y[1]","execution_count":32,"outputs":[{"output_type":"execute_result","execution_count":32,"data":{"text/plain":"array([ 8,  4,  4,  4,  4,  4,  4,  4,  4,  4,  4,  4,  4,  4,  4, 13,  4,\n        4,  4, 12,  4,  4,  4,  4,  4,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0], dtype=int32)"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Y","execution_count":33,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# TRAIN TEST SPLIT-->"},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.model_selection import train_test_split","execution_count":34,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"x_train,x_test,y_train,y_test=train_test_split(X,Y,test_size=0.1,random_state=42)","execution_count":35,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import keras.utils","execution_count":36,"outputs":[{"output_type":"stream","text":"Using TensorFlow backend.\n","name":"stderr"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_train = keras.utils.to_categorical(y_train)","execution_count":37,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print(x_train.shape, y_train.shape)","execution_count":38,"outputs":[{"output_type":"stream","text":"(43163, 104) (43163, 104, 17)\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"print(x_test.shape, y_test.shape)","execution_count":39,"outputs":[{"output_type":"stream","text":"(4796, 104) (4796, 104)\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_test = keras.utils.to_categorical(y_test)","execution_count":40,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print(x_test.shape, y_test.shape)","execution_count":41,"outputs":[{"output_type":"stream","text":"(4796, 104) (4796, 104, 17)\n","name":"stdout"}]},{"metadata":{"id":"P-r4PR85hpoF"},"cell_type":"markdown","source":"# BIDIRECTIONAL LSTM MODEL-->"},{"metadata":{"id":"Y2vM7IkXpkiH","trusted":true},"cell_type":"code","source":"from tensorflow.keras import Model, Input\nfrom tensorflow.keras.layers import LSTM, Embedding, Dense\nfrom tensorflow.keras.layers import TimeDistributed, SpatialDropout1D, Bidirectional","execution_count":42,"outputs":[]},{"metadata":{"id":"kOBpQg26pkqh","trusted":true},"cell_type":"code","source":"input=Input(shape=(104))\nmodel=Embedding(input_dim=n_words,output_dim=104,input_length=140)(input)\nmodel=SpatialDropout1D(0.1)(model)\nmodel=Bidirectional(LSTM(units=150,return_sequences=True, recurrent_dropout=0.1))(model)\noutput = TimeDistributed(Dense(n_tags, activation=\"softmax\"))(model)","execution_count":43,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# COMBINING INPUT AND OUTPUT LAYERS-->"},{"metadata":{"trusted":true},"cell_type":"code","source":"model=Model(input,output)","execution_count":44,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.summary()","execution_count":45,"outputs":[{"output_type":"stream","text":"Model: \"model\"\n_________________________________________________________________\nLayer (type)                 Output Shape              Param #   \n=================================================================\ninput_1 (InputLayer)         [(None, 104)]             0         \n_________________________________________________________________\nembedding (Embedding)        (None, 104, 104)          3658512   \n_________________________________________________________________\nspatial_dropout1d (SpatialDr (None, 104, 104)          0         \n_________________________________________________________________\nbidirectional (Bidirectional (None, 104, 300)          306000    \n_________________________________________________________________\ntime_distributed (TimeDistri (None, 104, 17)           5117      \n=================================================================\nTotal params: 3,969,629\nTrainable params: 3,969,629\nNon-trainable params: 