-
Notifications
You must be signed in to change notification settings - Fork 0
/
create_pretraining_data.py
430 lines (341 loc) · 15.8 KB
/
create_pretraining_data.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
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
# coding=utf-8
'''
@Software:PyCharm
@Time:2024/04/07 2:58 下午
@Author: fffan
'''
import os
import random
import torch
import pickle
import argparse
import collections
import click
from tqdm import tqdm
from model_info.tokenization import BertTokenizer
from torch.utils.data import Dataset
from loguru import logger
class TrainingInstance(object):
""" 句子对形式的单个训练数据实例类型 """
def __init__(self, tokens, segment_ids, masked_lm_positions, masked_lm_labels,
is_random_next):
self.tokens = tokens
self.segment_ids = segment_ids
self.masked_lm_positions = masked_lm_positions
self.masked_lm_labels = masked_lm_labels
self.is_random_next = is_random_next
def __str__(self):
s = ""
s += "tokens: %s\n" % (" ".join([x for x in self.tokens]))
s += "segment_ids: %s\n" % (" ".join([str(x) for x in self.segment_ids]))
s += "is_random_next: %s\n" % self.is_random_next
s += "masked_lm_positions: %s\n" % (" ".join([str(x) for x in self.masked_lm_positions]))
s += "masked_lm_labels: %s\n" % (" ".join([x for x in self.masked_lm_labels]))
s += "\n"
return s
def __repr__(self):
return self.__str__()
class PreTrainingDataset(Dataset):
""" 实际的供给 Dataloader 的数据类 """
def __init__(self):
self.data = []
def add_instance(self, features: collections.OrderedDict):
self.data.append((
features["input_ids"],
features["segment_ids"],
features["input_mask"],
features["masked_lm_ids"],
features["next_sentence_labels"]
))
def __len__(self):
return len(self.data)
def __getitem__(self, index):
input_ids, token_type_ids, attention_mask, masked_lm_labels, next_sentence_label = self.data[index]
return input_ids, token_type_ids, attention_mask, masked_lm_labels, next_sentence_label
def write_instance_to_example_files(instances, tokenizer, max_seq_length, output_file):
""" 以 TrainingInstance 创造 Dataset 训练实例"""
eval_set = PreTrainingDataset()
train_set = PreTrainingDataset()
total_written = 0
for (inst_index, instance) in enumerate(instances):
input_ids = tokenizer.convert_tokens_to_ids(instance.tokens)
input_ids_copy_for_replace = list(input_ids) # 用于替换 mask 原字符得到 label
input_mask = [1] * len(input_ids)
segment_ids = list(instance.segment_ids)
assert len(input_ids) <= max_seq_length
while len(input_ids) < max_seq_length:
input_ids.append(0)
input_mask.append(0)
segment_ids.append(0)
assert len(input_ids) == max_seq_length
assert len(input_mask) == max_seq_length
assert len(segment_ids) == max_seq_length
masked_lm_positions = list(instance.masked_lm_positions)
masked_lm_ids = tokenizer.convert_tokens_to_ids(instance.masked_lm_labels)
# masked_lm_weights = [1.0] * len(masked_lm_ids)
# 替换以及补齐长度
for (p, m) in zip(masked_lm_positions, masked_lm_ids):
input_ids_copy_for_replace[p] = m
while len(input_ids_copy_for_replace) < max_seq_length:
input_ids_copy_for_replace.append(-1)
# while len(masked_lm_positions) < max_predictions_per_seq:
# masked_lm_positions.append(0)
# masked_lm_ids.append(0)
# masked_lm_weights.append(0.0)
# 这里注意:按照原文思路,随机下一句的 label 为 1,而真实下一句的 label 为 0 。
next_sentence_label = 1 if instance.is_random_next else 0
features = collections.OrderedDict()
features["input_ids"] = create_long_tensor(input_ids)
features["segment_ids"] = create_long_tensor(segment_ids)
features["input_mask"] = create_float_tensor(input_mask)
features["masked_lm_ids"] = create_long_tensor(input_ids_copy_for_replace)
features["next_sentence_labels"] = create_long_tensor([next_sentence_label])
if total_written < len(instances) * 0.10:
eval_set.add_instance(features)
else:
train_set.add_instance(features)
total_written += 1
