-
Notifications
You must be signed in to change notification settings - Fork 0
/
data_loader.py
572 lines (489 loc) · 22.4 KB
/
data_loader.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
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
import pickle
from torch.utils.data import Dataset,DataLoader
import torch
import pandas as pd
import random
import os
import jieba
import re
import json
from tqdm import tqdm
from transformers import BertTokenizer, AutoTokenizer
import torch.nn as nn
from tokenizer import MyTokenizer
RANDOM_SEED = 16
torch.manual_seed(RANDOM_SEED)
def text_cleaner(text):
def load_stopwords(filename):
stopwords = []
with open(filename, "r", encoding="utf-8") as fr:
for line in fr:
line = line.replace("\n", "")
stopwords.append(line)
return stopwords
stop_words = load_stopwords("data/stopword.txt")
rules = [
{r'>\s+': u'>'}, # remove spaces after a tag opens or closes
{r'\s+': u' '}, # replace consecutive spaces
{r'\s*<br\s*/?>\s*': u'\n'}, # newline after a <br>
# newline after </p> and </div> and <h1/>...
{r'</(div)\s*>\s*': u'\n'},
# newline after </p> and </div> and <h1/>...
{r'</(p|h\d)\s*>\s*': u'\n\n'},
{r'<head>.*<\s*(/head|body)[^>]*>': u''}, # remove <head> to </head>
# show links instead of texts
{r'<a\s+href="([^"]+)"[^>]*>.*</a>': r'\1'},
{r'[ \t]*<[^<]*?/?>': u''}, # remove remaining tags
{r'^\s+': u''} # remove spaces at the beginning
]
# 替换html特殊字符
text = text.replace("“", "“").replace("”", "”")
text = text.replace(""", "\"").replace("×", "x")
text = text.replace(">", ">").replace("<", "<").replace("³", "")
text = text.replace("÷", "/").replace("…", "...")
text = text.replace("«", "《").replace("»", "》")
text = text.replace("‘", "‘").replace("’", '’')
text = text.replace(">；", ">").replace(
"<；", "<").replace("·", "")
text = text.replace("—", "—").replace("’", '’')
for rule in rules:
for (k, v) in rule.items():
regex = re.compile(k)
text = regex.sub(v, text)
text = text.rstrip()
text = text.strip()
text = text.replace('+', ' ').replace(',', ' ').replace(':', ' ')
text = re.sub("([0-9]+[年月日])+", "", text)
text = re.sub("[a-zA-Z]+", "", text)
text = re.sub("[0-9\.]+元", "", text)
stop_words_user = ["年", "月", "日", "时", "分", "许", "某", "甲", "乙", "丙"]
word_tokens = jieba.cut(text)
def str_find_list(string, words):
for word in words:
if string.find(word) != -1:
return True
return False
text = [w for w in word_tokens if w not in stop_words if not str_find_list(w, stop_words_user)
if len(w) >= 1 if not w.isspace()]
return " ".join(text)
#只清洗数据，不去停用词。
def text_cleaner2(text):
# 替换html特殊字符
text = text.replace("“", "“").replace("”", "”")
text = text.replace(""", "\"").replace("×", "x")
text = text.replace(">", ">").replace("<", "<").replace("³", "")
text = text.replace("÷", "/").replace("…", "...")
