-
Notifications
You must be signed in to change notification settings - Fork 0
/
dataset.py
72 lines (56 loc) · 2.7 KB
/
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
import torch
import pandas as pd
class EthicsDataset(torch.utils.data.Dataset):
def __init__(self, tokenizer, csv_path, data, max_length=64):
self.data = data
if data == 'cm':
df = pd.read_csv(csv_path)
self.scenarios = df['input'].tolist()
self.labels = df['label'].tolist()
self.encodings = tokenizer(self.scenarios,
max_length=max_length,
padding='max_length',
truncation=True)
elif data == 'deontology':
df = pd.read_csv(csv_path)
self.scenarios = df['scenario'].tolist()
self.execuses = df['excuse'].tolist()
self.labels = df['label'].tolist()
self.encodings = tokenizer(self.scenarios,
self.execuses,
max_length=max_length,
padding='max_length',
truncation=True)
elif data == 'util':
df = pd.read_csv(csv_path, header=None)
self.sentence1 = df[0].tolist()
self.sentence2 = df[1].tolist()
self.encodings1 = tokenizer(self.sentence1,
max_length=max_length,
padding='max_length',
truncation=True)
self.encodings2 = tokenizer(self.sentence2,
max_length=max_length,
padding='max_length',
truncation=True)
self.labels = torch.ones(len(self.encodings1['input_ids']))
else:
df = pd.read_csv(csv_path)
self.scenarios = df['scenario'].tolist()
self.labels = df['label'].tolist()
self.encodings = tokenizer(self.scenarios,
max_length=max_length,
padding='max_length',
truncation=True)
def __getitem__(self, idx):
if self.data == 'util':
item1 = {k: torch.Tensor(v[idx]) for k, v in self.encodings1.items()}
item2 = {k: torch.Tensor(v[idx]) for k, v in self.encodings2.items()}
labels = self.labels[idx]
return item1, item2, labels
else:
item = {k: torch.Tensor(v[idx]) for k, v in self.encodings.items()}
labels = self.labels[idx]
return item, labels
def __len__(self):
return len(self.labels)