-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathIRM_results.out
198 lines (188 loc) · 14.1 KB
/
IRM_results.out
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
./data/wildcam_denoised/train_43
== Found 858 items
== Found 2 classes
train environment: 43
['/coyote', '/raccoon']
class_indices: [(0, '/coyote'), (1, '/raccoon')]
./data/wildcam_denoised/train_46
== Found 753 items
== Found 2 classes
train environment: 46
['/coyote', '/raccoon']
class_indices: [(0, '/coyote'), (1, '/raccoon')]
./data/wildcam_denoised/test
== Found 522 items
== Found 2 classes
['/coyote', '/raccoon']
working with 2 training environments:
env['images']: 858
env['labels']: 858
class distribution: 582 coyotes and 276 raccoons. baseline accuracy 0.68 (always coyote).
env['images']: 753
env['labels']: 753
class distribution: 512 coyotes and 241 raccoons. baseline accuracy 0.68 (always coyote).
x_test: 522
y_test: 522
test class distribution: 144 coyotes and 378 raccoons. baseline accuracy 0.28 (always coyote).
Using GPU - True
========================================IRM========================================
{'n_restarts': 5, 'steps': 121, 'n_classes': 2, 'fc_only': True, 'model_path': './models/', 'transform': {'train': Compose(
CenterCrop(size=(224, 224))
ToTensor()
Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
), 'test': Compose(
CenterCrop(size=(224, 224))
ToTensor()
Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
)}, 'loader_tr_args': {'batch_size': 100, 'num_workers': 1}, 'loader_te_args': {'batch_size': 100, 'num_workers': 1}, 'loader_sample_args': {'batch_size': 100, 'num_workers': 1}, 'optimizer_args': {'lr': 0.001, 'l2_regularizer_weight': 0.001, 'penalty_anneal_iters': 40, 'penalty_weight': 10000.0}}
Restart 0
step train nll train acc train penalty test nll test acc test prec test rec
0 0.84201 0.29020 0.03132 0.82703 0.26864 0.56522 0.03439
10 0.54066 0.67895 0.00259 0.95627 0.27970 0.00000 0.00000
20 0.41900 0.79070 0.01581 0.85620 0.26273 0.63636 0.01852
30 0.36187 0.81747 0.01000 0.83011 0.29818 0.61538 0.02116
40 0.32415 0.85461 0.01120 0.73086 0.34545 0.72340 0.17989
50 0.31500 0.85451 0.00757 0.71289 0.39894 0.75969 0.25926
60 0.30527 0.86227 0.00709 0.72102 0.38818 0.74783 0.22751
70 0.30908 0.85298 0.00517 0.73591 0.39152 0.75214 0.23280
80 0.30120 0.86253 0.00473 0.69224 0.44182 0.75862 0.29101
90 0.30720 0.85848 0.00371 0.67274 0.50591 0.79397 0.41799
100 0.31127 0.85567 0.00265 0.60255 0.62379 0.81481 0.58201
110 0.33108 0.84932 0.00228 0.57398 0.71530 0.82955 0.77249
120 0.33016 0.84830 0.00258 0.54449 0.75045 0.83065 0.81746
Restart 1
step train nll train acc train penalty test nll test acc test prec test rec
0 0.77001 0.40002 0.01066 0.71149 0.51606 0.73387 0.48148
10 0.50554 0.68160 0.00324 0.91477 0.26364 0.50000 0.01323
20 0.40135 0.81262 0.01788 0.72849 0.35894 0.67769 0.21693
30 0.35353 0.80601 0.00835 0.79722 0.29864 0.67797 0.10582
40 0.32162 0.84847 0.00843 0.66641 0.51682 0.78037 0.44180
