-
Notifications
You must be signed in to change notification settings - Fork 28
/
Copy pathgenotypes.py
148 lines (134 loc) · 13 KB
/
genotypes.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
from collections import namedtuple
Genotype = namedtuple('Genotype', 'normal normal_concat reduce reduce_concat')
PRIMITIVES = [
'none',
'max_pool_3x3',
'avg_pool_3x3',
'skip_connect',
'sep_conv_3x3',
'sep_conv_5x5',
'dil_conv_3x3',
'dil_conv_5x5'
]
NASNet = Genotype(
normal = [
('sep_conv_5x5', 1),
('sep_conv_3x3', 0),
('sep_conv_5x5', 0),
('sep_conv_3x3', 0),
('avg_pool_3x3', 1),
('skip_connect', 0),
('avg_pool_3x3', 0),
('avg_pool_3x3', 0),
('sep_conv_3x3', 1),
('skip_connect', 1),
],
normal_concat = [2, 3, 4, 5, 6],
reduce = [
('sep_conv_5x5', 1),
('sep_conv_7x7', 0),
('max_pool_3x3', 1),
('sep_conv_7x7', 0),
('avg_pool_3x3', 1),
('sep_conv_5x5', 0),
('skip_connect', 3),
('avg_pool_3x3', 2),
('sep_conv_3x3', 2),
('max_pool_3x3', 1),
],
reduce_concat = [4, 5, 6],
)
AmoebaNet = Genotype(
normal = [
('avg_pool_3x3', 0),
('max_pool_3x3', 1),
('sep_conv_3x3', 0),
('sep_conv_5x5', 2),
('sep_conv_3x3', 0),
('avg_pool_3x3', 3),
('sep_conv_3x3', 1),
('skip_connect', 1),
('skip_connect', 0),
('avg_pool_3x3', 1),
],
normal_concat = [4, 5, 6],
reduce = [
('avg_pool_3x3', 0),
('sep_conv_3x3', 1),
('max_pool_3x3', 0),
('sep_conv_7x7', 2),
('sep_conv_7x7', 0),
('avg_pool_3x3', 1),
('max_pool_3x3', 0),
('max_pool_3x3', 1),
('conv_7x1_1x7', 0),
('sep_conv_3x3', 5),
],
reduce_concat = [3, 4, 6]
)
# **************************** SGAS CRITERION 1 ****************************** #
# Experiment Validation error (%) Params (M) Test error (%) Evaluation ranking
# Cri1_CIFAR_1 16.94 3.75 2.44 2
Cri1_CIFAR_1 = Genotype(normal=[('skip_connect', 0), ('sep_conv_3x3', 1), ('sep_conv_3x3', 0), ('sep_conv_5x5', 1), ('sep_conv_3x3', 1), ('sep_conv_3x3', 2), ('dil_conv_5x5', 2), ('dil_conv_3x3', 3)], normal_concat=range(2, 6), reduce=[('max_pool_3x3', 0), ('dil_conv_3x3', 1), ('sep_conv_3x3', 0), ('sep_conv_5x5', 2), ('max_pool_3x3', 1), ('skip_connect', 2), ('skip_connect', 2), ('skip_connect', 3)], reduce_concat=range(2, 6))
# Cri1_CIFAR_2 17.33 3.73 2.50 3
Cri1_CIFAR_2 = Genotype(normal=[('sep_conv_3x3', 0), ('dil_conv_3x3', 1), ('skip_connect', 0), ('dil_conv_3x3', 2), ('sep_conv_3x3', 1), ('sep_conv_3x3', 2), ('sep_conv_3x3', 2), ('sep_conv_5x5', 4)], normal_concat=range(2, 6), reduce=[('avg_pool_3x3', 0), ('sep_conv_3x3', 1), ('max_pool_3x3', 0), ('skip_connect', 2), ('max_pool_3x3', 0), ('sep_conv_5x5', 2), ('max_pool_3x3', 1), ('dil_conv_5x5', 3)], reduce_concat=range(2, 6))
# Cri1_CIFAR_3 17.90 3.80 2.39 1
Cri1_CIFAR_3 = Genotype(normal=[('sep_conv_3x3', 0), ('dil_conv_3x3', 1), ('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_5x5', 1), ('sep_conv_5x5', 3), ('skip_connect', 0), ('dil_conv_5x5', 2)], normal_concat=range(2, 6), reduce=[('max_pool_3x3', 0), ('dil_conv_5x5', 1), ('sep_conv_3x3', 0), ('skip_connect', 2), ('sep_conv_3x3', 1), ('skip_connect', 2), ('max_pool_3x3', 0), ('max_pool_3x3', 1)], reduce_concat=range(2, 6))