text = text.replace("«", "《").replace("»", "》")
text = text.replace("‘", "‘").replace("’", '’')
text = text.replace(">；", ">").replace(
"<；", "<").replace("·", "")
text = text.replace("—", "—").replace("’", '’')
# 换行替换为#, 空格替换为&
text = text.replace("#", "").replace("$", "").replace("&", "")
text = text.replace("\n", "").replace(" ", "")
return text
#用的时候改改属性即可，中文的需要bert来tokenizer
class myDataset(Dataset):
def __init__(self, justice, charge):
self.justice = justice
self.charge= torch.LongTensor(charge)
def __getitem__(self, idx):
#为了匹配bert_tokenizer的返回结果
return {"justice":
{
"input_ids": self.justice["input_ids"][idx],
"token_type_ids": self.justice["token_type_ids"][idx],
"attention_mask": self.justice["attention_mask"][idx],
}, "charge": self.charge[idx]}
def __len__(self):
return len(self.charge)
def get_split_dataset(idx,justice,charge):
#此处根据ID返回对应的数据，因为dataset需要根据id进行划分
justice_cur = {
"input_ids": justice["input_ids"][idx],
"token_type_ids": justice["token_type_ids"][idx],
"attention_mask": justice["attention_mask"][idx],
}
charge_cur=pd.Series(charge)[idx].tolist()
return myDataset(justice_cur,charge_cur)
def simple_load_data(data_path,seq_len,clean:bool):
data = pd.read_csv(data_path,sep="\t",encoding='utf-8',header=None,names=["fact","type"])
#建一个存clean后的dataset的文件夹并保存clean文件
path, file = os.path.split(data_path)
stem, suffix = os.path.splitext(file)
# print(path,file,stem,suffix)
if clean:
file = stem+"_clean"+suffix
clean_data_path = os.path.join(path, file)
if not os.path.exists(clean_data_path):
for i in tqdm(range(len(data))):
data["fact"][i] = text_cleaner(data["fact"][i])
data.to_csv(clean_data_path, index=False, sep=",")
data = pd.read_csv(clean_data_path,sep=",")
# print(data.head(10))
#清除数据中含有None的行
if len(data)!=len(data.dropna()):
print("before dropna, data num:",len(data))
print("after dropna, data num:",len(data.dropna()))
data=data.dropna()
data=data.reset_index()
# print(data.head(10))
#tokenizer
#根据clean参数看保存
pkl_path = "data/pkl/"+stem+"_clean.pkl" if clean else "data/pkl/"+stem+".pkl"
# print(data["fact"].tolist())
if not os.path.exists(pkl_path):
tokenizer=BertTokenizer.from_pretrained("bert-base-chinese")
fact=data["fact"].tolist()
fact=tokenizer(fact,return_tensors="pt",padding="max_length",max_length=seq_len,truncation=True)
with open(pkl_path,"wb") as f:
pickle.dump(fact,f,protocol=pickle.HIGHEST_PROTOCOL)
print("pkl saved: {}".format(pkl_path))
else:
with open(pkl_path,"rb") as f:
fact=pickle.load(f)
# print(fact)
type=data["type"].tolist()
# print(type)
#根据idx创建dataset
idx=list(range(len(type)))
# print(idx)
# print(fact)
# print(type)
dataset=get_split_dataset(idx,fact,type)
return dataset
#此时只用了justice预测judge
def load_confusing_data(filename, seq_len, embedding_path, train_set_ratio=0.8, text_clean=True):
#LA:将错误行跳过
df = pd.read_csv(filename, sep=",",encoding="gbk")
path, file = os.path.split(filename)
stem, suffix = os.path.splitext(file)
# 读入csv数据，根据上面text_clean参数是否过stopwords
if text_clean:
file = stem+"_clean"+suffix
clean_data_path = os.path.join(path, file)
if not os.path.exists(clean_data_path):
for i in tqdm(range(len(df))):
# df["charge"][i] = text_cleaner(df["charge"][i])
df["justice"][i] = text_cleaner(df["justice"][i])
