Hugging Face's logo

poloclub
/
robust-principles

Model card Files Community
AnthonyPeng commited on
Commit
9d14f67
·
1 Parent(s): 67325b2

Upload AA CIFAR-10 evaluation log

Browse files
Files changed (1) hide show
  1. eval_log.txt +139 -0
eval_log.txt ADDED
@@ -0,0 +1,139 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ../data/cifar-10-python.tar[85/5761]
2
+ 100%|█████████████████████████████████████████████████████| 170498071/170498071 [00:01<00:00, 86462452.39it
3
+ /s]
4
+ Extracting ../data/cifar-10-python.tar.gz to ../data
5
+ Files already downloaded and verified
6
+ Files already downloaded and verified
7
+ Clean accuracy: 93.27%
8
+ setting parameters for standard version
9
+ using standard version including apgd-ce, apgd-t, fab-t, square.
10
+ initial accuracy: 93.27%
11
+ apgd-ce - 1/37 - 53 out of 256 successfully perturbed
12
+ apgd-ce - 2/37 - 53 out of 256 successfully perturbed
13
+ apgd-ce - 3/37 - 55 out of 256 successfully perturbed
14
+ apgd-ce - 4/37 - 58 out of 256 successfully perturbed
15
+ apgd-ce - 5/37 - 45 out of 256 successfully perturbed
16
+ apgd-ce - 6/37 - 55 out of 256 successfully perturbed
17
+ apgd-ce - 7/37 - 45 out of 256 successfully perturbed
18
+ apgd-ce - 8/37 - 56 out of 256 successfully perturbed
19
+ apgd-ce - 9/37 - 49 out of 256 successfully perturbed
20
+ apgd-ce - 10/37 - 54 out of 256 successfully perturbed
21
+ apgd-ce - 11/37 - 42 out of 256 successfully perturbed
22
+ apgd-ce - 12/37 - 47 out of 256 successfully perturbed
23
+ apgd-ce - 13/37 - 57 out of 256 successfully perturbed
24
+ apgd-ce - 14/37 - 53 out of 256 successfully perturbed
25
+ apgd-ce - 15/37 - 60 out of 256 successfully perturbed
26
+ apgd-ce - 16/37 - 41 out of 256 successfully perturbed
27
+ apgd-ce - 17/37 - 60 out of 256 successfully perturbed
28
+ apgd-ce - 18/37 - 57 out of 256 successfully perturbed
29
+ apgd-ce - 19/37 - 64 out of 256 successfully perturbed
30
+ apgd-ce - 20/37 - 48 out of 256 successfully perturbed
31
+ apgd-ce - 21/37 - 43 out of 256 successfully perturbed
32
+ apgd-ce - 22/37 - 69 out of 256 successfully perturbed
33
+ apgd-ce - 23/37 - 53 out of 256 successfully perturbed
34
+ apgd-ce - 24/37 - 44 out of 256 successfully perturbed
35
+ apgd-ce - 25/37 - 58 out of 256 successfully perturbed
36
+ apgd-ce - 26/37 - 62 out of 256 successfully perturbed
37
+ apgd-ce - 27/37 - 57 out of 256 successfully perturbed
38
+ apgd-ce - 28/37 - 60 out of 256 successfully perturbed
39
+ apgd-ce - 29/37 - 50 out of 256 successfully perturbed
40
+ apgd-ce - 30/37 - 53 out of 256 successfully perturbed
41
+ apgd-ce - 31/37 - 57 out of 256 successfully perturbed
42
+ apgd-ce - 32/37 - 56 out of 256 successfully perturbed
43
+ apgd-ce - 33/37 - 48 out of 256 successfully perturbed
44
+ apgd-ce - 34/37 - 48 out of 256 successfully perturbed
45
+ apgd-ce - 35/37 - 54 out of 256 successfully perturbed
46
+ apgd-ce - 36/37 - 54 out of 256 successfully perturbed
47
+ apgd-ce - 37/37 - 19 out of 111 successfully perturbed
48
+ robust accuracy after APGD-CE: 73.90% (total time 1815.9 s)
49
+ apgd-t - 1/29 - 6 out of 256 successfully perturbed
50
+ apgd-t - 2/29 - 11 out of 256 successfully perturbed
51
+ apgd-t - 3/29 - 5 out of 256 successfully perturbed
52
+ apgd-t - 4/29 - 6 out of 256 successfully perturbed
53
+ apgd-t - 5/29 - 9 out of 256 successfully perturbed
54
+ apgd-t - 6/29 - 10 out of 256 successfully perturbed
55
+ apgd-t - 7/29 - 4 out of 256 successfully perturbed
56
+ apgd-t - 8/29 - 16 out of 256 successfully perturbed
57
+ apgd-t - 9/29 - 7 out of 256 successfully perturbed
58
+ apgd-t - 10/29 - 11 out of 256 successfully perturbed
59
+ apgd-t - 11/29 - 9 out of 256 successfully perturbed
60
+ apgd-t - 12/29 - 7 out of 256 successfully perturbed
61
+ apgd-t - 13/29 - 8 out of 256 successfully perturbed
62
+ apgd-t - 14/29 - 10 out of 256 successfully perturbed
63
+ apgd-t - 15/29 - 11 out of 256 successfully perturbed
64
+ apgd-t - 16/29 - 13 out of 256 successfully perturbed
65
+ apgd-t - 17/29 - 20 out of 256 successfully perturbed
66
+ apgd-t - 18/29 - 12 out of 256 successfully perturbed
67
+ apgd-t - 19/29 - 11 out of 256 successfully perturbed
