TextPrediction/example_gen_words.py at master · LeonHeTheFirst/TextPrediction · GitHub

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
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
from __future__ import print_function
from keras.models import Sequential
from keras.layers import Dense, Activation
from keras.layers import LSTM
from keras.optimizers import RMSprop
from keras.utils.data_utils import get_file
from keras.callbacks import ModelCheckpoint
import numpy as np
import random
import sys, re, os

filename = 'data_parsed/drseuss.txt'
output_filename = 'output_text/drseuss_out_epoch_1.txt'
weight_dir_name = 'weights/drseuss/words/'
weights_filename = 'weights/drseuss/weights-improvement-01-1.9294-drseuss-larger.hdf5'
# raw_text = open(filename, encoding='utf-8', errors='ignore').read().lower()

# load ascii text and covert to lowercase
filename = 'data_parsed/drseuss.txt'
raw_text = open(filename, encoding='utf8', errors='ignore').read()
raw_text = raw_text.lower()
# create mapping of unique chars to integers
wordList = re.sub("[^\w]", " ",  raw_text).split()
# print(len(wordList))
wordList = [w for w in wordList if re.match("^[a-z]*$", w)]
# print(len(wordList))
words = sorted(list(set(wordList)))
word_to_int = dict((w, i) for i, w in enumerate(words))

# text = text.lower()

print('corpus length:', len(raw_text))

# chars = sorted(list(set(text)))
print('total words:', len(words))
char_indices = dict((c, i) for i, c in enumerate(words))
indices_char = dict((i, c) for i, c in enumerate(words))

# print(char_indices)

# cut the text in semi-redundant sequences of maxlen characters
maxlen = 10
step = 3
sentences = []
next_words = []
for i in range(0, len(wordList) - maxlen, step):
    sentences.append(wordList[i: i + maxlen])
    next_words.append(wordList[i + maxlen])
print('nb sequences:', len(sentences))

print('Vectorization...')
x = np.zeros((len(sentences), maxlen, len(words)), dtype=np.bool)
y = np.zeros((len(sentences), len(words)), dtype=np.bool)
for i, sentence in enumerate(sentences):
    for t, char in enumerate(sentence):
        x[i, t, char_indices[char]] = 1
    y[i, char_indices[next_words[i]]] = 1


# build the model: a single LSTM
print('Build model...')
model = Sequential()
model.add(LSTM(128, input_shape=(maxlen, len(words))))
# model.add(LSTM(128, input_shape=(maxlen, len(words)), return_sequences=True))
# model.add(LSTM(128, return_sequences=True))
# model.add(LSTM(128, return_sequences=True))
# model.add(LSTM(128))
model.add(Dense(len(words)))
model.add(Activation('softmax'))

optimizer = RMSprop(lr=0.01)
# model.compile(loss='categorical_crossentropy', optimizer=optimizer)


def sample(preds, temperature=1.0):
    # helper function to sample an index from a probability array
    preds = np.asarray(preds).astype('float64')
    preds = np.log(preds) / temperature
    exp_preds = np.exp(preds)
    preds = exp_preds / np.sum(exp_preds)
    probas = np.random.multinomial(1, preds, 1)
    return np.argmax(probas)

print('-' * 50)

files = []
for file in os.listdir(weight_dir_name):
    if file.endswith(".hdf5"):
        files.append(file)

# outfile = open(output_filename, 'w')
diversity = 0.5

start_index = random.randint(0, len(wordList) - maxlen - 1)
generated = ''
sentence = wordList[start_index: start_index + maxlen]
for word in sentence:
	generated += word + ' '
print('----- Generating with seed: "' + ' '.join(word for word in sentence) + '"')
sys.stdout.write(generated)
# outfile.write(generated)

for weight_file in files:

    model.load_weights(weight_dir_name + weight_file)
    model.compile(loss='categorical_crossentropy', optimizer=optimizer)

    weight_file.replace('.', '-')
    substrings = weight_file.split('-')
    epoch_number = substrings[2]
    source = substrings[4]
    new_output_filename = 'output_text/' + 'drseuss' + '_epoch_' + epoch_number + '.txt'
    new_output_file = open(new_output_filename, 'w')

    for i in range(1000):
        x_pred = np.zeros((1, maxlen, len(words)))
        for t, char in enumerate(sentence):
            x_pred[0, t, char_indices[char]] = 1.

        preds = model.predict(x_pred, verbose=0)[0]
        next_index = sample(preds, diversity)
        next_word = indices_char[next_index]

        generated += next_word
        sentence = sentence[1:] + [next_word]

        sys.stdout.write(next_word + ' ')
        new_output_file.write(next_word + ' ')
    print()
    new_output_file.close()