df["opinion"][i] = text_cleaner(df["opinion"][i])
df.to_csv(clean_data_path, index=False, sep=",")
df = pd.read_csv(clean_data_path)
if len(df) != len(df.dropna()):
print("before drop nan, data num: ", len(df))
print("after drop nan, data num: ", len(df.dropna()))
df = df.dropna()
df = df.reset_index()
# 将fact和opinion文本进行tokenize
pkl_path = "data/pkl/"+stem+"_clean.pkl" if text_clean else "data/pkl/"+stem+".pkl"
if not os.path.exists(pkl_path):
#用字划分数据集得到id，使用现有的Bert_tokenizer
# tokenizer = BertTokenizer.from_pretrained('bert-base-chinese')
#用词划分，使用自己的tokenizer,替换时注意删除原有的pkl文件
tokenizer=MyTokenizer(embedding_path)
justice = df["justice"].tolist()
opinion = df["opinion"].tolist()
justice = tokenizer(justice, return_tensors="pt",
padding="max_length", max_length=seq_len, truncation=True)
opinion = tokenizer(opinion, return_tensors="pt",
padding="max_length", max_length=seq_len, truncation=True)
with open(pkl_path, "wb") as f:
pickle.dump((justice, opinion), f,
protocol=pickle.HIGHEST_PROTOCOL)
with open(pkl_path, "rb") as f:
justice, opinion = pickle.load(f)
charge = df["charge"].tolist()
# judge = df["judge"].tolist() if "judge" in df.columns else [-1]*len(charge)
# article = df["article"].tolist() if "article" in df.columns else [-1]*len(charge)
# use_violence = df['use_violence'].tolist()
# violence_target = df['violence_target'].tolist()
# violence_victim = df['violence_victim'].tolist()
# violate_victim = df['violate_victim'].tolist()
# use_contract = df['use_contract'].tolist()
# 标签转数字id
def label2idx(label):
st = set(label)
lst = sorted(list(st)) # 按照字符串顺序排列
mp_label2idx, mp_idx2label = dict(), dict()
for i in range(len(lst)):
mp_label2idx[lst[i]] = i
mp_idx2label[i] = lst[i]
return [mp_label2idx[i] for i in label], mp_label2idx, mp_idx2label
charge, mp_charge2idx, mp_idx2charge = label2idx(charge)
# article, mp_article2idx, mp_idx2article = label2idx(article)
maps = {}
maps["charge2idx"] = mp_charge2idx
maps["idx2charge"] = mp_idx2charge
# maps["article2idx"] = mp_article2idx
# maps["idx2article"] = mp_idx2article
# 划分trainset, validset
data_split_dir = "data/data_split/20211216/"
if not os.path.exists(data_split_dir):
os.mkdir(data_split_dir)
tot_size = len(df)
train_size = int(tot_size*train_set_ratio)
valid_size = int((tot_size-train_size)/2)
test_size = tot_size-train_size-valid_size
random.seed(RANDOM_SEED)
shuffle_idx = list(range(len(charge)))
random.shuffle(shuffle_idx)
train_idx, valid_idx, test_idx = shuffle_idx[:train_size], shuffle_idx[
train_size:train_size+valid_size], shuffle_idx[train_size+valid_size:]
train_df, valid_df, test_df = df.iloc[train_idx], df.iloc[valid_idx], df.iloc[test_idx]
train_df.to_csv(data_split_dir+"train.csv", index=False, sep=",")
valid_df.to_csv(data_split_dir+"valid.csv", index=False, sep=",")
test_df.to_csv(data_split_dir+"test.csv", index=False, sep=",")
trainset = get_split_dataset(
train_idx, justice, charge)
validset = get_split_dataset(
valid_idx, justice, charge)
testset = get_split_dataset(
test_idx, justice, charge)
return trainset, validset, testset, maps
def simple_load_confusing_data(filename, seq_len, embedding_path, text_clean:bool):
