-
Notifications
You must be signed in to change notification settings - Fork 11
/
gen_data.py
227 lines (209 loc) · 8.68 KB
/
gen_data.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
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
import argparse
import os
import json
from data.grade_data.data_loader import load_grade_data
from data.usr_data.data_loader import load_usr_data
from data.dstc6_data.data_loader import load_dstc6_data
from data.fed_data.data_loader import load_fed_data, load_fed_dialog_data
from data.dstc9_data.data_loader import load_dstc9_data
from data.holistic_data.data_loader import load_holistic_data
from data.engage_data.data_loader import load_engage_data
'''
Adding new data:
from data.data.data_laoder import load_new_data # the customized data loader function
'''
from maude.data_parser import gen_maude_data
from grade.data_parser import gen_grade_data
from ruber.data_parser import gen_ruber_data
from holistic_eval.data_parser import gen_hostilic_data
from predictive_engagement.data_parser import gen_engagement_data
from am_fm.data_parser import gen_amfm_data
from usl_dialogue_metric.data_parser import gen_usl_data
from deb.data_parser import gen_deb_data
from dynaeval.data_parser import gen_dynaeval_data
from fbd.data_parser import gen_fbd_data
from dialogrpt.data_parser import gen_dialogrpt_data
def parse_args():
parser = argparse.ArgumentParser(description='')
parser.add_argument('--source_data', type=str, default=None)
parser.add_argument('--target_format', type=str, default=None)
args = parser.parse_args()
return args
def gen_baseline_data(data, data_path):
with open(data_path, 'w') as fout:
json.dump(data, fout)
def main(source_data, target_format):
format_type = 0 # two types of generating data
max_words = None # for models using pretrained language models with input length constraint such as BERT
if target_format == 'maude':
metric = 'maude'
output_dir = f'{os.getcwd()}/maude/eval_data'
gen_data = gen_maude_data
suffix = '.csv'
max_words = 500
elif target_format == 'hostilic':
metric = 'hostilic'
output_dir = f'{os.getcwd()}/holistic_eval/eval_data'
gen_data = gen_hostilic_data
suffix = '.csv'
max_words = 500
elif target_format == 'baseline':
metric = 'baseline'
output_dir = f'{os.getcwd()}/baseline_data'
gen_data = gen_baseline_data
suffix = '.json'
elif target_format == 'usr_fed':
metric = 'usr_fed'
output_dir = f'{os.getcwd()}/usr_fed_data'
gen_data = gen_baseline_data
suffix = '.json'
max_words = 500
elif target_format == 'ruber':
metric = 'ruber'
output_dir = f'{os.getcwd()}/ruber_and_bert_ruber/RUBER/data'
gen_data = gen_ruber_data
suffix = ''
elif target_format == 'bert_ruber':
metric = 'bert_ruber'
output_dir = f'{os.getcwd()}/PONE/PONE/data'
gen_data = gen_ruber_data
suffix = ''
elif target_format == 'grade':
metric = 'grade'
output_dir = f'{os.getcwd()}/grade'
gen_data = gen_grade_data
suffix = '.json'
format_type = 1
elif target_format == 'predictive_engagement':
metric = 'predictive_engagement'
output_dir = f'{os.getcwd()}/predictive_engagement/data'
gen_data = gen_engagement_data
suffix = '.csv'
elif target_format =='amfm':
metric = 'amfm'
output_dir = f'{os.getcwd()}/am_fm/examples/dstc6/test_data'
gen_data = gen_amfm_data
suffix = ''
elif target_format == 'flowscore':
metric = 'flowscore'
output_dir = f'{os.getcwd()}/FlowScore/eval_data'
gen_data = gen_baseline_data
suffix = '.json'
elif target_format == 'usl':
metric = 'usl'
output_dir = f'{os.getcwd()}/usl_dialogue_metric/usl_score/datasets'
gen_data = gen_usl_data
suffix = ''
elif target_format == 'questeval':
metric = 'questeval'
output_dir = f'{os.getcwd()}/questeval/test_data'
gen_data = gen_baseline_data
suffix = '.json'
elif target_format == 'deb':
metric = 'deb'
output_dir = f'{os.getcwd()}/deb/dataset'
gen_data = gen_deb_data
