-
Notifications
You must be signed in to change notification settings - Fork 5
/
roledataset.py
141 lines (119 loc) · 5.83 KB
/
roledataset.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
import torch
from torch.utils.data import DataLoader
from torch.utils.data.dataset import Dataset
import pandas as pd
import config
from transformers import BertTokenizer
import ipdb
def create_dataloader(dataset, batch_size, shuffle=False):
if shuffle:
data_loader = DataLoader(dataset, batch_size=batch_size, shuffle=shuffle, num_workers=4)
else:
data_loader = DataLoader(dataset, batch_size=batch_size, shuffle=shuffle, num_workers=4) # 测试集还是别drop_last
return data_loader
class RoleDataset(Dataset):
def __init__(self, tokenizer, max_len, mode='train'):
super(RoleDataset, self).__init__()
if mode == 'train':
self.data = pd.read_csv('data/train.csv',sep='\t')
else:
self.data = pd.read_csv('data/test.csv',sep='\t')
self.mode=mode
self.text=self.data['text'].tolist()
self.labels=self.data[config.target_cols].to_dict('records')
self.tokenizer = tokenizer
self.max_len = max_len
self.id =self.data['id'].tolist()
self.character = self.data['character'].tolist()
def __getitem__(self, index):
text=str(self.text[index]) # 天空下着暴雨，o2正在给c1穿雨衣，他自己却只穿着单薄的军装，完全暴露在大雨之中。角色: o2'
label=self.labels[index] # {'love': 0, 'joy': 0, 'fright': 0, 'anger': 0, 'fear': 0, 'sorrow': 0}
id = self.id[index]
character = self.character[index]
encoding=self.tokenizer.encode_plus(text,
add_special_tokens=True, # Add '[CLS]' and '[SEP]'
truncation=True,
padding = 'max_length',
max_length=self.max_len,
return_token_type_ids=True,
return_attention_mask=True,
return_tensors='pt',)
# ipdb.set_trace()
# tokens = self.tokenizer.decode(encoding['input_ids'].flatten()).split(' ')
pos = None
try:
vocab_index = self.tokenizer.convert_tokens_to_ids(character)
pos = encoding['input_ids'][0].tolist().index(vocab_index)
# pos = tokens.index(character)
except:
print(self.mode, id, text, character)
ipdb.set_trace()
# print(tokens)
# assert tokens[pos] == character
# assert self.tokenizer.convert_ids_to_tokens(encoding['input_ids'][0][pos].item()) == character
assert pos < self.max_len and pos >= 0
sample = {
'id': id,
'text': text,
'character': character,
'pos': pos,
'input_ids': encoding['input_ids'].flatten(), # [max_length]
'attention_mask': encoding['attention_mask'].flatten(), # [max_length]
}
# ipdb.set_trace()
labels = []
for label_col in config.target_cols:
labels.append(label[label_col])
sample['labels'] = torch.tensor(labels, dtype=torch.float)
# sample['labels'] = labels
return sample
def __len__(self):
return len(self.text)
if __name__ == "__main__":
print("loading tokenizer...")
tokenizer = BertTokenizer.from_pretrained(config.PRE_TRAINED_MODEL_NAME)
print("loading finish!")
trainset = RoleDataset(tokenizer, config.max_len, mode='train')
train_loader = create_dataloader(trainset, batch_size=1)
for step, batch in enumerate(train_loader):
#print(step, batch['text'])
pass
testset = RoleDataset(tokenizer, config.max_len, mode='test')
test_loader = create_dataloader(testset, batch_size=1)
for step, batch in enumerate(test_loader):
pass
#print(step, batch['text'])
# ipdb.set_trace()
print(testset.__len__()) # 36612
print(len(test_loader)) # 36612/32=1250 ?
print(testset.__getitem__(0))
'''
sampel = {
'text': '天空下着暴雨，o2正在给c1穿雨衣，他自己却只穿着单薄的军装，完全暴露在大雨之中。角色: o2',
'input_ids': torch.Tensor([ 101, 1921, 4958, 678, 4708, 3274, 7433, 8024, 157, 8144,
3633, 1762, 5314, 10905, 4959, 7433, 6132, 8024, 800, 5632,
2346, 1316, 1372, 4959, 4708, 1296, 5946, 4638, 1092, 6163,
8024, 2130, 1059, 3274, 7463, 1762, 1920, 7433, 722, 704,
511, 6235, 5682, 131, 157, 8144, 102, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0]),
'attention_mask': torch.Tensor([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0]),
'love': torch.Tensor(0),
'joy': torch.Tensor(0),
'fright': torch.Tensor(0),
'anger': torch.Tensor(0),
'fear': torch.Tensor(0),
'sorrow': torch.Tensor(0)
}
'''