-
Notifications
You must be signed in to change notification settings - Fork 0
/
keras_example.py
74 lines (61 loc) · 2.22 KB
/
keras_example.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
"""Simple Keras RNN example.
Builds, trains, and tests an RNN discriminator that distinguishes between
English and German/French.
"""
import warnings
with warnings.catch_warnings():
warnings.simplefilter('ignore', RuntimeWarning)
warnings.simplefilter('ignore', FutureWarning)
from tensorflow import keras
import text
def keras_example(args):
"""Build, train, and test the discriminator using the Keras frontend."""
# Get one-hot encoded English and German/French words.
x, y = text.get_data(args.length, args.language)
implementation = 2 if args.gpu else 0, # Optimize matrix sizes.
# The Keras neural network model.
discriminator = keras.models.Sequential()
# The first LSTM layer.
discriminator.add(
keras.layers.LSTM(
args.n_state,
return_sequences=True, # Keep outputs from all time steps.
input_shape=(x.shape[1], x.shape[2]),
implementation=implementation,
recurrent_activation='sigmoid',
unroll=True # Faster performance but more memory consumption.
)
)
#The second LSTM layer. Only keep the output from the final time step.
discriminator.add(
keras.layers.LSTM(
args.n_state,
return_sequences=False,
implementation=implementation,
recurrent_activation='sigmoid',
unroll=True
)
)
# Dense layer.
discriminator.add(keras.layers.Dense(1, activation='sigmoid'))
# Compile the Keras model.
discriminator.summary()
discriminator.compile('adam', 'binary_crossentropy', metrics=['accuracy'])
# Train the Keras model.
discriminator.fit(
x,
y,
args.batch_size,
args.n_epochs,
validation_split=args.validation_split
)
# Save the model.
discriminator.save('keras_example.h5')
# Label test words.
test_words = text.get_test_data(args.language)
words_encoded = text.one_hot(test_words, args.length)
test_labels = discriminator.predict(words_encoded)
# Print predictions.
print('\nWord: P({})'.format(args.language.capitalize()))
for word, label in zip(test_words, test_labels):
print('{}: {:.3f}'.format(word, float(label)))