# Cri1_CIFAR_4 17.90 3.32 2.63 6
Cri1_CIFAR_4 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('skip_connect', 0), ('skip_connect', 2), ('sep_conv_3x3', 1), ('sep_conv_3x3', 2), ('dil_conv_3x3', 2), ('dil_conv_5x5', 4)], normal_concat=range(2, 6), reduce=[('avg_pool_3x3', 0), ('sep_conv_5x5', 1), ('avg_pool_3x3', 0), ('dil_conv_5x5', 2), ('max_pool_3x3', 0), ('dil_conv_5x5', 2), ('max_pool_3x3', 0), ('skip_connect', 2)], reduce_concat=range(2, 6))
# Cri1_CIFAR_5 17.99 3.45 2.78 8
Cri1_CIFAR_5 = Genotype(normal=[('dil_conv_5x5', 0), ('sep_conv_3x3', 1), ('skip_connect', 0), ('sep_conv_3x3', 1), ('skip_connect', 0), ('sep_conv_3x3', 2), ('dil_conv_5x5', 2), ('sep_conv_5x5', 3)], normal_concat=range(2, 6), reduce=[('max_pool_3x3', 0), ('sep_conv_5x5', 1), ('max_pool_3x3', 0), ('skip_connect', 2), ('sep_conv_3x3', 2), ('sep_conv_5x5', 3), ('avg_pool_3x3', 0), ('skip_connect', 2)], reduce_concat=range(2, 6))
# Cri1_CIFAR_6 18.43 3.47 2.68 7
Cri1_CIFAR_6 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('dil_conv_3x3', 1), ('dil_conv_3x3', 2), ('sep_conv_3x3', 1), ('dil_conv_5x5', 2), ('skip_connect', 1), ('sep_conv_3x3', 2)], normal_concat=range(2, 6), reduce=[('avg_pool_3x3', 0), ('max_pool_3x3', 1), ('max_pool_3x3', 1), ('dil_conv_5x5', 2), ('max_pool_3x3', 0), ('dil_conv_3x3', 3), ('skip_connect', 2), ('dil_conv_5x5', 3)], reduce_concat=range(2, 6))
# Cri1_CIFAR_7 18.72 3.83 2.51 4
Cri1_CIFAR_7 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_5x5', 1), ('skip_connect', 0), ('sep_conv_5x5', 2), ('sep_conv_3x3', 2), ('dil_conv_5x5', 3), ('sep_conv_3x3', 0), ('dil_conv_3x3', 2)], normal_concat=range(2, 6), reduce=[('dil_conv_3x3', 0), ('avg_pool_3x3', 1), ('max_pool_3x3', 0), ('sep_conv_3x3', 2), ('max_pool_3x3', 1), ('dil_conv_5x5', 3), ('avg_pool_3x3', 0), ('sep_conv_5x5', 2)], reduce_concat=range(2, 6))
# Cri1_CIFAR_8 19.82 3.66 2.61 5
Cri1_CIFAR_8 = Genotype(normal=[('sep_conv_5x5', 0), ('sep_conv_3x3', 1), ('dil_conv_3x3', 1), ('dil_conv_3x3', 2), ('skip_connect', 0), ('sep_conv_3x3', 1), ('dil_conv_5x5', 1), ('sep_conv_3x3', 2)], normal_concat=range(2, 6), reduce=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('max_pool_3x3', 1), ('skip_connect', 2), ('max_pool_3x3', 1), ('sep_conv_5x5', 2), ('max_pool_3x3', 0), ('sep_conv_5x5', 3)], reduce_concat=range(2, 6))
# Cri1_CIFAR_9 19.93 3.98 3.18 10
Cri1_CIFAR_9 = Genotype(normal=[('sep_conv_5x5', 0), ('sep_conv_5x5', 1), ('sep_conv_3x3', 1), ('dil_conv_3x3', 2), ('sep_conv_3x3', 1), ('dil_conv_5x5', 3), ('sep_conv_3x3', 1), ('dil_conv_3x3', 3)], normal_concat=range(2, 6), reduce=[('avg_pool_3x3', 0), ('sep_conv_3x3', 1), ('max_pool_3x3', 0), ('dil_conv_5x5', 2), ('max_pool_3x3', 0), ('sep_conv_3x3', 2), ('max_pool_3x3', 1), ('skip_connect', 2)], reduce_concat=range(2, 6))