#LA:将错误行跳过
df = pd.read_csv(filename, sep=",")
path, file = os.path.split(filename)
stem, suffix = os.path.splitext(file)
# 读入csv数据，根据上面text_clean参数是否过stopwords
if text_clean:
file = stem+"_clean"+suffix
clean_data_path = os.path.join(path, file)
if not os.path.exists(clean_data_path):
for i in tqdm(range(len(df))):
# df["charge"][i] = text_cleaner(df["charge"][i])
df["justice"][i] = text_cleaner2(df["justice"][i])
# df["opinion"][i] = text_cleaner2(df["opinion"][i])
df.to_csv(clean_data_path, index=False, sep=",")
df = pd.read_csv(clean_data_path)
if len(df) != len(df.dropna()):
print("before drop nan, data num: ", len(df))
print("after drop nan, data num: ", len(df.dropna()))
df = df.dropna()
df = df.reset_index()
# 将fact和opinion文本进行tokenize
# pkl_path = "data/pkl_4sin/"+stem+"_clean.pkl" if text_clean else "data/pkl_4sin/"+stem+".pkl"
pkl_path = "data/pkl_4sin_el/"+stem+"_clean.pkl" if text_clean else "data/pkl_4sin_el/"+stem+".pkl"
if not os.path.exists(pkl_path):
#用字划分数据集得到id，使用现有的Bert_tokenizer
tokenizer = BertTokenizer.from_pretrained('bert-base-chinese')
# 用词划分，使用自己的tokenizer,替换时注意删除原有的pkl文件
# tokenizer=MyTokenizer(embedding_path)
justice = df["justice"].tolist()
# opinion = df["opinion"].tolist()
justice = tokenizer(justice, return_tensors="pt",
padding="max_length", max_length=seq_len, truncation=True)
# opinion = tokenizer(opinion, return_tensors="pt",
# padding="max_length", max_length=seq_len, truncation=True)
with open(pkl_path, "wb") as f:
pickle.dump(justice, f,
protocol=pickle.HIGHEST_PROTOCOL)
else:
with open(pkl_path, "rb") as f:
justice= pickle.load(f)
charge = df["charge"].tolist()
# 标签转数字id
def label2idx(label):
maps = {}
c2i_path = "data/charge2idx.json"
with open(c2i_path) as f:
c2i = json.load(f)
maps["charge2idx"] = c2i
maps["idx2charge"] = {str(v): k for k, v in c2i.items()}
return [maps["charge2idx"][i] for i in label], maps["charge2idx"], maps["idx2charge"]
charge, mp_charge2idx, mp_idx2charge = label2idx(charge)
# article, mp_article2idx, mp_idx2article = label2idx(article)
maps = {}
maps["charge2idx"] = mp_charge2idx
maps["idx2charge"] = mp_idx2charge
# maps["article2idx"] = mp_article2idx
# maps["idx2article"] = mp_idx2article
idx=list(range(len(charge)))
dataset = get_split_dataset(idx, justice, charge)
return dataset, maps
def load_cail2018_data(filename, seq_len, embedding_path, text_clean:bool):
#LA:将错误行跳过
source=[]
with open(filename,'r',encoding="utf-8") as f:
for line in f:
source.append((json.loads(line)))
df_data=pd.DataFrame(source)
charge=[]
justice=[]
for i in range(len(df_data)):
# print(df_data["meta"].iloc[i]["accusation"])
if len(df_data["meta"][i]["accusation"])==1:
tmp=str(df_data["meta"][i]["accusation"])
tmp=tmp.replace("[","").replace("]","").replace("'","")
charge.append(tmp+"罪")
justice.append(df_data["fact"][i])
df=pd.DataFrame()
df["charge"]=charge
df["justice"]=justice
path, file = os.path.split(filename)
stem, suffix = os.path.splitext(file)
# 读入csv数据，根据上面text_clean参数是否过stopwords
if text_clean:
file = stem+"_clean"+suffix
clean_data_path = os.path.join(path, file)
if not os.path.exists(clean_data_path):
for i in tqdm(range(len(df))):
# df["charge"][i] = text_cleaner(df["charge"][i])
df["justice"][i] = text_cleaner2(df["justice"][i])
# df["opinion"][i] = text_cleaner2(df["opinion"][i])
df.to_csv(clean_data_path, index=False, sep=",")
df = pd.read_csv(clean_data_path)
if len(df) != len(df.dropna()):
print("before drop nan, data num: ", len(df))
print("after drop nan, data num: ", len(df.dropna()))
df = df.dropna()
df = df.reset_index()
# 将fact和opinion文本进行tokenize
pkl_path = "data/pkl18_new/"+stem+"_clean.pkl" if text_clean else "data/pkl18_new/"+stem+".pkl"
if not os.path.exists(pkl_path):
#用字划分数据集得到id，使用现有的Bert_tokenizer
# tokenizer = BertTokenizer.from_pretrained('bert-base-chinese')
#用词划分，使用自己的tokenizer,替换时注意删除原有的pkl文件
tokenizer=MyTokenizer(embedding_path)
justice = df["justice"].tolist()
# opinion = df["opinion"].tolist()
justice = tokenizer(justice, return_tensors="pt",
padding="max_length", max_length=seq_len, truncation=True)
# opinion = tokenizer(opinion, return_tensors="pt",
# padding="max_length", max_length=seq_len, truncation=True)
with open(pkl_path, "wb") as f:
pickle.dump(justice, f,
protocol=pickle.HIGHEST_PROTOCOL)
else:
with open(pkl_path, "rb") as f:
justice= pickle.load(f)
charge = df["charge"].tolist()
# 标签转数字id
def label2idx(label):
st = set(label)
lst = sorted(list(st)) # 按照字符串顺序排列
mp_label2idx, mp_idx2label = dict(), dict()
for i in range(len(lst)):
mp_label2idx[lst[i]] = i
mp_idx2label[i] = lst[i]
return [mp_label2idx[i] for i in label], mp_label2idx, mp_idx2label
charge, mp_charge2idx, mp_idx2charge = label2idx(charge)
# article, mp_article2idx, mp_idx2article = label2idx(article)
maps = {}
maps["charge2idx"] = mp_charge2idx
maps["idx2charge"] = mp_idx2charge
# 划分trainset, validset
data_split_dir = "data/cail2018train_split/"
if not os.path.exists(data_split_dir):
os.mkdir(data_split_dir)
train_set_ratio=0.8
tot_size = len(df)
train_size = int(tot_size*train_set_ratio)
valid_size = int((tot_size-train_size)/2)
test_size = tot_size-train_size-valid_size
random.seed(RANDOM_SEED)
shuffle_idx = list(range(len(charge)))
random.shuffle(shuffle_idx)
train_idx, valid_idx, test_idx = shuffle_idx[:train_size], shuffle_idx[
train_size:train_size+valid_size], shuffle_idx[train_size+valid_size:]
train_df, valid_df, test_df = df.iloc[train_idx], df.iloc[valid_idx], df.iloc[test_idx]
train_df.to_csv(data_split_dir+"train.csv", index=False, sep=",")
valid_df.to_csv(data_split_dir+"valid.csv", index=False, sep=",")
test_df.to_csv(data_split_dir+"test.csv", index=False, sep=",")
train_set = get_split_dataset(train_idx, justice, charge)
valid_set = get_split_dataset(valid_idx, justice, charge)
test_set = get_split_dataset(test_idx, justice, charge)
return train_set, valid_set, test_set, maps
def load_law_data(filename, seq_len, embedding_path, text_clean:bool):
#LA:将错误行跳过
df=pd.read_csv(filename, sep=",")
path, file = os.path.split(filename)
stem, suffix = os.path.splitext(file)
# 读入csv数据，根据上面text_clean参数是否过stopwords
if text_clean:
file = stem+"_clean"+suffix
clean_data_path = os.path.join(path, file)
if not os.path.exists(clean_data_path):
for i in tqdm(range(len(df))):
# df["charge"][i] = text_cleaner(df["charge"][i])
df["justice"][i] = text_cleaner2(df["justice"][i])
