In [1]:

    

'''Trains a LSTM on the IMDB sentiment classification task.
The dataset is actually too small for LSTM to be of any advantage
compared to simpler, much faster methods such as TF-IDF + LogReg.
Notes:
- RNNs are tricky. Choice of batch size is important,
choice of loss and optimizer is critical, etc.
Some configurations won't converge.
- LSTM loss decrease patterns during training can be quite different
from what you see with CNNs/MLPs/etc.
'''
from __future__ import print_function

from keras.preprocessing import sequence
from keras.models import Sequential
from keras.layers import Dense, Embedding
from keras.layers import LSTM
from keras.datasets import imdb

max_features = 20000
maxlen = 80  # cut texts after this number of words (among top max_features most common words)
batch_size = 32

print('Loading data...')
(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=max_features)
print(len(x_train), 'train sequences')
print(len(x_test), 'test sequences')

print('Pad sequences (samples x time)')
x_train = sequence.pad_sequences(x_train, maxlen=maxlen)
x_test = sequence.pad_sequences(x_test, maxlen=maxlen)
print('x_train shape:', x_train.shape)
print('x_test shape:', x_test.shape)

print('Build model...')
model = Sequential()
model.add(Embedding(max_features, 128))
model.add(LSTM(128, dropout=0.2, recurrent_dropout=0.2))
model.add(Dense(1, activation='sigmoid'))

# try using different optimizers and different optimizer configs
model.compile(loss='binary_crossentropy',
              optimizer='adam',
              metrics=['accuracy'])

print('Train...')
model.fit(x_train, y_train,
          batch_size=batch_size,
          epochs=15,
          validation_data=(x_test, y_test))
score, acc = model.evaluate(x_test, y_test,
                            batch_size=batch_size)
print('Test score:', score)
print('Test accuracy:', acc)


    

    




Using TensorFlow backend.






    




Loading data...
Downloading data from https://s3.amazonaws.com/text-datasets/imdb.npz
25000 train sequences
25000 test sequences
Pad sequences (samples x time)
x_train shape: (25000, 80)
x_test shape: (25000, 80)
Build model...
Train...
Train on 25000 samples, validate on 25000 samples
Epoch 1/15
25000/25000 [==============================] - 181s - loss: 0.4552 - acc: 0.7838 - val_loss: 0.3707 - val_acc: 0.8392
Epoch 2/15
25000/25000 [==============================] - 179s - loss: 0.2964 - acc: 0.8801 - val_loss: 0.3803 - val_acc: 0.8304
Epoch 3/15
25000/25000 [==============================] - 179s - loss: 0.2121 - acc: 0.9182 - val_loss: 0.4190 - val_acc: 0.8274
Epoch 4/15
25000/25000 [==============================] - 179s - loss: 0.1486 - acc: 0.9447 - val_loss: 0.5269 - val_acc: 0.8250
Epoch 5/15
25000/25000 [==============================] - 178s - loss: 0.1022 - acc: 0.9626 - val_loss: 0.5790 - val_acc: 0.8206
Epoch 6/15
25000/25000 [==============================] - 178s - loss: 0.0728 - acc: 0.9739 - val_loss: 0.6543 - val_acc: 0.8205
Epoch 7/15
25000/25000 [==============================] - 179s - loss: 0.0547 - acc: 0.9815 - val_loss: 0.7418 - val_acc: 0.8146
Epoch 8/15
25000/25000 [==============================] - 178s - loss: 0.0403 - acc: 0.9874 - val_loss: 0.8405 - val_acc: 0.8149
Epoch 9/15
25000/25000 [==============================] - 178s - loss: 0.0359 - acc: 0.9890 - val_loss: 0.8116 - val_acc: 0.8108
Epoch 10/15
25000/25000 [==============================] - 178s - loss: 0.0202 - acc: 0.9935 - val_loss: 1.0708 - val_acc: 0.8098
Epoch 11/15
25000/25000 [==============================] - 178s - loss: 0.0250 - acc: 0.9922 - val_loss: 0.8883 - val_acc: 0.8074
Epoch 12/15
25000/25000 [==============================] - 178s - loss: 0.0187 - acc: 0.9941 - val_loss: 1.0817 - val_acc: 0.8087
Epoch 13/15
25000/25000 [==============================] - 178s - loss: 0.0128 - acc: 0.9956 - val_loss: 1.0535 - val_acc: 0.8121
Epoch 14/15
25000/25000 [==============================] - 178s - loss: 0.0131 - acc: 0.9960 - val_loss: 1.0574 - val_acc: 0.8112
Epoch 15/15
25000/25000 [==============================] - 178s - loss: 0.0102 - acc: 0.9971 - val_loss: 1.1064 - val_acc: 0.8110
24992/25000 [============================>.] - ETA: 0sTest score: 1.10635716881
Test accuracy: 0.81104

In [ ]: