-
Notifications
You must be signed in to change notification settings - Fork 1
/
helpers.py
267 lines (224 loc) · 10.7 KB
/
helpers.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
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
from collections import Counter
from torch.autograd import Variable
import torch
import numpy as np
import random
from torch.utils.data.dataset import Dataset
from torch.utils.data import DataLoader
from torch.utils.data.sampler import WeightedRandomSampler
def add_noise(l):
words = l.split()
length = len(words)
i = 0
new_words = []
while i < length:
a = random.random()
w = words[i]
if a <= 0.1:
i += 1
continue
elif a <= 0.2:
new_words.append(w)
new_words.append(w)
i += 1
elif a <= 0.3 and i != length - 1:
new_words.append(words[i + 1])
new_words.append(w)
i += 2
else:
new_words.append(w)
i += 1
return ' '.join(new_words)
class MyDataset(Dataset):
def __init__(self, file_path, tag, word2idx, noise=False, debug=False):
seqs = open(file_path, "r", encoding="utf-8").readlines()
seqs = list(map(lambda line: line.strip(), seqs))
if tag == 1:
labels = list(np.ones((len(seqs),)))
else:
labels = list(np.zeros((len(seqs),)))
self.seqs = seqs
self.labels = labels
self.num_total_seqs = len(self.seqs)
self.word2idx = word2idx
self.noise = noise
if debug:
self.seqs = self.seqs[:100]
self.labels = self.labels[:100]
self.num_total_seqs = len(self.seqs)
def __getitem__(self, index):
seq = self.seqs[index]
if self.noise:
seq = add_noise(seq)
label = self.labels[index]
seq = self.words2ids(seq)
return seq, label
def __len__(self):
return self.num_total_seqs
def words2ids(self, sentence):
tokens = sentence.lower().split()
sequence = []
sequence.append(self.word2idx['<sos>'])
for token in tokens:
if token in self.word2idx:
sequence.append(self.word2idx[token])
else:
sequence.append(self.word2idx['<unk>'])
sequence.append(self.word2idx['<eos>'])
sequence = torch.LongTensor(sequence)
return sequence
class DisDataset(Dataset):
def __init__(self, data_path, label_path, word2idx, debug=False):
seqs = open(data_path, "r", encoding="utf-8").readlines()
labels = [int(l) for l in open(label_path, "r", encoding="utf-8").read().split('\n')]
self.ls = [int(l) for l in open(label_path, "r", encoding="utf-8").read().split('\n')]
seqs = list(map(lambda line: line.strip(), seqs))
self.seqs = seqs
self.labels = labels
self.num_total_seqs = len(self.seqs)
self.word2idx = word2idx
if debug:
self.seqs = self.seqs[:100]
self.labels = self.labels[:100]
self.num_total_seqs = len(self.seqs)
def __getitem__(self, index):
seq = self.seqs[index]
label = self.labels[index]
seq = self.words2ids(seq)
return seq, label
def __len__(self):
return self.num_total_seqs
def words2ids(self, sentence):
tokens = sentence.lower().split()
sequence = []
sequence.append(self.word2idx['<sos>'])
for token in tokens:
if token in self.word2idx:
sequence.append(self.word2idx[token])
else:
sequence.append(self.word2idx['<unk>'])
sequence.append(self.word2idx['<eos>'])
sequence = torch.LongTensor(sequence)
return sequence
def make_vocab(hps):
words = []
lines = open(hps.dis_train_data_path, 'r').read().split('\n')
lines += open(hps.dis_dev_data_path, 'r').read().split('\n')
lines += open(hps.irony_path, 'r').read().split('\n')
lines += open(hps.non_path, 'r').read().split('\n')
lines += open(hps.senti_train_data_path, 'r').read().split('\n')
lines += open(hps.senti_dev_data_path, 'r').read().split('\n')
lines += open(hps.non_senti_train_data_path, 'r').read().split('\n')
lines += open(hps.non_senti_dev_data_path, 'r').read().split('\n')
for l in lines:
words += l.split()
c = Counter(words)
top_k_words = sorted(c.keys(), reverse=True, key=c.get)#[:hps.vocab_size - 4]
words = ['<pad>', '<unk>', '<sos>', '<eos>'] + [w for w in top_k_words if c[w] > 2]
print('vocab size {}'.format(len(words)))
vocab = {w:i for i, w in enumerate(words)}
idx2word = {i:w for i, w in enumerate(words)}
open(hps.test_path + hps.vocab_path, 'w', encoding='utf-8').write('\n'.join([w + '\t' + str(vocab[w]) for w in vocab]))
return vocab, idx2word
def make_weights_for_balanced_classes(samples, nclasses, PosOverNeg=1):
count = [0] * nclasses
for item in samples:
count[item] += 1
weight_per_class = [0.] * nclasses
N = float(sum(count))
for i in range(nclasses):
weight_per_class[i] = N/float(count[i])
weight_per_class[1] *= PosOverNeg
weight = [0] * len(samples)
for idx, val in enumerate(samples):
weight[idx] = weight_per_class[val]
return weight
def collate_fn(data):
def merge(sequences):
lengths = torch.LongTensor([len(seq) for seq in sequences])
padded_seq = torch.zeros(len(sequences), 40, dtype=torch.long)
for i, seq in enumerate(sequences):
end = min(40, lengths[i])
padded_seq[i, :end] = seq[:end]
return padded_seq
seqs, labels = zip(*data) # tuples
seqs = merge(seqs)
labels = torch.LongTensor(list(labels))
return seqs, labels
def prepare_non_senti_data(hps, vocab):
print('preparing non senti data...')
