-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathargs.py
232 lines (220 loc) · 9.51 KB
/
args.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
228
229
230
231
232
import argparse
def initialize_args():
parser = argparse.ArgumentParser(description='Arguments')
parser.add_argument('--project_name', type=str, default='MoGERNN')
# Model
parser.add_argument('--model', type=str, default='MoGERNN')
parser.add_argument('--d_model', type=int, default=64)
parser.add_argument('--k_hop', type=int, default=1, help='k_hop for graph')
parser.add_argument('--activation', type=str, default='gelu')
parser.add_argument('--e_layers',
type=int,
default=2,
help='num of encoder layers')
parser.add_argument('--d_layers',
type=int,
default=1,
help='num of decoder layers')
parser.add_argument('--mean_expert',
action='store_true',
help='use mean expert')
parser.add_argument('--weight_expert',
action='store_true',
help='use distane-based weighted expert')
parser.add_argument('--max_expert',
action='store_true',
help='use max pooling expert')
parser.add_argument('--min_expert',
action='store_true',
help='use min pooling expert')
parser.add_argument('--diffusion_expert',
action='store_true',
help='use diffusion expert')
parser.add_argument('--num_used_experts', type=int, default=3)
parser.add_argument('--t_mark',
action='store_true',
help='do not add time stamp to feature')
parser.add_argument('--pos_mark',
action='store_true',
help='do not add pos stamp to feature')
# train
parser.add_argument('--step_type',
type=str,
default='step',
help='step type for the model')
parser.add_argument('--max_epoch', type=int, default=50)
parser.add_argument('--loss',
type=str,
default='mse',
choices=[
'mse', 'mae', 'rmse', 'huber', 'msepinball',
'multipinball', 'nll'
],
help='loss function used for backpropagation')
parser.add_argument('--device',
type=str,
default='cuda:0',
help='device to train on')
parser.add_argument('--seed',
type=int,
default=42,
help='seed for reproducibility')
parser.add_argument('--lr',
type=float,
default=0.001,
help='learning rate')
parser.add_argument('--weight_decay',
type=float,
default=0.0001,
help='weight decay')
parser.add_argument('--optimizer',
type=str,
default='adamw',
choices=['adam', 'sgd', 'adamw'],
help='optimizer')
parser.add_argument('--scheduler',
type=str,
default='plateau',
choices=['plateau', 'step', 'cosine', 'None'],
help='scheduler')
parser.add_argument("--scheduler_patience",
type=int,
default=1,
help="scheduler patience")
parser.add_argument('--patience',
type=int,
default=3,
help='early stopping epochs')
parser.add_argument(
'--stop_based',
type=str,
default='train_mask',
choices=['val_total', 'val_mask', 'train_total', 'train_mask'],
help='early stopping criteria')
parser.add_argument('--return_best',
action='store_true',
help='return best model')
parser.add_argument('--test_for_changed',
action='store_true',
help='test pretrained model in changed sensor network')
parser.add_argument('--read_only', action='store_true', help='read only')
parser.add_argument('--inference_only',
action='store_true',
help='inference only')
parser.add_argument('--pretrained_model_path',
default=None,
help='path to pretrained model')
parser.add_argument('--batch_size', type=int, default=32)
parser.add_argument('--no_scale', action='store_true')
parser.add_argument('--num_workers', type=int, default=2)
# task
parser.add_argument('--input_len', type=int, required=True)
parser.add_argument('--pred_len', type=int, required=True)
parser.add_argument('--look_back', type=int, default=0)
parser.add_argument('--slide_step',
type=int,
default=1,
help='step for sliding window')
parser.add_argument('--target_features',
type=str,
required=True,
choices=['q', 'o', 'v', 'qv', 'qov'],
help='q is flow, v is speed, o is occupancy')
parser.add_argument('--input_features',
type=str,
required=True,
choices=['q', 'o', 'v', 'qv', 'qov'],
help='q is flow, v is speed, o is occupancy')
parser.add_argument('--train_ratio',
type=float,
default=0.6,
help='train ratio')
parser.add_argument('--valid_ratio',
type=float,
default=0.2,
help='validation ratio')
parser.add_argument(
'--inverse',
action='store_true',
help='inverse the data for better results presentation')
# save and logging settings
parser.add_argument('--wandb',
action='store_true',
help='enable wandb logging')
parser.add_argument('--wandb_entity',
type=str,
default='zqslalala',
help='user name of wandb')
parser.add_argument('--best_checkpoint_path',
type=str,
default='./check_points/',
help='best val checkpoint path')
parser.add_argument('--best_prediction_path',
type=str,
default='./predictions/',
help='results path')
parser.add_argument('--best_metrics_path',
type=str,
default='./',
help='results path')
parser.add_argument('--best_fig_path',
type=str,
default='./figures/',
help='figures path')
parser.add_argument('--log_dir',
type=str,
default='./logs/',
help='log path')
parser.add_argument('--verbose', action='store_true', help='verbose')
parser.add_argument('--log_interval', type=int, default=100)
parser.add_argument('--save_model',
action='store_true',
help='save model not model state dict')
# dataset
parser.add_argument(
'--unknown_nodes_path',
type=str,
help='path for unobserved sensors id in the train data')
parser.add_argument('--unknown_nodes_list',
type=str,
default="",
help='unobserved sensors in the train data')
parser.add_argument(
'--unknown_nodes_num',
type=int,
default=0,
help='number of unobserved sensors in the training data')
parser.add_argument('--ori_unknown_nodes_list',
type=str,
default="",
help='unobserved sensors in the original data')
parser.add_argument(
'--dloader_name',
type=str,
default='METRLA',
choices=['METRLA', 'PEMSBAY', 'METRLA-dynamic', 'PEMSBAY-dynamic'],
help='dynamic denotes the scenario where the scensor is changing')
parser.add_argument('--data_path', type=str)
parser.add_argument('--freq', type=str, default='min')
args = parser.parse_args()
args.unknown_nodes_list = [
int(x) for x in args.unknown_nodes_list.split(',') if x
]
args.ori_unknown_nodes_list = [
int(x) for x in args.ori_unknown_nodes_list.split(',') if x
]
args.unknown_nodes_list = [
x for x in args.unknown_nodes_list
if x not in args.ori_unknown_nodes_list
]
assert args.input_features in args.target_features, 'input_features should be in target_features'
args.c_out = len(args.target_features)
args.enc_in = len(args.input_features)
args.enc_mark_in = 3 * (args.t_mark) + 2 * (args.pos_mark)
print(
f'enc_in enc_mark_in, and c_out is set automatically based on input_features,target_features and t_mark,pos_mark',
)
args.num_experts = 1 * (args.mean_expert) + 1 * (
args.weight_expert) + 1 * (args.max_expert) + 1 * (
args.min_expert) + 1 * (args.diffusion_expert)
return args