50 0.31128 0.86299 0.00669 0.65222 0.52985 0.78862 0.51323
60 0.30963 0.85908 0.00444 0.61316 0.59833 0.80986 0.60847
70 0.31299 0.85294 0.00396 0.59488 0.66697 0.82428 0.68254
80 0.31454 0.85752 0.00210 0.52549 0.78894 0.84293 0.85185
90 0.33164 0.85114 0.00137 0.50592 0.80636 0.84500 0.89418
100 0.33392 0.84957 0.00206 0.49313 0.81985 0.84539 0.89683
110 0.33262 0.85043 0.00173 0.52783 0.78955 0.84733 0.88095
120 0.32581 0.85013 0.00243 0.55593 0.74197 0.83696 0.81481
Restart 2
step train nll train acc train penalty test nll test acc test prec test rec
0 0.68354 0.56027 0.00036 0.76839 0.29106 0.55618 0.26190
10 0.45686 0.71761 0.00968 0.76576 0.29530 0.64179 0.11376
20 0.36461 0.81921 0.01235 0.71850 0.36576 0.71739 0.17460
30 0.32438 0.83935 0.00871 0.66251 0.46167 0.74479 0.37831
40 0.30207 0.85187 0.00571 0.64449 0.51515 0.75983 0.46032
50 0.28921 0.86064 0.00511 0.67794 0.43152 0.74850 0.33069
60 0.28712 0.85942 0.00373 0.68062 0.43318 0.74706 0.33598
70 0.28430 0.86687 0.00328 0.68967 0.44318 0.75281 0.35450
80 0.29041 0.85785 0.00269 0.65703 0.51242 0.77119 0.48148
90 0.29091 0.86024 0.00172 0.60005 0.60864 0.78378 0.61376
100 0.30046 0.85759 0.00215 0.57720 0.69455 0.81214 0.74339
110 0.29921 0.85762 0.00205 0.55420 0.73576 0.81408 0.76455
120 0.30072 0.85294 0.00159 0.57015 0.68697 0.81159 0.74074
Restart 3
step train nll train acc train penalty test nll test acc test prec test rec
0 0.77014 0.41664 0.01233 0.73993 0.42076 0.71345 0.32275
10 0.50340 0.68309 0.00254 0.99198 0.26530 1.00000 0.00265
20 0.39765 0.83078 0.01902 0.77469 0.31455 0.96552 0.07407
30 0.34992 0.82967 0.00951 0.85256 0.27015 0.92308 0.03175
40 0.31512 0.85426 0.00957 0.71173 0.41636 0.92632 0.23280
50 0.30176 0.85508 0.00769 0.68903 0.44136 0.88710 0.29101
60 0.30162 0.85668 0.00526 0.66175 0.49576 0.86875 0.36772
70 0.30523 0.85203 0.00449 0.63369 0.54742 0.86341 0.46825
80 0.31269 0.84951 0.00393 0.59046 0.64773 0.84946 0.62698
90 0.32310 0.84647 0.00176 0.52499 0.77030 0.84416 0.85979
100 0.32483 0.84776 0.00226 0.51887 0.79970 0.84343 0.88360
110 0.32417 0.84360 0.00129 0.53481 0.78136 0.84987 0.83862
120 0.32188 0.85104 0.00263 0.54493 0.73030 0.84330 0.78307
Restart 4
step train nll train acc train penalty test nll test acc test prec test rec
0 0.85345 0.33348 0.04186 0.62254 0.69455 0.72979 0.90741
10 0.55439 0.68045 0.00360 0.84897 0.26364 0.50000 0.00794
20 0.44384 0.79003 0.01649 0.59896 0.74136 0.83761 0.77778
30 0.38544 0.79516 0.00860 0.70349 0.40818 0.81081 0.23810
40 0.34726 0.83438 0.00904 0.62724 0.51333 0.81095 0.43122
50 0.33030 0.83764 0.00791 0.62927 0.53424 0.80455 0.46825
60 0.33537 0.83369 0.00473 0.71459 0.41409 0.81081 0.23810
70 0.35069 0.79866 0.00255 0.82923 0.31576 0.77778 0.03704
80 0.37122 0.76631 0.00425 0.92717 0.27121 0.66667 0.00529
90 0.35889 0.78758 0.00288 0.88191 0.27955 0.71429 0.02646
100 0.34741 0.80489 0.00164 0.84057 0.30894 0.80952 0.04497
110 0.35773 0.76904 0.00235 0.86070 0.30061 0.75000 0.03175
