-
Notifications
You must be signed in to change notification settings - Fork 5
/
nli_utils.py
69 lines (57 loc) · 3.15 KB
/
nli_utils.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
from transformers.data.processors.utils import DataProcessor, InputExample, InputFeatures
import pandas as pd
import numpy as np
class ExpProcessor(DataProcessor):
s1 = 'sentence1'
s2 = 'sentence2'
index_col = "pairID"
labels = ["entailment", "contradiction", "neutral"]
gold_label = "gold_label"
def get_train_examples(self, filepath, data_format="instance", to_drop=[]):
data = pd.read_csv(filepath, index_col=self.index_col)
examples = self._create_examples(data, 'train', data_format=data_format, to_drop=to_drop)
return examples
def get_dev_examples(self, filepath, data_format="instance", to_drop=[]):
data = pd.read_csv(filepath, index_col=self.index_col)
examples = self._create_examples(data, 'dev', data_format=data_format, to_drop=to_drop)
return examples
def get_labels(self):
return self.labels
data_formats = ["instance", "independent", "append", "instance_independent", "instance_append",
"all_explanation", "Explanation_1"]
#aggregate uses the same format as independent
def _create_examples(self, labeled_examples, set_type, data_format="instance", to_drop=[]):
"""Creates examples for the training and dev sets."""
if data_format not in self.data_formats:
raise ValueError("Data format {} not supported".format(data_format))
if 'explanation' in to_drop: to_drop = self.labels
keep_labels = [True if l not in to_drop else False for l in self.labels]
exp_names = ["{}_explanation".format(l) for l in self.labels]
examples = []
for (idx, le) in labeled_examples.iterrows():
guid = idx
label = le[self.gold_label]
if data_format in ["independent", "instance_independent"]:
exp_text = [le[exp_name] if keep else ""
for l, keep, exp_name in zip(self.labels, keep_labels, exp_names)]
elif data_format in ["append", "instance_append"]:
exp_text = " ".join(["{}: {}".format(l, le[exp_name]) if keep else ""
for l, keep, exp_name in zip(self.labels, keep_labels, exp_names)])
if data_format == "instance":
text_a, text_b = le[self.s1], le[self.s2]
elif data_format in ["Explanation_1", "all_explanation"]:
text_a, text_b = le[data_format], None
elif data_format in ["independent", "append"]:
text_a, text_b = exp_text, None
elif data_format in ["instance_independent", "instance_append"]:
instance_text = "Premise: {} Hypothesis: {}".format(
le[self.s1], le[self.s2]) if "instance" not in to_drop else "Premise: Hypothesis:"
text_a, text_b = instance_text, exp_text
examples.append(
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
return examples
def simple_accuracy(preds, labels):
return (preds == labels).mean()
def exp_compute_metrics(preds, labels):
assert len(preds) == len(labels)
return {"acc": simple_accuracy(preds, labels)}