WEEK2 : NLP in Tensorflow (Embedding)

데이터 로드

import tensorflow_datasets as tfds
import numpy as np

imdb, info = tfds.load("imdb_reviews", with_info=True, as_supervised=True)

train_data, test_data = imdb['train'], imdb['test']

training_sentences = []
training_labels = []

testing_sentences = []
testing_labels = []

for s, l in train_data:
    training_sentences.append(s.numpy().decode('utf8'))
    training_labels.append(l.numpy())

for s, l in test_data:
    testing_sentences.append(s.numpy().decode('utf8'))
    testing_labels.append(l.numpy())
    
training_labels = np.array(training_labels)
testing_labels = np.array(testing_labels)

 

tokenizing

• vocab_size, embeding_dim, max_length 조절하며 학습 진행

# hyperparameter 
vocab_size = 10000
embedding_dim = 16
max_length = 120
trunc_type = 'post'
oov_token = '<OOV>'

from tensorflow.keras.preprocessing.text import Tokenizer
from tensorflow.keras.preprocessing.sequence import pad_sequences

tokenizer = Tokenizer(num_words=vocab_size, oov_token=oov_token)
tokenizer.fit_on_texts(training_sentences)
word_index = tokenizer.word_index

training_sequences = tokenizer.texts_to_sequences(training_sentences)
training_padded = pad_sequences(training_sequences, maxlen=max_length, truncating=trunc_type)

testing_sequences = tokenizer.texts_to_sequences(testing_sentences)
testing_padded = pad_sequences(testing_sequences, maxlen=max_length, truncating=trunc_type)

 

model

import tensorflow as tf

model = tf.keras.models.Sequential([
    tf.keras.layers.Embedding(vocab_size, embedding_dim, input_length=max_length),
    #tf.keras.layers.Flatten(),
    tf.keras.layers.GlobalAveragePooling1D(),
    tf.keras.layers.Dense(6, activation='relu'),
    tf.keras.layers.Dense(1, activation='sigmoid')
    
])

 

compile

model.compile(
    optimizer='adam', 
    loss='binary_crossentropy', 
    metrics=['acc'])

 

fit

class myCallback(tf.keras.callbacks.Callback):
    def on_epoch_end(self, epoch, logs={}):
        if logs.get('acc')>0.95:
            self.model.stop_training=True
            
callbacks = myCallback()
            
history = model.fit(
    x=training_padded, 
    y=training_labels, 
    batch_size=32, 
    epochs=5, 
    validation_data=(testing_padded, testing_labels), callbacks=[callbacks])

 

+)

# embedding layer
e = model.layers[0]

weights = e.get_weights()[0]

print(weights.shape) # (vocab_size, embedding_dim)

index_word = dict([(value, key) for (key, value) in word_index.items()])
index_word[0] = '<PAD>'

def decode_review(text):
    return " ".join([index_word[i] for i in text])

print(decode_review(training_padded[3]))
print(training_sentences[3])