120 0.35277 0.79150 0.00122 0.83655 0.30894 0.78261 0.04762
Final train acc (mean/std across restarts so far):
0.839 0.024
Final test acc (mean/std across restarts so far):
0.644 0.169
Final test precision:
0.7826
Final test recall:
0.0476
confusion matrix:
[[139 5]
[360 18]]
tn = 139, fp = 5, fn = 360, tp = 18
./data/wildcam_denoised/train_43
== Found 858 items
== Found 2 classes
train environment: 43
['/coyote', '/raccoon']
class_indices: [(0, '/coyote'), (1, '/raccoon')]
./data/wildcam_denoised/train_46
== Found 753 items
== Found 2 classes
train environment: 46
['/coyote', '/raccoon']
class_indices: [(0, '/coyote'), (1, '/raccoon')]
./data/wildcam_denoised/test
== Found 522 items
== Found 2 classes
['/coyote', '/raccoon']
working with 2 training environments:
env['images']: 858
env['labels']: 858
class distribution: 582 coyotes and 276 raccoons. baseline accuracy 0.68 (always coyote).
env['images']: 753
env['labels']: 753
class distribution: 512 coyotes and 241 raccoons. baseline accuracy 0.68 (always coyote).
x_test: 522
y_test: 522
test class distribution: 144 coyotes and 378 raccoons. baseline accuracy 0.28 (always coyote).
Using GPU - True
========================================IRM========================================
{'n_restarts': 1, 'steps': 121, 'n_classes': 2, 'fc_only': True, 'model_path': './models/', 'transform': {'train': Compose(
CenterCrop(size=(224, 224))
ToTensor()
Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
), 'test': Compose(
CenterCrop(size=(224, 224))
ToTensor()
Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
)}, 'loader_tr_args': {'batch_size': 100, 'num_workers': 1}, 'loader_te_args': {'batch_size': 100, 'num_workers': 1}, 'loader_sample_args': {'batch_size': 100, 'num_workers': 1}, 'optimizer_args': {'lr': 0.001, 'l2_regularizer_weight': 0.001, 'penalty_anneal_iters': 40, 'penalty_weight': 10000.0}}
Restart 0
step train nll train acc train penalty test nll test acc test prec test rec
0 0.84201 0.29020 0.03132 0.81487 0.28045 0.56522 0.03439
10 0.54066 0.67895 0.00259 0.98345 0.26197 0.00000 0.00000
20 0.41900 0.79070 0.01581 0.83753 0.28636 0.63636 0.01852
30 0.36187 0.81747 0.01000 0.85654 0.27455 0.61538 0.02116
40 0.32415 0.85461 0.01120 0.73950 0.33955 0.72340 0.17989
50 0.31500 0.85451 0.00757 0.70585 0.40485 0.75969 0.25926
60 0.30527 0.86227 0.00709 0.73972 0.35273 0.74783 0.22751
70 0.30908 0.85298 0.00517 0.70374 0.40924 0.75214 0.23280
80 0.30120 0.86253 0.00473 0.70319 0.42409 0.75862 0.29101
90 0.30720 0.85848 0.00371 0.67267 0.49409 0.79397 0.41799
100 0.31127 0.85567 0.00265 0.61584 0.58833 0.81481 0.58201
110 0.33108 0.84932 0.00228 0.55162 0.73303 0.82955 0.77249
120 0.33016 0.84830 0.00258 0.54170 0.75636 0.83065 0.81746
Final train acc (mean/std across restarts so far):
0.848 0.0
Final test acc (mean/std across restarts so far):
0.756 0.0
Final test precision:
0.8306
Final test recall:
0.8175
confusion matrix:
[[ 81 63]
[ 69 309]]
tn = 81, fp = 63, fn = 69, tp = 309