/
main.py
163 lines (131 loc) · 5.52 KB
/
main.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
import logging
import torch
import numpy as np
from wrench.dataset import load_dataset
from wrench._logging import LoggingHandler
from wrench.labelmodel import FlyingSquid
from wrench.endmodel import EndClassifierModel
from utils.RefiningModule import get_graph, get_removing_list
from utils.Utils import init_random, remove_LF
from scipy.stats import sem
import os
import torch
import argparse
import time
parser = argparse.ArgumentParser()
# required arguments
parser.add_argument('--dataset_path', default = '')
parser.add_argument('--data', default = 'youtube', type =str)
parser.add_argument('--total_lf', default = 10, type = int)
parser.add_argument('--batch_size', default = 8, type = int)
parser.add_argument('--test_batch_size', default = 512, type = int)
parser.add_argument('--optimizer_lr', default = 5e-5, type = float)
parser.add_argument('--metric', default = 'f1_binary', type = str)
parser.add_argument('--patience', default = 50, type = int)
parser.add_argument('--random_seed', nargs='+', type = int)
# parser.add_argument('--random_seed', default = [1,2])
parser.add_argument('--threshold_structure', default = 0.2, type = float)
parser.add_argument('--threshold_removing', default = 1, type =float)
parser.add_argument('--save_dir', default = './results/', type = str)
parser.add_argument('--endModel_weight_decay', default = 0.0, type = float)
args = parser.parse_args()
args = vars(args)
dataset_path = args['dataset_path']
data = args['data']
save_file = os.path.join(args['save_dir'], (args['dataset_path'].split('/')[2]+'_'+args['data']+'.txt'))
with open(save_file, 'a') as file:
file.write( 'batch size: '+ str(args['batch_size']) + ', test batch size: '+ str(args['test_batch_size'])+ ', learning rate: '+ str(args['optimizer_lr'])+ ', weight decay: '+ str(args['endModel_weight_decay']) +', threshold_structure: '+ str(args['threshold_structure']) +', threshold_removing: '+ str(args['threshold_removing'])+'\n')
logging.basicConfig(format='%(asctime)s - %(message)s',
datefmt='%Y-%m-%d %H:%M:%S',
level=logging.INFO,
handlers=[LoggingHandler()])
logger = logging.getLogger(__name__)
result_dict = {}
device = torch.device('cuda')
result_List = []
for pos, seed in enumerate(args['random_seed']):
init_random(seed)
logging.info('random seed used: %d', seed)
logging.info('loading data...')
result_dict['seed_{}'.format(seed)] ={}
result_dict['seed_{}'.format(seed)]['metric'] = []
label_model = FlyingSquid()
LFs_removed = get_removing_list(args)
with open(save_file, 'a') as file:
file.write(f'Seed: {seed} \n')
with open(save_file, 'a') as file:
file.write('LF removed: ')
file.write('[ ')
for i in LFs_removed:
file.write( str(i) +', ')
file.write(' ]')
file.write('\n')
train_data, valid_data, test_data = load_dataset(
dataset_path,
data,
extract_feature=False,
extract_fn='bert',
model_name='bert-base-cased',
cache_name='bert'
)
train_data, valid_data, test_data = remove_LF(train_data, LFs_removed), remove_LF(valid_data, LFs_removed), remove_LF(test_data, LFs_removed)
dependency_graph = get_graph(args, LFs_removed)
with open(save_file, 'a') as file:
file.write('dependencies: ')
file.write('[ ')
for i in dependency_graph:
file.write( '( '+ str(i[0]) + ', '+str(i[1])+' )' +', ')
file.write(' ]')
file.write('\n')
model_start = time.time()
label_model.fit(
dataset_train=train_data,
dataset_valid=valid_data,
dependency_graph = dependency_graph
)
with open(save_file, 'a') as file:
file.write(f'Fit label model time: {time.time()-model_start}\n')
total_data_num = len(train_data)
train_data = train_data.get_covered_subset()
aggregated_hard_labels = label_model.predict(train_data)
aggregated_soft_labels = label_model.predict_proba(train_data)
model = EndClassifierModel(
batch_size=args['batch_size'],
real_batch_size=args['batch_size'], # for accumulative gradient update
test_batch_size=args['test_batch_size'],
n_steps=1000,
backbone='BERT',
backbone_model_name='roberta-base',
backbone_max_tokens=128,
backbone_fine_tune_layers=-1, # fine tune all
optimizer='AdamW',
optimizer_lr=args['optimizer_lr'],
optimizer_weight_decay=0.0,
)
model.fit(
dataset_train=train_data,
y_train=aggregated_soft_labels,
dataset_valid=valid_data,
evaluation_step=10,
metric=args['metric'],
patience=args['patience'],
device=device
)
endmodel_test = model.test(test_data, args['metric'])
with open(save_file, 'a') as file:
file.write('End model' + args['metric'] +': '+ str(endmodel_test) +'\n\n')
acc = model.test(test_data, args['metric'])
logging.info('end model (Roberta) test {}: {}'.format(args['metric'], acc))
result_dict['seed_{}'.format(seed)]['metric'].append(float(acc))
del model
result_List.append(result_dict['seed_{}'.format(seed)]['metric'])
with open(save_file, 'a') as file:
file.write('Means: ')
for i in np.mean(result_List,0):
file.write( str(i) + ', ')
file.write('\n')
file.write('STD:')
for i in sem(result_List,0):
file.write( str(i) + ', ')
file.write('\n\n')
print('Result', result_dict)