68
+ apgd-t - 20/29 - 14 out of 256 successfully perturbed
69
+ apgd-t - 21/29 - 15 out of 256 successfully perturbed
70
+ apgd-t - 22/29 - 5 out of 256 successfully perturbed
71
+ apgd-t - 23/29 - 5 out of 256 successfully perturbed
72
+ apgd-t - 24/29 - 8 out of 256 successfully perturbed
73
+ apgd-t - 25/29 - 17 out of 256 successfully perturbed
74
+ apgd-t - 26/29 - 12 out of 256 successfully perturbed
75
+ apgd-t - 27/29 - 8 out of 256 successfully perturbed
76
+ apgd-t - 28/29 - 6 out of 256 successfully perturbed
77
+ apgd-t - 29/29 - 6 out of 222 successfully perturbed
78
+ robust accuracy after APGD-T: 71.08% (total time 14360.6 s)
79
+ fab-t - 1/28 - 0 out of 256 successfully perturbed
80
+ fab-t - 2/28 - 0 out of 256 successfully perturbed
81
+ fab-t - 3/28 - 0 out of 256 successfully perturbed
82
+ fab-t - 4/28 - 0 out of 256 successfully perturbed
83
+ fab-t - 5/28 - 0 out of 256 successfully perturbed
84
+ fab-t - 6/28 - 0 out of 256 successfully perturbed
85
+ fab-t - 7/28 - 0 out of 256 successfully perturbed
86
+ fab-t - 8/28 - 0 out of 256 successfully perturbed
87
+ fab-t - 9/28 - 0 out of 256 successfully perturbed
88
+ fab-t - 10/28 - 0 out of 256 successfully perturbed
89
+ fab-t - 11/28 - 1 out of 256 successfully perturbed
90
+ fab-t - 12/28 - 0 out of 256 successfully perturbed
91
+ fab-t - 13/28 - 0 out of 256 successfully perturbed
92
+ fab-t - 14/28 - 0 out of 256 successfully perturbed
93
+ fab-t - 15/28 - 0 out of 256 successfully perturbed
94
+ fab-t - 16/28 - 0 out of 256 successfully perturbed
95
+ fab-t - 17/28 - 0 out of 256 successfully perturbed
96
+ fab-t - 18/28 - 0 out of 256 successfully perturbed
97
+ fab-t - 19/28 - 0 out of 256 successfully perturbed
98
+ fab-t - 20/28 - 0 out of 256 successfully perturbed
99
+ fab-t - 21/28 - 0 out of 256 successfully perturbed
100
+ fab-t - 22/28 - 0 out of 256 successfully perturbed
101
+ fab-t - 23/28 - 0 out of 256 successfully perturbed
102
+ fab-t - 24/28 - 0 out of 256 successfully perturbed
103
+ fab-t - 25/28 - 0 out of 256 successfully perturbed
104
+ fab-t - 26/28 - 0 out of 256 successfully perturbed
105
+ fab-t - 27/28 - 0 out of 256 successfully perturbed
106
+ fab-t - 28/28 - 0 out of 196 successfully perturbed
107
+ robust accuracy after FAB-T: 71.07% (total time 37176.5 s)
108
+ square - 1/28 - 0 out of 256 successfully perturbed
109
+ square - 2/28 - 0 out of 256 successfully perturbed
110
+ square - 3/28 - 0 out of 256 successfully perturbed
111
+ square - 4/28 - 0 out of 256 successfully perturbed
112
+ square - 5/28 - 0 out of 256 successfully perturbed
113
+ square - 6/28 - 0 out of 256 successfully perturbed
114
+ square - 7/28 - 0 out of 256 successfully perturbed
115
+ square - 8/28 - 0 out of 256 successfully perturbed
116
+ square - 9/28 - 0 out of 256 successfully perturbed
117
+ square - 10/28 - 0 out of 256 successfully perturbed
118
+ square - 11/28 - 0 out of 256 successfully perturbed
119
+ square - 12/28 - 0 out of 256 successfully perturbed
120
+ square - 13/28 - 0 out of 256 successfully perturbed
121
+ square - 14/28 - 0 out of 256 successfully perturbed
122
+ square - 15/28 - 0 out of 256 successfully perturbed
123
+ square - 16/28 - 0 out of 256 successfully perturbed
124
+ square - 17/28 - 0 out of 256 successfully perturbed
125
+ square - 18/28 - 0 out of 256 successfully perturbed
126
+ square - 19/28 - 0 out of 256 successfully perturbed
127
+ square - 20/28 - 0 out of 256 successfully perturbed
128
+ square - 21/28 - 0 out of 256 successfully perturbed
129
+ square - 22/28 - 0 out of 256 successfully perturbed
130
+ square - 23/28 - 0 out of 256 successfully perturbed
131
+ square - 24/28 - 0 out of 256 successfully perturbed
132
+ square - 25/28 - 0 out of 256 successfully perturbed
133
+ square - 26/28 - 0 out of 256 successfully perturbed
134
+ square - 27/28 - 0 out of 256 successfully perturbed
135
+ square - 28/28 - 0 out of 195 successfully perturbed
136
+ robust accuracy after SQUARE: 71.07% (total time 84616.4 s)
137
+ max Linf perturbation: 0.03137, nan in tensor: 0, max: 1.00000, min: 0.00000
138
+ robust accuracy: 71.07%
139
+ Adversarial accuracy: 71.07%