dataset = DisDataset(hps.non_senti_train_data_path, hps.non_senti_train_label_path, vocab, debug=False)
weights = make_weights_for_balanced_classes(dataset.ls, 2, PosOverNeg=1)
sampler = WeightedRandomSampler(weights, len(weights))
train_data_loader = DataLoader(dataset,\
batch_size=hps.non_senti_batch_size,\
shuffle=False,\
collate_fn=collate_fn, drop_last=False, sampler=sampler)
dataset = DisDataset(hps.non_senti_dev_data_path, hps.non_senti_dev_label_path, vocab, debug=False)
weights = make_weights_for_balanced_classes(dataset.ls, 2, PosOverNeg=1)
sampler = WeightedRandomSampler(weights, len(weights))
dev_data_loader = DataLoader(dataset,\
batch_size=hps.non_senti_batch_size,\
shuffle=False,\
collate_fn=collate_fn, drop_last=False, sampler=sampler)
return train_data_loader, dev_data_loader
def prepare_senti_data(hps, vocab):
print('preparing senti data...')
dataset = DisDataset(hps.senti_train_data_path, hps.senti_train_label_path, vocab, debug=False)
weights = make_weights_for_balanced_classes(dataset.ls, 2, PosOverNeg=1)
sampler = WeightedRandomSampler(weights, len(weights))
train_data_loader = DataLoader(dataset,\
batch_size=hps.senti_batch_size,\
shuffle=False,\
collate_fn=collate_fn, drop_last=False, sampler=sampler)
dataset = DisDataset(hps.senti_dev_data_path, hps.senti_dev_label_path, vocab, debug=False)
weights = make_weights_for_balanced_classes(dataset.ls, 2, PosOverNeg=1)
sampler = WeightedRandomSampler(weights, len(weights))
dev_data_loader = DataLoader(dataset,\
batch_size=hps.senti_batch_size,\
shuffle=False,\
collate_fn=collate_fn, drop_last=False, sampler=sampler)
return train_data_loader, dev_data_loader
def prepare_discriminator_data(hps, vocab):
print('preparing dis data...')
train_dataset = DisDataset(hps.dis_train_data_path, hps.dis_train_label_path, vocab, debug=False)
train_weights = make_weights_for_balanced_classes(train_dataset.ls, 2, PosOverNeg=1)
train_sampler = WeightedRandomSampler(train_weights, len(train_weights))
train_data_loader = DataLoader(train_dataset,\
batch_size=hps.dis_batch_size,\
shuffle=False,\
collate_fn=collate_fn, drop_last=False, sampler=train_sampler)
dev_dataset = DisDataset(hps.dis_dev_data_path, hps.dis_dev_label_path, vocab, debug=False)
dev_weights = make_weights_for_balanced_classes(dev_dataset.ls, 2, PosOverNeg=1)
dev_sampler = WeightedRandomSampler(dev_weights, len(dev_weights))
dev_data_loader = DataLoader(dev_dataset,\
batch_size=hps.dis_batch_size,\
shuffle=False,\
collate_fn=collate_fn, drop_last=False, sampler=dev_sampler)
return train_data_loader, dev_data_loader
def prepare_seq2seq_data(hps,vocab, noise=False):
print('preparing seq data...')
irony_dataset = MyDataset(hps.irony_path, 1, vocab, noise, debug=False)
irony_data_loader = DataLoader(irony_dataset,\
batch_size=hps.batch_size,\
shuffle=True,\
collate_fn=collate_fn, drop_last=True)
non_dataset = MyDataset(hps.non_path, 0, vocab, noise, debug=False)
non_data_loader = DataLoader(non_dataset,\
batch_size=hps.batch_size,\
shuffle=True,\
collate_fn=collate_fn, drop_last=True)
return irony_data_loader, non_data_loader
def prepare_test_data(hps, vocab):
print('preparing test data...')
non_dataset = MyDataset(hps.non_test_path, 0, vocab, False, debug=False)
non_data_loader = DataLoader(non_dataset,\
batch_size=hps.test_batch_size,\
shuffle=False,\
collate_fn=collate_fn, drop_last=False)
irony_dataset = MyDataset(hps.irony_test_path, 1, vocab, False, debug=False)
irony_data_loader = DataLoader(irony_dataset,\
batch_size=hps.test_batch_size,\
shuffle=False,\
collate_fn=collate_fn, drop_last=False)
return irony_data_loader, non_data_loader