-
Notifications
You must be signed in to change notification settings - Fork 0
/
ExpCLIP_B32.txt
190 lines (189 loc) · 8.18 KB
/
ExpCLIP_B32.txt
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
Variable OMP_NUM_THREADS has been set to 8
Training date: 24-05-29 15:56
************************
workers = 8
epochs = 5
batch_size = 512
batch_size_test_image = 512
batch_size_test_video = 64
lr = 0.001
weight_decay = 0.0001
momentum = 0.9
print_freq = 10
milestones = [30]
seed = 1
job_id = 3596133
instruction = Please play the role of a facial action describer. Objectively describe the detailed facial actions of the person in the image.
load_model = CLIP_B32
************************
Loading checkpoint shards: 0%| | 0/5 [00:00<?, ?it/s]Loading checkpoint shards: 20%|██ | 1/5 [00:17<01:08, 17.16s/it]Loading checkpoint shards: 40%|████ | 2/5 [00:34<00:52, 17.53s/it]Loading checkpoint shards: 60%|██████ | 3/5 [00:53<00:36, 18.11s/it]Loading checkpoint shards: 80%|████████ | 4/5 [01:13<00:18, 18.65s/it]Loading checkpoint shards: 100%|██████████| 5/5 [01:24<00:00, 15.99s/it]Loading checkpoint shards: 100%|██████████| 5/5 [01:24<00:00, 16.90s/it]
********************0********************
Epoch: [0][ 0/95] Loss 6.4295 (6.4295) Accuracy 0.391 ( 0.391)
Epoch: [0][10/95] Loss 6.3365 (6.3875) Accuracy 0.195 ( 0.178)
Epoch: [0][20/95] Loss 6.2966 (6.3510) Accuracy 0.391 ( 0.205)
Epoch: [0][30/95] Loss 6.2216 (6.3206) Accuracy 0.586 ( 0.239)
Epoch: [0][40/95] Loss 6.1781 (6.2924) Accuracy 0.391 ( 0.257)
Epoch: [0][50/95] Loss 6.1391 (6.2652) Accuracy 0.391 ( 0.299)
Epoch: [0][60/95] Loss 6.0878 (6.2411) Accuracy 0.391 ( 0.333)
Epoch: [0][70/95] Loss 6.0382 (6.2163) Accuracy 0.586 ( 0.377)
Epoch: [0][80/95] Loss 5.9958 (6.1932) Accuracy 0.781 ( 0.444)
Epoch: [0][90/95] Loss 5.9534 (6.1699) Accuracy 0.977 ( 0.487)
The train accuracy: 0.508
An epoch time: 551.99s
********************1********************
Epoch: [1][ 0/95] Loss 5.9335 (5.9335) Accuracy 1.367 ( 1.367)
Epoch: [1][10/95] Loss 5.8979 (5.9258) Accuracy 0.781 ( 0.923)
Epoch: [1][20/95] Loss 5.8667 (5.9056) Accuracy 1.562 ( 1.070)
Epoch: [1][30/95] Loss 5.8007 (5.8884) Accuracy 2.148 ( 1.184)
Epoch: [1][40/95] Loss 5.7862 (5.8708) Accuracy 1.367 ( 1.272)
Epoch: [1][50/95] Loss 5.7604 (5.8527) Accuracy 1.953 ( 1.325)
Epoch: [1][60/95] Loss 5.7227 (5.8359) Accuracy 2.539 ( 1.409)
Epoch: [1][70/95] Loss 5.6631 (5.8179) Accuracy 1.562 ( 1.485)
Epoch: [1][80/95] Loss 5.6538 (5.7994) Accuracy 2.344 ( 1.558)
Epoch: [1][90/95] Loss 5.6351 (5.7813) Accuracy 2.734 ( 1.603)
The train accuracy: 1.653
An epoch time: 529.88s
********************2********************
Epoch: [2][ 0/95] Loss 5.5970 (5.5970) Accuracy 3.711 ( 3.711)
Epoch: [2][10/95] Loss 5.5712 (5.5805) Accuracy 2.148 ( 2.486)
Epoch: [2][20/95] Loss 5.4842 (5.5653) Accuracy 2.148 ( 2.586)
Epoch: [2][30/95] Loss 5.4541 (5.5455) Accuracy 2.930 ( 2.684)
Epoch: [2][40/95] Loss 5.4756 (5.5274) Accuracy 3.320 ( 2.749)
Epoch: [2][50/95] Loss 5.4358 (5.5090) Accuracy 2.734 ( 2.845)
Epoch: [2][60/95] Loss 5.3922 (5.4880) Accuracy 3.516 ( 2.894)
Epoch: [2][70/95] Loss 5.3402 (5.4675) Accuracy 4.688 ( 3.020)
Epoch: [2][80/95] Loss 5.3263 (5.4487) Accuracy 4.688 ( 3.142)
Epoch: [2][90/95] Loss 5.2460 (5.4291) Accuracy 3.125 ( 3.213)
The train accuracy: 3.259
An epoch time: 525.98s
********************3********************
Epoch: [3][ 0/95] Loss 5.2462 (5.2462) Accuracy 4.492 ( 4.492)
Epoch: [3][10/95] Loss 5.1310 (5.2050) Accuracy 4.688 ( 4.794)
Epoch: [3][20/95] Loss 5.1730 (5.1842) Accuracy 5.859 ( 4.874)
Epoch: [3][30/95] Loss 5.0671 (5.1593) Accuracy 5.469 ( 5.028)
Epoch: [3][40/95] Loss 5.0216 (5.1349) Accuracy 6.641 ( 5.097)
Epoch: [3][50/95] Loss 4.9832 (5.1114) Accuracy 7.812 ( 5.224)
Epoch: [3][60/95] Loss 4.9513 (5.0872) Accuracy 5.664 ( 5.450)
Epoch: [3][70/95] Loss 4.8684 (5.0599) Accuracy 7.617 ( 5.620)
Epoch: [3][80/95] Loss 4.8130 (5.0321) Accuracy 8.789 ( 5.823)
Epoch: [3][90/95] Loss 4.7349 (5.0051) Accuracy 6.836 ( 6.042)
The train accuracy: 6.112
An epoch time: 527.53s
********************4********************
Epoch: [4][ 0/95] Loss 4.7617 (4.7617) Accuracy 8.984 ( 8.984)
Epoch: [4][10/95] Loss 4.7024 (4.6919) Accuracy 7.617 ( 8.274)