# Cri1_CIFAR_10 21.53 3.61 2.87 9
Cri1_CIFAR_10 = Genotype(normal=[('sep_conv_3x3', 0), ('dil_conv_5x5', 1), ('sep_conv_3x3', 0), ('dil_conv_3x3', 2), ('sep_conv_3x3', 1), ('dil_conv_3x3', 3), ('dil_conv_3x3', 2), ('dil_conv_3x3', 4)], normal_concat=range(2, 6), reduce=[('max_pool_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_3x3', 0), ('sep_conv_5x5', 2), ('max_pool_3x3', 0), ('max_pool_3x3', 1), ('max_pool_3x3', 0), ('sep_conv_3x3', 2)], reduce_concat=range(2, 6))
Cri1_CIFAR_Best = Cri1_CIFAR_3
# -------------------------------------------------------------------------------#
# Experiment Test error top-1 (%) Test error top-5 (%) Params (M) ×+
# Cri1_ImageNet_1 24.47 7.23 5.25 578
Cri1_ImageNet_1 = Cri1_CIFAR_1
# Cri1_ImageNet_2 24.53 7.40 5.23 574
Cri1_ImageNet_2 = Cri1_CIFAR_2
# Cri1_ImageNet_3 24.22 7.25 5.29 585
Cri1_ImageNet_3 = Cri1_CIFAR_3
Cri1_ImageNet_Best = Cri1_ImageNet_3
# **************************** SGAS CRITERION 1 ****************************** #
# **************************** SGAS CRITERION 2 ****************************** #
# Experiment Validation error (%) Params (M) Test error (%) Evaluation ranking
# Cri2_CIFAR_1 16.48 4.14 2.57 4
Cri2_CIFAR_1 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_5x5', 0), ('sep_conv_5x5', 1), ('skip_connect', 0), ('sep_conv_5x5', 2), ('sep_conv_3x3', 1), ('dil_conv_5x5', 2)], normal_concat=range(2, 6), reduce=[('max_pool_3x3', 0), ('sep_conv_5x5', 1), ('max_pool_3x3', 1), ('sep_conv_3x3', 2), ('max_pool_3x3', 0), ('sep_conv_3x3', 3), ('sep_conv_3x3', 0), ('dil_conv_5x5', 2)], reduce_concat=range(2, 6))
# Cri2_CIFAR_2 17.26 3.88 2.60 6
Cri2_CIFAR_2 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_3x3', 0), ('skip_connect', 2), ('sep_conv_5x5', 1), ('sep_conv_3x3', 2), ('dil_conv_5x5', 1), ('sep_conv_3x3', 2)], normal_concat=range(2, 6), reduce=[('avg_pool_3x3', 0), ('max_pool_3x3', 1), ('max_pool_3x3', 0), ('sep_conv_5x5', 2), ('max_pool_3x3', 0), ('dil_conv_5x5', 2), ('avg_pool_3x3', 0), ('max_pool_3x3', 1)], reduce_concat=range(2, 6))
# Cri2_CIFAR_3 17.31 4.09 2.44 1
Cri2_CIFAR_3 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_3x3', 0), ('sep_conv_3x3', 2), ('max_pool_3x3', 0), ('sep_conv_3x3', 3), ('sep_conv_3x3', 2), ('sep_conv_3x3', 3)], normal_concat=range(2, 6), reduce=[('sep_conv_3x3', 0), ('max_pool_3x3', 1), ('dil_conv_3x3', 0), ('max_pool_3x3', 1), ('max_pool_3x3', 0), ('sep_conv_3x3', 2), ('max_pool_3x3', 0), ('sep_conv_5x5', 2)], reduce_concat=range(2, 6))
# Cri2_CIFAR_4 17.47 3.91 2.49 2
Cri2_CIFAR_4 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('dil_conv_5x5', 0), ('dil_conv_3x3', 2), ('sep_conv_3x3', 1), ('dil_conv_5x5', 3)], normal_concat=range(2, 6), reduce=[('sep_conv_5x5', 0), ('dil_conv_3x3', 1), ('max_pool_3x3', 0), ('max_pool_3x3', 1), ('dil_conv_3x3', 0), ('dil_conv_5x5', 2), ('max_pool_3x3', 0), ('avg_pool_3x3', 1)], reduce_concat=range(2, 6))
# Cri2_CIFAR_5 17.53 3.69 2.52 3
Cri2_CIFAR_5 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_5x5', 1), ('skip_connect', 0), ('sep_conv_5x5', 1), ('dil_conv_3x3', 1), ('dil_conv_3x3', 2), ('sep_conv_3x3', 2), ('dil_conv_5x5', 3)], normal_concat=range(2, 6), reduce=[('sep_conv_5x5', 0), ('skip_connect', 1), ('max_pool_3x3', 1), ('sep_conv_5x5', 2), ('max_pool_3x3', 0), ('sep_conv_5x5', 3), ('max_pool_3x3', 0), ('skip_connect', 2)], reduce_concat=range(2, 6))
