-
Notifications
You must be signed in to change notification settings - Fork 30
/
Copy pathbinarize_snli_dataset.py
183 lines (131 loc) · 5.59 KB
/
binarize_snli_dataset.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
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
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
import numpy as np
import pickle
import argparse
import os
import re
def str2bool(answer):
answer = answer.lower()
if answer in ['y', 'yes']:
return True
if answer in ['n', 'no']:
return False
print('Invalid answer: ' + answer)
print('Exiting..')
exit()
parser = argparse.ArgumentParser()
# Path to a folder called snli_dataset/ where all binary data will be placed.
# This folder should contain the original SNLI dataset downloaded and unzipped from https://nlp.stanford.edu/projects/snli/
parser.add_argument("--root_path", type=str, help="Root path", default='/path/to/your/snli_dataset/')
# Original SNLI files.
parser.add_argument("--train_file", type=str, help="Train path", default='snli_1.0/snli_1.0_train.txt')
parser.add_argument("--dev_file", type=str, default='snli_1.0/snli_1.0_dev.txt')
parser.add_argument("--test_file", type=str, default='snli_1.0/snli_1.0_test.txt')
parser.add_argument("--save_train_file", type=str, default='train')
parser.add_argument("--save_dev_file", type=str, default='dev')
parser.add_argument("--save_test_file", type=str, default='test')
parser.add_argument("--save_word_to_id", type=str, default='word_to_id')
parser.add_argument("--save_id_to_word", type=str, default='id_to_word')
parser.add_argument("--num_classes", type=int, default=2)
parser.add_argument("--use_vocab", type=str2bool,
help="Whether to use already stored word_to_id and id_to_word dicts",
default="no")
args = parser.parse_args()
CLASS_TO_ID = {
'contradiction': 2,
'neutral': 0,
'entailment': 1
}
if args.num_classes == 2:
CLASS_TO_ID['contradiction'] = 0
original_filepath_train = os.path.join(args.root_path, args.train_file)
original_filepath_dev = os.path.join(args.root_path, args.dev_file)
original_filepath_test = os.path.join(args.root_path, args.test_file)
suffix = '_' + str(args.num_classes) + 'class'
save_filepath_train = os.path.join(args.root_path, args.save_train_file + suffix)
save_filepath_dev = os.path.join(args.root_path, args.save_dev_file + suffix)
save_filepath_test = os.path.join(args.root_path, args.save_test_file + suffix)
save_filepath_word2id = os.path.join(args.root_path, args.save_word_to_id)
save_filepath_id2word = os.path.join(args.root_path, args.save_id_to_word)
def get_vocab(file_paths):
if args.use_vocab:
word_to_id = pickle.load(open(save_filepath_word2id, 'rb'))
id_to_word = pickle.load(open(save_filepath_id2word, 'rb'))
print('USING SAVED VOCABULARY')
return word_to_id, id_to_word
word_to_id = {}
id_to_word = []
curr_word_id = 0
word_to_id['<UNK>'] = 0
id_to_word.append('<UNK>')
curr_word_id += 1
for file_path in file_paths:
for line in open(file_path, 'r'):
parts = line.strip().lower().split('\t')
sent1 = parts[5]
sent2 = parts[6]
words_1 = re.split('[^a-zA-Z]', sent1)
words_2 = re.split('[^a-zA-Z]', sent2)
for w in words_1 + words_2:
if w not in word_to_id:
word_to_id[w] = curr_word_id
id_to_word.append(w)
curr_word_id += 1
assert len(word_to_id) == len(id_to_word)
return word_to_id, id_to_word
def transform(file_path, word_to_id):
transformed_data = []
for line in open(file_path, 'r'):
parts = line.strip().lower().split('\t')
sent1 = parts[5]
sent2 = parts[6]
label = parts[0]
if label not in CLASS_TO_ID: continue
words_1 = re.split('[^a-zA-Z]', sent1)
words_2 = re.split('[^a-zA-Z]', sent2)
word_ids_1 = []
word_ids_2 = []
for w in words_1:
word_ids_1.append(word_to_id[w])
for w in words_2:
word_ids_2.append(word_to_id[w])
# list of words for s1 , list of words for s2 , class id
transformed_data.append((word_ids_1, len(word_ids_1), word_ids_2, len(word_ids_2), CLASS_TO_ID[label]))
return transformed_data
def print_stats(data):
stats = {}
for k in CLASS_TO_ID.keys():
stats[CLASS_TO_ID[k]] = 0
for _, _, _, _, label in data:
stats[label] += 1
for k in CLASS_TO_ID.keys():
print(' --> ' + k + ' : ' + str(stats[CLASS_TO_ID[k]]))
def save_files():
word_to_id, id_to_word = get_vocab([original_filepath_train, original_filepath_dev, original_filepath_test])
train_data = transform(original_filepath_train, word_to_id)
dev_data = transform(original_filepath_dev, word_to_id)
test_data = transform(original_filepath_test, word_to_id)
print('Vocab size: %d' % len(word_to_id))
print('Training data size: %d' % len(train_data))
print_stats(train_data)
print('Dev data size: %d' % len(dev_data))
print_stats(dev_data)
print('Test data size: %d' % len(test_data))
print_stats(test_data)
print('Some samples: ')
for i in range(0, 20):
ind = np.random.randint(0, len(train_data))
for word_id in train_data[ind][0]:
print(id_to_word[word_id] + ' ' )
print('')
for word_id in train_data[ind][2]:
print(id_to_word[word_id] + ' ')
print(train_data[ind][4])
print('===================')
if not args.use_vocab:
pickle.dump(word_to_id, open(save_filepath_word2id, 'wb'))
pickle.dump(id_to_word, open(save_filepath_id2word, 'wb'))
pickle.dump(train_data, open(save_filepath_train, 'wb'))
pickle.dump(dev_data, open(save_filepath_dev, 'wb'))
pickle.dump(test_data, open(save_filepath_test, 'wb'))
if __name__ == '__main__':
save_files()