# if inst_index < 20:
# logger.info("*** Example ***")
# logger.info("tokens: %s" % " ".join([x for x in instance.tokens]))
# for feature_name in features.keys():
# feature = features[feature_name]
# values = []
# if feature.int64_list.value:
# values = feature.int64_list.value
# elif feature.float_list.value:
# values = feature.float_list.value
# logger.info("%s: %s" % (feature_name, " ".join([str(x) for x in values])))
with open(output_file + '.eval', 'wb') as f:
pickle.dump(eval_set, f)
with open(output_file + '.train', 'wb') as f:
pickle.dump(train_set, f)
logger.info("Wrote %d total instances" % total_written)
def create_long_tensor(values):
tensor = torch.LongTensor(list(values))
return tensor
def create_float_tensor(values):
tensor = torch.FloatTensor(list(values))
return tensor
def create_training_instances(input_files, tokenizer, max_seq_length,
dupe_factor, short_seq_prob, masked_lm_prob,
max_predictions_per_seq, rng):
""" 从原始语料生成 TrainingInstance 类实例"""
all_documents = [[]]
# 输入语料的数据格式:
# (1) 每行是一整句完整句子。
# (2) 每个段落之间用一个空行隔开。
for input_file in input_files:
with open(input_file, "r") as reader:
while True:
# TODO 这里没有实现判断是否为 unicode 编码
line = reader.readline()
if not line:
break
line = line.strip()
# 如果是个空行,则表示新的段落开始。
if not line:
all_documents.append([])
tokens = tokenizer.tokenize(line)
if tokens:
all_documents[-1].append(tokens)
# 去除空段落
all_documents = [x for x in all_documents if x]
rng.shuffle(all_documents)
vocab_words = list(tokenizer.vocab.keys())
instances = []
for _ in range(dupe_factor):
for document_index in range(len(all_documents)):
instances.extend(
create_instances_from_document(
all_documents, document_index, max_seq_length, short_seq_prob,
masked_lm_prob, max_predictions_per_seq, vocab_words, rng))
rng.shuffle(instances)
return instances
def create_instances_from_document(
all_documents, document_index, max_seq_length, short_seq_prob,
masked_lm_prob, max_predictions_per_seq, vocab_words, rng):
""" 从一个文本段落生成 TrainingInstances 类型实例"""
document = all_documents[document_index]
# 因为句子包含 [CLS], [SEP], [SEP] 所以长度减三
max_num_tokens = max_seq_length - 3
# 有概率在满足 max_seq_length 的前提下随机决定新的句子总长,
# 以增加数据的任意性。
target_seq_length = max_num_tokens
if rng.random() < short_seq_prob:
target_seq_length = rng.randint(2, max_num_tokens)
# 这里采取的策略是先预选中最大长度的几个句子片段,然后将片段以完整句子为单位随机分为 A B 两部分,
# 如果是进入了随机下一句的 case ,则在选完 A 中的句子后,下一句从其他段落中随机挑选句子填满该 instance。
instances = []
current_chunk = []
current_length = 0
i = 0
#while i < len(document):
for i in tqdm(range(len(document))):
segment = document[i]
current_chunk.append(segment)
current_length += len(segment)
if i == len(document) - 1 or current_length >= target_seq_length:
if current_chunk:
# a_end 代表从 current_chunk 中选择多少个句子放入 A 片段
# A 代表第一句话
a_end = 1
if len(current_chunk) >= 2:
a_end = rng.randint(1, len(current_chunk) - 1)
tokens_a = []
for j in range(a_end):
tokens_a.extend(current_chunk[j])
tokens_b = []
# 随机的 “下一句话”
if len(current_chunk) == 1 or rng.random() < 0.5:
is_random_next = True
target_b_length = target_seq_length - len(tokens_a)
# 随机选取一个其他的段落以摘取一个“任意的”下一句话
random_document_index = rng.randint(0, len(all_documents) - 1)
for _ in range(10):
if random_document_index != document_index:
break
random_document_index = rng.randint(0, len(all_documents) - 1)
random_document = all_documents[random_document_index]
random_start = rng.randint(0, len(random_document) - 1)
for j in range(random_start, len(random_document)):
tokens_b.extend(random_document[j])
if len(tokens_b) >= target_b_length:
break
# 这里由于采取了随机的下一句话,所以原本想当作第二句话的句子都可以放回去等待下一轮使用。
num_unused_segments = len(current_chunk) - a_end
i -= num_unused_segments
# 真实的 “下一句话”
else:
is_random_next = False
for j in range(a_end, len(current_chunk)):
tokens_b.extend(current_chunk[j])
truncate_seq_pair(tokens_a, tokens_b, max_num_tokens, rng)
assert len(tokens_a) >= 1
assert len(tokens_b) >= 1
tokens = []
segment_ids = []
tokens.append("[CLS]")
segment_ids.append(0)
for token in tokens_a:
tokens.append(token)
segment_ids.append(0)
tokens.append("[SEP]")
segment_ids.append(0)
for token in tokens_b:
tokens.append(token)
segment_ids.append(1)
tokens.append("[SEP]")
segment_ids.append(1)
(tokens, masked_lm_positions, masked_lm_labels) = create_masked_lm_predictions(