suffix = ''
elif target_format == 'dynaeval':
metric = 'dynaeval'
output_dir = f'{os.getcwd()}/dynaeval/data'
gen_data = gen_dynaeval_data
suffix = ''
elif target_format == 'fbd':
metric = 'fbd'
output_dir = f'{os.getcwd()}/fbd/datasets'
gen_data = gen_fbd_data
suffix = ''
elif target_format == 'dialogrpt':
metric = 'dialogrpt'
output_dir = f'{os.getcwd()}/dialogrpt/test_data'
gen_data = gen_dialogrpt_data
suffix = ''
else:
raise Exception
'''
Adding a new metric
elif target_format == 'metric':
metirc = 'METRIC_NAME'
output_dir = 'PATH/TO/OUTPUT/DIR'
gen_data = gen_metric_data # the customized function
suffix = '' # the suffix of generated data
'''
if source_data == 'convai2_grade':
model_names = ['bert_ranker', 'dialogGPT', 'transformer_generator', 'transformer_ranker']
for model in model_names:
data_path = f'{os.getcwd()}/data/grade_data'
data = load_grade_data(data_path, 'convai2', model)
if format_type == 0:
output_path = f'{output_dir}/convai2_grade_{model}{suffix}'
gen_data(data, output_path)
elif format_type == 1:
gen_data(data, output_dir, f'{source_data}_{model}')
elif source_data == 'dailydialog_grade':
model_names = ['transformer_generator', 'transformer_ranker']
for model in model_names:
data_path = f'{os.getcwd()}/data/grade_data'
data = load_grade_data(data_path, 'dailydialog', model)
if format_type == 0:
output_path = f'{output_dir}/dailydialog_grade_{model}{suffix}'
gen_data(data, output_path)
elif format_type == 1:
gen_data(data, output_dir, f'{source_data}_{model}')
elif source_data == 'empatheticdialogues_grade':
model_names = ['transformer_generator', 'transformer_ranker']
for model in model_names:
data_path = f'{os.getcwd()}/data/grade_data'
data = load_grade_data(data_path, 'empatheticdialogues', model)
if format_type == 0:
output_path = f'{output_dir}/empatheticdialogues_grade_{model}{suffix}'
gen_data(data, output_path)
elif format_type == 1:
gen_data(data, output_dir, f'{source_data}_{model}')
elif source_data == 'personachat_usr':
data_path = f'{os.getcwd()}/data/usr_data'
data = load_usr_data(data_path, 'personachat')
if format_type == 0:
output_path = f'{output_dir}/personachat_usr{suffix}'
gen_data(data, output_path)
elif format_type == 1:
gen_data(data, output_dir, 'personachat_usr')
elif source_data == 'topicalchat_usr':
data_path = f'{os.getcwd()}/data/usr_data'
data = load_usr_data(data_path, 'topicalchat')
if format_type == 0:
output_path = f'{output_dir}/topicalchat_usr{suffix}'
gen_data(data, output_path)
elif format_type == 1:
gen_data(data, output_dir, 'topicalchat_usr')
elif source_data == 'fed_dialog':
data_path = f'{os.getcwd()}/data/fed_data'
data = load_fed_dialog_data(data_path)
if format_type == 0:
output_path = f'{output_dir}/fed_dialog{suffix}'
gen_data(data, output_path)
elif format_type == 1:
gen_data(data, output_dir, 'fed_dialog')
else:
data_path = f'{os.getcwd()}/data/{source_data}_data'
data = eval(f'load_{source_data}_data')(data_path)
if format_type == 0:
output_path = f'{output_dir}/{source_data}{suffix}'
gen_data(data, output_path)
elif format_type == 1:
gen_data(data, output_dir, source_data)
if __name__ == '__main__':
args = parse_args()
all_data = ['convai2_grade', 'dailydialog_grade', 'empatheticdialogues_grade',
'personachat_usr', 'topicalchat_usr', 'dstc6', 'fed', 'fed_dialog', 'holistic', 'dstc9', 'engage']
if args.source_data is not None:
assert args.source_data in all_data
all_data = [args.source_data]
if args.target_format is not None:
metrics = [args.target_format]
else:
metrics = ['maude', 'hostilic', 'baseline', 'usr_fed', 'ruber', 'bert_ruber', 'grade', 'predictive_engagement', 'amfm', 'flowscore',
'usl', 'questeval', 'deb', 'dynaeval', 'dialogrpt']
for data in all_data:
for target in metrics:
print(f'Generating {data} to {target}')
main(data, target)