-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathexperiments_hm_script_ensembles_QA_accuracy.py
78 lines (69 loc) · 2.83 KB
/
experiments_hm_script_ensembles_QA_accuracy.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
from utils_data import create_calibrated_df
import numpy as np
from utils_constants import CORRECTNESS
splits = ['eval', 'test']
for split in splits:
# BERT
df = create_calibrated_df([
'output_bert_seed0_%s.csv' % split,
'output_bert_seed3_%s.csv' % split,
'output_bert_seed42_%s.csv' % split,
])
print('BERT: QA %s ACCURACY = %.4f' % (split, float(np.mean(df[CORRECTNESS]))))
# XLNet
df = create_calibrated_df([
'output_xlnet_seed_2_%s.csv' % split,
'output_xlnet_seed_3_%s.csv' % split,
'output_xlnet_seed_4_%s.csv' % split,
])
print('XLNet: QA %s ACCURACY = %.4f' % (split, float(np.mean(df[CORRECTNESS]))))
# DistilBERT
df = create_calibrated_df([
'output_distilbert_seed1_%s.csv' % split,
'output_distilbert_seed3_%s.csv' % split,
'output_distilbert_seed42_%s.csv' % split,
])
print('DistilBERT: QA %s ACCURACY = %.4f' % (split, float(np.mean(df[CORRECTNESS]))))
# BERT DistilBERT
df = create_calibrated_df([
'output_bert_seed0_%s.csv' % split,
'output_bert_seed3_%s.csv' % split,
'output_bert_seed42_%s.csv' % split,
'output_distilbert_seed1_%s.csv' % split,
'output_distilbert_seed3_%s.csv' % split,
'output_distilbert_seed42_%s.csv' % split,
])
print('BERT-DistilBERT: QA %s ACCURACY = %.4f' % (split, float(np.mean(df[CORRECTNESS]))))
# BERT XLNet
df = create_calibrated_df([
'output_bert_seed0_%s.csv' % split,
'output_bert_seed3_%s.csv' % split,
'output_bert_seed42_%s.csv' % split,
'output_xlnet_seed_2_%s.csv' % split,
'output_xlnet_seed_3_%s.csv' % split,
'output_xlnet_seed_4_%s.csv' % split,
])
print('BERT-XLNet: QA %s ACCURACY = %.4f' % (split, float(np.mean(df[CORRECTNESS]))))
# DistilBERT XLNet
df = create_calibrated_df([
'output_distilbert_seed1_%s.csv' % split,
'output_distilbert_seed3_%s.csv' % split,
'output_distilbert_seed42_%s.csv' % split,
'output_xlnet_seed_2_%s.csv' % split,
'output_xlnet_seed_3_%s.csv' % split,
'output_xlnet_seed_4_%s.csv' % split,
])
print('DistilBERT-XLNet: QA %s ACCURACY = %.4f' % (split, float(np.mean(df[CORRECTNESS]))))
# BERT - DistilBERT - XLNet
df = create_calibrated_df([
'output_bert_seed0_%s.csv' % split,
'output_bert_seed3_%s.csv' % split,
'output_bert_seed42_%s.csv' % split,
'output_distilbert_seed1_%s.csv' % split,
'output_distilbert_seed3_%s.csv' % split,
'output_distilbert_seed42_%s.csv' % split,
'output_xlnet_seed_2_%s.csv' % split,
'output_xlnet_seed_3_%s.csv' % split,
'output_xlnet_seed_4_%s.csv' % split,
])
print('BERT-DistilBERT-XLNet: QA %s ACCURACY = %.4f' % (split, float(np.mean(df[CORRECTNESS]))))