-
Notifications
You must be signed in to change notification settings - Fork 12
/
model_param_space.py
132 lines (120 loc) · 3.46 KB
/
model_param_space.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
import numpy as np
from hyperopt import hp
param_space_base = {
"wpe_size": hp.quniform("wpe_size", 5, 100, 5),
"lr": hp.qloguniform("lr", np.log(1e-4), np.log(1e-3), 1e-4),
"state_size": hp.quniform("state_size", 100, 500, 10),
"hidden_layers": 1,
"hidden_size": hp.quniform("hidden_size", 100, 500, 10),
"dense_keep_prob": hp.quniform("dense_keep_prob", 0.5, 1, 0.1),
"rnn_keep_prob": hp.quniform("rnn_keep_prob", 0.5, 1, 0.1),
"l2_reg_lambda": hp.quniform("l2_reg_lambda", 0, 1e-3, 1e-4),
"batch_size": 128,
"num_epochs": 20,
}
param_space_best_base = {
"wpe_size": 25,
"lr": 0.0005,
"state_size": 320,
"hidden_layers": 1,
"hidden_size": 190,
"dense_keep_prob": 0.9,
"rnn_keep_prob": 0.7,
"l2_reg_lambda": 0.0003,
"batch_size": 128,
"num_epochs": 30,
}
param_space_complex = {
"wpe_size": 25,
"lr": 0.0005,
"lr2": 1e-6,
"state_size": 320,
"hidden_layers": 1,
"hidden_size": 190,
"dense_keep_prob": 0.9,
"rnn_keep_prob": 0.7,
"l2_reg_lambda": 0.0003,
"batch_size": 128,
"num_epochs": 30,
"alpha": hp.quniform("alpha", 0.5, 0.9, 0.05),
"lambda1": hp.quniform("lambda1", 0.1, 2.0, 0.1),
"lambda2": hp.quniform("lambda2", 0.1, 2.0, 0.1),
}
param_space_best_complex = {
"wpe_size": 25,
"lr": 0.0005,
"lr2": 1e-6,
"state_size": 320,
"hidden_layers": 1,
"hidden_size": 190,
"dense_keep_prob": 0.9,
"rnn_keep_prob": 0.7,
"l2_reg_lambda": 0.0003,
"batch_size": 128,
"num_epochs": 30,
"alpha": 0.9,
"lambda1": 1.0,
"lambda2": 1.0,
}
param_space_real = {
"wpe_size": 25,
"lr": 0.0005,
"lr2": 1e-6,
"state_size": 320,
"hidden_layers": 1,
"hidden_size": 190,
"dense_keep_prob": 0.9,
"rnn_keep_prob": 0.7,
"l2_reg_lambda": 0.0003,
"batch_size": 128,
"num_epochs": 30,
"alpha": hp.quniform("alpha", 0.5, 0.9, 0.05),
"lambda1": hp.quniform("lambda1", 0.1, 2.0, 0.1),
"lambda2": hp.quniform("lambda2", 0.1, 2.0, 0.1),
}
param_space_best_real = {
"wpe_size": 25,
"lr": 0.0005,
"lr2": 1e-6,
"state_size": 320,
"hidden_layers": 1,
"hidden_size": 190,
"dense_keep_prob": 0.9,
"rnn_keep_prob": 0.7,
"l2_reg_lambda": 0.0003,
"batch_size": 128,
"num_epochs": 30,
"alpha": 0.9,
"lambda1": 1.0,
"lambda2": 1.0,
}
param_space_dict = {
"base": param_space_base,
"best_base": param_space_best_base,
"complex_hrere": param_space_complex,
"best_complex_hrere": param_space_best_complex,
"real_hrere": param_space_real,
"best_real_hrere": param_space_best_real,
}
int_params = [
"wpe_size", "state_size", "batch_size", "num_epochs",
"hidden_size", "hidden_layers",
]
class ModelParamSpace:
def __init__(self, learner_name):
s = "Invalid model name!"
assert learner_name in param_space_dict, s
self.learner_name = learner_name
def _build_space(self):
return param_space_dict[self.learner_name]
def _convert_into_param(self, param_dict):
if isinstance(param_dict, dict):
for k, v in param_dict.items():
if k in int_params:
param_dict[k] = int(v)
elif isinstance(v, list) or isinstance(v, tuple):
for i in range(len(v)):
self._convert_into_param(v[i])
elif isinstance(v, dict):
self._convert_into_param(v)
return param_dict