0\n_________________________________________________________________\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.compile(optimizer='adam',loss='categorical_crossentropy',metrics=['accuracy'])","execution_count":46,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"pip install livelossplot","execution_count":47,"outputs":[{"output_type":"stream","text":"Collecting livelossplot\n  Downloading livelossplot-0.5.0-py3-none-any.whl (26 kB)\nRequirement already satisfied: matplotlib; python_version >= \"3.6\" in /opt/conda/lib/python3.7/site-packages (from livelossplot) (3.1.3)\nRequirement already satisfied: ipython in /opt/conda/lib/python3.7/site-packages (from livelossplot) (7.12.0)\nRequirement already satisfied: kiwisolver>=1.0.1 in /opt/conda/lib/python3.7/site-packages (from matplotlib; python_version >= \"3.6\"->livelossplot) (1.1.0)\nRequirement already satisfied: cycler>=0.10 in /opt/conda/lib/python3.7/site-packages (from matplotlib; python_version >= \"3.6\"->livelossplot) (0.10.0)\nRequirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /opt/conda/lib/python3.7/site-packages (from matplotlib; python_version >= \"3.6\"->livelossplot) (2.4.6)\nRequirement already satisfied: numpy>=1.11 in /opt/conda/lib/python3.7/site-packages (from matplotlib; python_version >= \"3.6\"->livelossplot) (1.18.1)\nRequirement already satisfied: python-dateutil>=2.1 in /opt/conda/lib/python3.7/site-packages (from matplotlib; python_version >= \"3.6\"->livelossplot) (2.8.1)\nRequirement already satisfied: pickleshare in /opt/conda/lib/python3.7/site-packages (from ipython->livelossplot) (0.7.5)\nRequirement already satisfied: pygments in /opt/conda/lib/python3.7/site-packages (from ipython->livelossplot) (2.5.2)\nRequirement already satisfied: pexpect; sys_platform != \"win32\" in /opt/conda/lib/python3.7/site-packages (from ipython->livelossplot) (4.8.0)\nRequirement already satisfied: prompt-toolkit!=3.0.0,!=3.0.1,<3.1.0,>=2.0.0 in /opt/conda/lib/python3.7/site-packages (from ipython->livelossplot) (2.0.10)\nRequirement already satisfied: decorator in /opt/conda/lib/python3.7/site-packages (from ipython->livelossplot) (4.4.1)\nRequirement already satisfied: setuptools>=18.5 in /opt/conda/lib/python3.7/site-packages (from ipython->livelossplot) (45.2.0.post20200209)\nRequirement already satisfied: jedi>=0.10 in /opt/conda/lib/python3.7/site-packages (from ipython->livelossplot) (0.14.1)\nRequirement already satisfied: traitlets>=4.2 in /opt/conda/lib/python3.7/site-packages (from ipython->livelossplot) (4.3.3)\nRequirement already satisfied: backcall in /opt/conda/lib/python3.7/site-packages (from ipython->livelossplot) (0.1.0)\nRequirement already satisfied: six in /opt/conda/lib/python3.7/site-packages (from cycler>=0.10->matplotlib; python_version >= \"3.6\"->livelossplot) (1.14.0)\nRequirement already satisfied: ptyprocess>=0.5 in /opt/conda/lib/python3.7/site-packages (from pexpect; sys_platform != \"win32\"->ipython->livelossplot) (0.6.0)\nRequirement already satisfied: wcwidth in /opt/conda/lib/python3.7/site-packages (from prompt-toolkit!=3.0.0,!=3.0.1,<3.1.0,>=2.0.0->ipython->livelossplot) (0.1.8)\nRequirement already satisfied: parso>=0.5.0 in /opt/conda/lib/python3.7/site-packages (from jedi>=0.10->ipython->livelossplot) (0.6.1)\nRequirement already satisfied: ipython-genutils in /opt/conda/lib/python3.7/site-packages (from traitlets>=4.2->ipython->livelossplot) (0.2.0)\nInstalling collected packages: livelossplot\nSuccessfully installed livelossplot-0.5.0\nNote: you may need to restart the kernel to use updated packages.