-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathconfig_woz.py
193 lines (182 loc) · 5.14 KB
/
config_woz.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
import os
args = dict()
'''
configurations for basic model without label dependency
'''
# # root experiment folder
# args['dexp'] = 'TEN-XH'
# # dataset dir
# args['data_dir'] = 'woz'
# # args['data_dir'] = 'dstc2'
# # which model to use
# args['model'] = 'dst'
# # max epoch to run for
# args['epoch'] = 100
# # word embedding size
# args['demb'] = 400
# # hidden state size
# args['dhid'] = 50
# # tracker hidden state size
# args['hidden_s'] = 50
# # batch size
# args['batch_size'] = 10
# # learning rate
# args['lr'] = 1e-3
# # slot to early stop on
# args['stop'] = 'joint_goal'
# # save directory to resume from
# args['resume'] = None
# # args['resume'] = 'exp/dst/base/epoch=45,iter=2760,train_joint_goal=0.991325,dev_joint_goal=0.901205.t7'
# # label dependency
# args['label_depend'] = False
# # submodel to use
# args['submodel'] = 'base'
# # args['submodel'] = 'state'
# # args['submodel'] = 'FGT'
# # random seed
# args['seed'] = 42
# # run in evaluation only mode
# args['test'] = False
# # which device to use
# args['device'] = 'cuda'
# # dropout rates
# args['dropout'] = {'emb': 0.2, 'att_in': 0.2, 'att_out': 0.2, 'tracker': 0.0}
# # output file
# args['dout'] = os.path.join(args['dexp'], args['model'], args['submodel'])
# if not os.path.isdir(args['dout']):
# os.makedirs(args['dout'])
# '''
# configurations for basic model with label dependency
# '''
# # root experiment folder
# args['dexp'] = 'TEN-X'
# # dataset dir
# args['data_dir'] = 'woz'
# # args['data_dir'] = 'dstc2'
# # which model to use
# args['model'] = 'dst'
# # max epoch to run for
# args['epoch'] = 100
# # word embedding size
# args['demb'] = 400
# # hidden state size
# args['dhid'] = 50
# # tracker hidden state size
# args['hidden_s'] = 50
# # batch size
# args['batch_size'] = 10
# # learning rate
# args['lr'] = 1e-3
# # slot to early stop on
# args['stop'] = 'joint_goal'
# # save directory to resume from
# args['resume'] = None
# # args['resume'] = 'exp/dst/base/epoch=45,iter=2760,train_joint_goal=0.991325,dev_joint_goal=0.901205.t7'
# # label dependency
# args['label_depend'] = True
# # submodel to use
# args['submodel'] = 'base'
# # args['submodel'] = 'state'
# # args['submodel'] = 'FGT'
# # random seed
# args['seed'] = 42
# # run in evaluation only mode
# args['test'] = False
# # which device to use
# args['device'] = 'cuda'
# # dropout rates
# args['dropout'] = {'emb': 0.2, 'att_in': 0.2, 'att_out': 0.2, 'tracker': 0.0}
# # output file
# args['dout'] = os.path.join(args['dexp'], args['model'], args['submodel'])
# if not os.path.isdir(args['dout']):
# os.makedirs(args['dout'])
# '''
# configurations for model trained with states
# '''
# # root experiment folder
# args['dexp'] = 'TEN-Y'
# # dataset dir
# args['data_dir'] = 'woz'
# # args['data_dir'] = 'dstc2'
# # which model to use
# args['model'] = 'dst'
# # max epoch to run for
# args['epoch'] = 100
# # word embedding size
# args['demb'] = 400
# # hidden state size
# args['dhid'] = 50
# # tracker hidden state size
# args['hidden_s'] = 50
# # batch size
# args['batch_size'] = 10
# # learning rate
# args['lr'] = 1e-3
# # slot to early stop on
# args['stop'] = 'joint_goal'
# # save directory to resume from
# args['resume'] = None
# # args['resume'] = 'exp/dst/base/epoch=45,iter=2760,train_joint_goal=0.991325,dev_joint_goal=0.901205.t7'
# # label dependency
# args['label_depend'] = True
# # submodel to use
# # args['submodel'] = 'base'
# args['submodel'] = 'state'
# # args['submodel'] = 'FGT'
# # random seed
# args['seed'] = 42
# # run in evaluation only mode
# args['test'] = False
# # which device to use
# args['device'] = 'cuda'
# # dropout rates
# args['dropout'] = {'emb': 0.2, 'att_in': 0.2, 'att_out': 0.2, 'tracker': 0.0}
# # output file
# args['dout'] = os.path.join(args['dexp'], args['model'], args['submodel'])
# if not os.path.isdir(args['dout']):
# os.makedirs(args['dout'])
'''
Configurarions for FGT
'''
# root experiment folder
args['dexp'] = 'TEN'
# dataset dir
args['data_dir'] = 'woz'
# args['data_dir'] = 'dstc2'
# which model to use
args['model'] = 'dst'
# max epoch to run for
args['epoch'] = 100
# word embedding size
args['demb'] = 400
# hidden state size
args['dhid'] = 50
# tracker hidden state size
args['hidden_s'] = 50
# batch size
args['batch_size'] = 30
# learning rate
args['lr'] = 1e-3
# slot to early stop on
args['stop'] = 'joint_goal'
# save directory to resume from
# args['resume'] = None
args['resume'] = 'exp/dst/base/epoch=45,iter=2760,train_joint_goal=0.991325,dev_joint_goal=0.901205.t7'
# label dependency
args['label_depend'] = True
# submodel to use
# args['submodel'] = 'base'
# args['submodel'] = 'state'
args['submodel'] = 'FGT'
# random seed
args['seed'] = 42
# run in evaluation only mode
args['test'] = False
# which device to use
args['device'] = 'cuda'
# dropout rates
args['dropout'] = {'emb': 0.3, 'att_in': 0.3, 'att_out': 0.3, 'tracker': 0.0}
# output file
args['dout'] = os.path.join(args['dexp'], args['model'], args['submodel'])
if not os.path.isdir(args['dout']):
os.makedirs(args['dout'])