tokens, masked_lm_prob, max_predictions_per_seq, vocab_words, rng)
instance = TrainingInstance(
tokens=tokens,
segment_ids=segment_ids,
masked_lm_positions=masked_lm_positions,
masked_lm_labels=masked_lm_labels,
is_random_next=is_random_next)
instances.append(instance)
current_chunk = []
current_length = 0
i += 1
return instances
MaskedLmInstance = collections.namedtuple("MaskedLmInstance", ["index", "label"])
def create_masked_lm_predictions(tokens, masked_lm_prob,
max_predictions_per_seq, vocab_words, rng):
""" 按照 masked LM 的设定生成输入向量 """
cand_indexes = []
for (i, token) in enumerate(tokens):
if token == "[CLS]" or token == "[SEP]":
continue
cand_indexes.append([i])
# 先打乱,后面直接去前百分之几做[MASK]
rng.shuffle(cand_indexes)
output_tokens = list(tokens)
num_to_predict = min(max_predictions_per_seq,
max(1, int(round(len(tokens) * masked_lm_prob))))
masked_lms = []
covered_indexes = set()
for index_set in cand_indexes:
if len(masked_lms) >= num_to_predict:
break
is_any_index_covered = False
for index in index_set:
if index in covered_indexes:
is_any_index_covered = True
break
if is_any_index_covered:
continue
for index in index_set:
covered_indexes.add(index)
# 80% 的概率替换为 [MASK]
if rng.random() < 0.8:
masked_token = "[MASK]"
else:
# 10% 的概率保持原始字符
if rng.random() < 0.5:
masked_token = tokens[index]
# 10% 的概率替换为随机字符
else:
masked_token = vocab_words[rng.randint(0, len(vocab_words) - 1)]
output_tokens[index] = masked_token
masked_lms.append(MaskedLmInstance(index=index, label=tokens[index]))
assert len(masked_lms) <= num_to_predict
masked_lms = sorted(masked_lms, key=lambda x: x.index)
masked_lm_positions = []
masked_lm_labels = []
for p in masked_lms:
masked_lm_positions.append(p.index)
masked_lm_labels.append(p.label)
return output_tokens, masked_lm_positions, masked_lm_labels
def truncate_seq_pair(tokens_a, tokens_b, max_num_tokens, rng):
""" 将两个句子修剪至总长度小于等于设定的最大长度 """
while True:
total_length = len(tokens_a) + len(tokens_b)
if total_length <= max_num_tokens:
break
trunc_tokens = tokens_a if len(tokens_a) > len(tokens_b) else tokens_b
assert len(trunc_tokens) >= 1
# 以二分之一的概率随机选择从句首或句尾截短当前句子
if rng.random() < 0.5:
del trunc_tokens[0]
else:
trunc_tokens.pop()
def get_parser():
parser = argparse.ArgumentParser()
parser.add_argument("--input_file",
default="/data1/fffan/0_data/0_original_data/3_NLP相关数据/0_data_wudao/wudao_data_3B_test.txt",
type=str)
# Required parameters
parser.add_argument("--output_file", default="./output_dir/bert_chinese_fffan_data", type=str)
parser.add_argument("--model_path", default="./pretrain_models/bert-base-chinese", type=str)
# Other parameters
parser.add_argument("--max_seq_length", default=128, type=int, help="")
parser.add_argument("--max_predictions_per_seq", default=20, type=int, help=" ")
parser.add_argument("--random_seed", default=20, type=int,help=" ")
parser.add_argument("--dupe_factor", default=10, type=int, help=" ")
parser.add_argument("--masked_lm_prob", default=0.15, type=float, help=" ")
parser.add_argument("--short_seq_prob", default=0.1, type=float, help=" ")
args = parser.parse_args()
return args
def main():
#####
args = get_parser()
os.makedirs(args.output_file, exist_ok=True)
args.output_file = os.path.join(args.output_file,"data")
logger.info("*** Loading the tokenizer ***")
tokenizer = BertTokenizer.from_pretrained(args.model_path)
# TODO 这里没有实现原始的给定文件 pattern 来批量匹配文件
input_files = args.input_file.split(",")
logger.info("*** Reading from input files ***")
for args.input_file in input_files:
logger.info(" %s" % args.input_file)
rng = random.Random(args.random_seed)
instances = create_training_instances(
input_files, tokenizer, args.max_seq_length, args.dupe_factor,
args.short_seq_prob, args.masked_lm_prob, args.max_predictions_per_seq, rng)
logger.info("*** Writing to output file ***")
logger.info(" %s.train %s.eval" % (args.output_file, args.output_file))
write_instance_to_example_files(instances, tokenizer, args.max_seq_length, args.output_file)
if __name__ == "__main__":
main()