Epoch: [4][20/95] Loss 4.6241 (4.6667) Accuracy 8.203 ( 8.557)
Epoch: [4][30/95] Loss 4.5465 (4.6333) Accuracy 8.008 ( 8.795)
Epoch: [4][40/95] Loss 4.4851 (4.6033) Accuracy 9.961 ( 8.913)
Epoch: [4][50/95] Loss 4.3949 (4.5756) Accuracy 11.523 ( 9.161)
Epoch: [4][60/95] Loss 4.3670 (4.5465) Accuracy 9.961 ( 9.391)
Epoch: [4][70/95] Loss 4.3640 (4.5211) Accuracy 11.328 ( 9.645)
Epoch: [4][80/95] Loss 4.2706 (4.4937) Accuracy 9.766 ( 9.797)
Epoch: [4][90/95] Loss 4.2431 (4.4687) Accuracy 14.844 (10.094)
The train accuracy: 10.210
An epoch time: 529.80s
************************
load_model = CLIP_B32
job_id = 3596133
************************
************************************************************************** Zero-shot Prompt Type: Class Name
******************** Static FER Zero-shot Performance ********************
************************* RAFDB
UAR/WAR: 36.00/44.95
************************* AffectNet7
UAR/WAR: 29.26/29.27
************************* AffectNet8
UAR/WAR: 25.56/25.56
************************* FERPlus
UAR/WAR: 25.75/37.26
******************** Dynamic FER Zero-shot Performance ********************
************************* DFEW
UAR/WAR: 20.91/27.37
************************* FERV39k
UAR/WAR: 19.39/20.30
************************* MAFW
UAR/WAR: 16.05/20.10
************************* AFEW
UAR/WAR: 32.09/33.33
************************************************************************** Zero-shot Prompt Type: An Expression of Name
******************** Static FER Zero-shot Performance ********************
************************* RAFDB
UAR/WAR: 42.00/51.27
************************* AffectNet7
UAR/WAR: 33.01/33.01
************************* AffectNet8
UAR/WAR: 28.98/28.98
************************* FERPlus
UAR/WAR: 40.62/45.20
******************** Dynamic FER Zero-shot Performance ********************
************************* DFEW
UAR/WAR: 28.08/26.37
************************* FERV39k
UAR/WAR: 22.36/19.79
************************* MAFW
UAR/WAR: 19.59/22.66
************************* AFEW
UAR/WAR: 31.34/31.50
************************************************************************** Zero-shot Prompt Type: A Photo of A Face with An Expression of Name
******************** Static FER Zero-shot Performance ********************
************************* RAFDB
UAR/WAR: 38.68/45.05
************************* AffectNet7
UAR/WAR: 31.12/31.12
************************* AffectNet8
UAR/WAR: 27.41/27.41
************************* FERPlus
UAR/WAR: 39.36/36.98
******************** Dynamic FER Zero-shot Performance ********************
************************* DFEW
UAR/WAR: 25.11/29.41
************************* FERV39k
UAR/WAR: 21.94/24.14
************************* MAFW
UAR/WAR: 17.87/21.05
************************* AFEW
UAR/WAR: 31.64/33.60
************************************************************************** Zero-shot Prompt Type: Expression Ensemble Five
******************** Static FER Zero-shot Performance ********************
************************* RAFDB
UAR/WAR: 40.34/48.57
************************* AffectNet7
UAR/WAR: 33.49/33.50
************************* AffectNet8
UAR/WAR: 29.33/29.33
************************* FERPlus
UAR/WAR: 40.27/42.97
******************** Dynamic FER Zero-shot Performance ********************
************************* DFEW
UAR/WAR: 24.96/26.96
************************* FERV39k
UAR/WAR: 21.58/21.01
************************* MAFW
UAR/WAR: 18.92/23.02
************************* AFEW
UAR/WAR: 34.20/35.70
************************************************************************** Zero-shot Prompt Type: Expression Ensemble Ten
******************** Static FER Zero-shot Performance ********************
************************* RAFDB
UAR/WAR: 39.82/50.49
************************* AffectNet7
UAR/WAR: 33.98/33.98
************************* AffectNet8
UAR/WAR: 29.76/29.76
************************* FERPlus
UAR/WAR: 40.94/44.09
******************** Dynamic FER Zero-shot Performance ********************
************************* DFEW
UAR/WAR: 25.32/26.78
************************* FERV39k
UAR/WAR: 22.36/22.19
************************* MAFW
UAR/WAR: 18.75/23.77
************************* AFEW
UAR/WAR: 32.37/34.12