# Cri2_CIFAR_6 17.98 3.95 3.12 10
Cri2_CIFAR_6 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_3x3', 0), ('dil_conv_5x5', 2), ('sep_conv_3x3', 1), ('sep_conv_3x3', 3), ('dil_conv_5x5', 2), ('dil_conv_3x3', 3)], normal_concat=range(2, 6), reduce=[('sep_conv_5x5', 0), ('sep_conv_5x5', 1), ('max_pool_3x3', 0), ('max_pool_3x3', 1), ('avg_pool_3x3', 0), ('sep_conv_5x5', 3), ('max_pool_3x3', 0), ('skip_connect', 2)], reduce_concat=range(2, 6))
# Cri2_CIFAR_7 18.28 3.69 2.58 5
Cri2_CIFAR_7 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('skip_connect', 1), ('dil_conv_3x3', 2), ('sep_conv_5x5', 1), ('dil_conv_5x5', 3)], normal_concat=range(2, 6), reduce=[('max_pool_3x3', 0), ('avg_pool_3x3', 1), ('max_pool_3x3', 1), ('dil_conv_5x5', 2), ('max_pool_3x3', 1), ('skip_connect', 2), ('max_pool_3x3', 0), ('sep_conv_3x3', 2)], reduce_concat=range(2, 6))
# Cri2_CIFAR_8 18.28 4.33 2.85 8
Cri2_CIFAR_8 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_5x5', 1), ('sep_conv_3x3', 0), ('sep_conv_3x3', 1), ('sep_conv_5x5', 1), ('sep_conv_5x5', 2), ('sep_conv_3x3', 1), ('dil_conv_5x5', 3)], normal_concat=range(2, 6), reduce=[('avg_pool_3x3', 0), ('skip_connect', 1), ('max_pool_3x3', 0), ('skip_connect', 2), ('max_pool_3x3', 0), ('sep_conv_5x5', 2), ('max_pool_3x3', 1), ('skip_connect', 2)], reduce_concat=range(2, 6))
# Cri2_CIFAR_9 19.48 3.73 2.85 9
Cri2_CIFAR_9 = Genotype(normal=[('max_pool_3x3', 0), ('sep_conv_5x5', 1), ('sep_conv_5x5', 0), ('sep_conv_5x5', 1), ('dil_conv_3x3', 1), ('dil_conv_3x3', 3), ('sep_conv_3x3', 1), ('dil_conv_3x3', 3)], normal_concat=range(2, 6), reduce=[('sep_conv_3x3', 0), ('sep_conv_5x5', 1), ('max_pool_3x3', 0), ('dil_conv_5x5', 2), ('dil_conv_5x5', 2), ('sep_conv_3x3', 3), ('avg_pool_3x3', 0), ('max_pool_3x3', 1)], reduce_concat=range(2, 6))
# Cri2_CIFAR_10 19.98 3.68 2.66 7
Cri2_CIFAR_10 = Genotype(normal=[('sep_conv_3x3', 0), ('sep_conv_5x5', 1), ('sep_conv_5x5', 1), ('dil_conv_5x5', 2), ('sep_conv_3x3', 2), ('dil_conv_5x5', 3), ('max_pool_3x3', 0), ('dil_conv_3x3', 3)], normal_concat=range(2, 6), reduce=[('avg_pool_3x3', 0), ('skip_connect', 1), ('sep_conv_5x5', 0), ('skip_connect', 2), ('max_pool_3x3', 0), ('sep_conv_5x5', 2), ('max_pool_3x3', 1), ('dil_conv_3x3', 2)], reduce_concat=range(2, 6))
Cri2_CIFAR_Best = Cri2_CIFAR_3
# -------------------------------------------------------------------------------#
# Experiment Test error top-1 (%) Test error top-5 (%) Params (M) ×+
# Cri2_ImageNet_1 24.44 7.41 5.70 621
Cri2_ImageNet_1 = Cri2_CIFAR_3
# Cri2_ImageNet_2 24.13 7.31 5.44 598
Cri2_ImageNet_2 = Cri2_CIFAR_4
# Cri2_ImageNet_3 24.55 7.44 5.20 571
Cri2_ImageNet_3 = Cri2_CIFAR_5
Cri2_ImageNet_Best = Cri2_ImageNet_2
# **************************** SGAS CRITERION 2 ****************************** #