# df["opinion"][i] = text_cleaner2(df["opinion"][i])
df.to_csv(clean_data_path, index=False, sep=",")
df = pd.read_csv(clean_data_path)
if len(df) != len(df.dropna()):
print("before drop nan, data num: ", len(df))
print("after drop nan, data num: ", len(df.dropna()))
df = df.dropna()
df = df.reset_index()
# 将fact和opinion文本进行tokenize
pkl_path = "data/pkl_6sin/"+stem+"_clean.pkl" if text_clean else "data/pkl_6sin/"+stem+".pkl"
if not os.path.exists(pkl_path):
#用字划分数据集得到id，使用现有的Bert_tokenizer
# tokenizer = BertTokenizer.from_pretrained('bert-base-chinese')
#用词划分，使用自己的tokenizer,替换时注意删除原有的pkl文件
tokenizer=MyTokenizer(embedding_path)
justice = df["justice"].tolist()
# opinion = df["opinion"].tolist()
justice = tokenizer(justice, return_tensors="pt",
padding="max_length", max_length=seq_len, truncation=True)
# opinion = tokenizer(opinion, return_tensors="pt",
# padding="max_length", max_length=seq_len, truncation=True)
with open(pkl_path, "wb") as f:
pickle.dump(justice, f,
protocol=pickle.HIGHEST_PROTOCOL)
else:
with open(pkl_path, "rb") as f:
justice= pickle.load(f)
charge = df["charge"].tolist()
# 标签转数字id
def label2idx(label):
st = set(label)
lst = sorted(list(st)) # 按照字符串顺序排列
mp_label2idx, mp_idx2label = dict(), dict()
for i in range(len(lst)):
mp_label2idx[lst[i]] = i
mp_idx2label[i] = lst[i]
return [mp_label2idx[i] for i in label], mp_label2idx, mp_idx2label
charge, mp_charge2idx, mp_idx2charge = label2idx(charge)
# article, mp_article2idx, mp_idx2article = label2idx(article)
maps = {}
maps["charge2idx"] = mp_charge2idx
maps["idx2charge"] = mp_idx2charge
# 划分trainset, validset
data_split_dir = "data/6sin_split/"
if not os.path.exists(data_split_dir):
os.mkdir(data_split_dir)
RANDOM_SEED=22
random.seed(RANDOM_SEED)
shuffle_idx = list(range(len(df)))
random.shuffle(shuffle_idx)
print(df.shape)
train_set_ratio=0.8
tot_size = len(df)
train_size = int(tot_size*train_set_ratio)
valid_size = int((tot_size-train_size)/2)
train_idx = shuffle_idx[:train_size]
valid_idx = shuffle_idx[train_size:train_size+valid_size]
test_idx=shuffle_idx[train_size+valid_size:]
train_df, valid_df, test_df = df.iloc[train_idx], df.iloc[valid_idx],df.iloc[test_idx]
train_df.to_csv(data_split_dir+"train.csv", index=False, sep=",")
valid_df.to_csv(data_split_dir+"valid.csv", index=False, sep=",")
test_df.to_csv(data_split_dir+"test.csv", index=False, sep=",")
train_set = get_split_dataset(train_idx, justice, charge)
valid_set = get_split_dataset(valid_idx, justice, charge)
test_set = get_split_dataset(test_idx, justice, charge)
return train_set, valid_set, test_set, maps
#测试dataloader
if __name__=="__main__":
total_data_path="/mnt/data/wuyiquan/liang/charge_prediction/data/total_6sin.csv"
train_set, valid_set, test_set, maps=load_law_data(total_data_path,seq_len=512,embedding_path="/mnt/data/wuyiquan/liang/charge_prediction/gensim_train/word2vec_6sin.model",text_clean=False)
train_iter=DataLoader(train_set,batch_size=128,shuffle=False,drop_last=True)
valid_iter=DataLoader(valid_set,batch_size=128,shuffle=False,drop_last=False)
test_iter=DataLoader(test_set,batch_size=128,shuffle=False,drop_last=False)
#测试dataiter
cnt=0
for data in enumerate(valid_iter):
print(data)
cnt+=1
if cnt>1:
break