\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping\nfrom livelossplot.tf_keras import PlotLossesCallback","execution_count":48,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# FITTING OUR MODEL-->"},{"metadata":{"id":"Q9HWH06Ypkxh","outputId":"e83ba281-0c14-4bca-dacb-cb708edacba5","trusted":true},"cell_type":"code","source":"early_stop= EarlyStopping(monitor='val_accuracy',patience=1,verbose=0,mode='max',restore_best_weights=False)\ncallbacks=[PlotLossesCallback(),early_stop]\n\nhistory=model.fit(x_train,np.array(y_train),validation_split=0.2,batch_size=32,epochs=2,verbose=1,callbacks=callbacks)","execution_count":49,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 864x576 with 2 Axes>","image/png":"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\n"},"metadata":{}},{"output_type":"stream","text":"Log-loss (cost function):\ntraining   (min:    0.023, max:    0.103, cur:    0.023)\nvalidation (min:    0.025, max:    0.030, cur:    0.025)\n\naccuracy:\ntraining   (min:    0.976, max:    0.993, cur:    0.993)\nvalidation (min:    0.991, max:    0.993, cur:    0.993)\n\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r34530/34530 [==============================] - 415s 12ms/sample - loss: 0.0229 - accuracy: 0.9932 - val_loss: 0.0247 - val_accuracy: 0.9926\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.evaluate(x_test,np.array(y_test))","execution_count":50,"outputs":[{"output_type":"stream","text":"4796/4796 [==============================] - 10s 2ms/sample - loss: 0.0249 - accuracy: 0.9926\n","name":"stdout"},{"output_type":"execute_result","execution_count":50,"data":{"text/plain":"[0.024935387255153225, 0.99258596]"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"# The Accuracy turns upto 99.3%"},{"metadata":{"trusted":true},"cell_type":"code","source":"i = np.random.randint(0,x_test.shape[0])\np = model.predict(np.array([x_test[i]]))\np = np.argmax(p, axis=-1)\n\ny_true=np.argmax(np.array(y_test),axis=-1)[i]\nprint(\"{:15} ({:5}): {}\".format(\"Word\", \"True\", \"Predicted by our Model\"))\n\nprint(\"--\"*20)\n\nfor w,true,pred in zip(x_test[i],y_true,p[0]):\n    print(\"{:15}{}\\t {}\".format(words[w-1],tags[true],tags[pred]))","execution_count":51,"outputs":[{"output_type":"stream","text":"Word            (True ): Predicted by our Model\n----------------------------------------\nMusharraf      B-per\t B-per\nAzeglio        I-per\t I-per\nseventh-seeded O\t O\nsocio-economic O\t O\nDollar-priced  O\t O\nMosques        O\t O\nAleppo         O\t O\ngemstone       O\t O\nretroactively  O\t O\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"p[0]","execution_count":52,"outputs":[{"output_type":"execute_result","execution_count":52,"data":{"text/plain":"array([10, 16,  4,  4,  4,  4,  4,  4,  4,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,\n        0,  0])"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"i = np.random.randint(0,x_test.shape[0])\np = model.predict(np.array([x_test[i]]))\np = np.argmax(p, axis=-1)\n\ny_true=np.argmax(np.array(y_test),axis=-1)[i]\nprint(\"{:15} ({:5}): {}\".format(\"Word\", \"True\", \"Predicted by our Model\"))\n\nprint(\"--\"*10)\n\nfor w,true,pred in zip(x_test[i],y_true,p[0]):\n    print(\"{:15}{}\\t {}\".format(words[w-1],tags[true],tags[pred]))","execution_count":53,"outputs":[{"output_type":"stream","text":"Word            (True ): Predicted by our Model\n--------------------\nMusharraf      B-per\t B-per\napostolic      I-per\t I-per\nsack           O\t O\nplummeting     O\t O\ninclusion      B-tim\t B-tim\nBuddhist       O\t O\nstick          B-per\t O\nMassey         I-per\t I-per\nauctioneers    O\t O\nbraved         O\t O\nSangakkara     O\t O\nRafidain       O\t O\nbraved         O\t O\nGINI           O\t O\ntorn           O\t I-org\nnulcear        O\t I-org\nTony           O\t I-org\nretroactively  O\t O\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\ncrises         B-art\t B-art\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"# See our model is performing quite well"},{"metadata":{},"cell_type":"markdown","source":"# Now saving our model"},{"metadata":{"id":"6euqX7UHplG7","outputId":"7222c24c-52c5-454b-a5d4-03d6df4173f0","trusted":true},"cell_type":"code","source":"model.save_weights('final_model.h5')","execution_count":54,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}