-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathmodel_performance.txt
233 lines (217 loc) · 13.6 KB
/
model_performance.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
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
VGG-16 -> 8k Dataset (Batch size-32) Trained from scratch -> 10M params
4 [2.5870352, 0.73944414, 0.92379546, 0.4375, 0.609375]
Iteration 4, source accuracy = 0.50850, target accuracy = 0.34935
Iteration 4, source accuracy = 0.09359, target accuracy = 0.04146
9 [2.3001287, 0.6663174, 0.8169056, 0.515625, 0.453125]
Iteration 9, source accuracy = 0.51013, target accuracy = 0.35156
Iteration 9, source accuracy = 0.11714, target accuracy = 0.04725
14 [2.9030704, 0.72558314, 1.0887437, 0.546875, 0.515625]
Iteration 14, source accuracy = 0.50600, target accuracy = 0.34854
Iteration 14, source accuracy = 0.06039, target accuracy = 0.03513
19 [2.5591428, 0.6745342, 0.9423043, 0.46875, 0.53125]
Iteration 19, source accuracy = 0.49950, target accuracy = 0.34231
Iteration 19, source accuracy = 0.00150, target accuracy = 0.00183
24 [2.819502, 0.6645438, 1.0774791, 0.546875, 0.546875]
Iteration 24, source accuracy = 0.49962, target accuracy = 0.34251
Iteration 24, source accuracy = 0.00744, target accuracy = 0.00366
29 [2.2795331, 0.69764733, 0.7909429, 0.46875, 0.5625]
Iteration 29, source accuracy = 0.50087, target accuracy = 0.34854
Iteration 29, source accuracy = 0.07247, target accuracy = 0.04872
34 [2.2584624, 0.6991379, 0.77966225, 0.46875, 0.5625]
Iteration 34, source accuracy = 0.49800, target accuracy = 0.46432
Iteration 34, source accuracy = 0.42349, target accuracy = 0.48622
39 [2.0976262, 0.6971944, 0.7002159, 0.53125, 0.59375]
Iteration 39, source accuracy = 0.48888, target accuracy = 0.55457
Iteration 39, source accuracy = 0.54491, target accuracy = 0.66505
44 [2.5285757, 0.6839916, 0.922292, 0.53125, 0.53125]
Iteration 44, source accuracy = 0.48800, target accuracy = 0.48362
Iteration 44, source accuracy = 0.44044, target accuracy = 0.53753
49 [2.5763657, 0.73201793, 0.9221739, 0.453125, 0.46875]
Iteration 49, source accuracy = 0.49300, target accuracy = 0.46111
Iteration 49, source accuracy = 0.40736, target accuracy = 0.48630
VGG-16 -> 5k Dataset (Batch size-32) Trained from scratch -> 10M params
4 [3.0941358, 0.73829037, 1.1779227, 0.515625, 0.46875]
Iteration 4, source accuracy = 0.49467, target accuracy = 0.51397
Iteration 4, source accuracy = 0.54637, target accuracy = 0.60046
9 [2.991872, 0.68256736, 1.1546524, 0.5625, 0.46875]
Iteration 9, source accuracy = 0.49688, target accuracy = 0.60060
Iteration 9, source accuracy = 0.61284, target accuracy = 0.73282
14 [2.9684834, 0.742146, 1.1131687, 0.484375, 0.484375]
Iteration 14, source accuracy = 0.49447, target accuracy = 0.57789
Iteration 14, source accuracy = 0.58395, target accuracy = 0.70331
19 [2.8132186, 0.71812755, 1.0475456, 0.515625, 0.5]
Iteration 19, source accuracy = 0.51497, target accuracy = 0.44724
Iteration 19, source accuracy = 0.44337, target accuracy = 0.45480
24 [2.813531, 0.70227885, 1.055626, 0.515625, 0.5]
Iteration 24, source accuracy = 0.50935, target accuracy = 0.41146
Iteration 24, source accuracy = 0.36216, target accuracy = 0.33151
29 [2.733358, 0.6810949, 1.0261315, 0.46875, 0.453125]
Iteration 29, source accuracy = 0.50412, target accuracy = 0.34995
Iteration 29, source accuracy = 0.07568, target accuracy = 0.05824
34 [2.3665345, 0.74769557, 0.80941945, 0.4375, 0.609375]
Iteration 34, source accuracy = 0.50372, target accuracy = 0.34673
Iteration 34, source accuracy = 0.04561, target accuracy = 0.03446
39 [3.15806, 0.6748207, 1.2416197, 0.5, 0.5]
Iteration 39, source accuracy = 0.50251, target accuracy = 0.34633
Iteration 39, source accuracy = 0.03883, target accuracy = 0.03272
44 [2.9966292, 0.71058726, 1.143021, 0.53125, 0.421875]
Iteration 44, source accuracy = 0.50392, target accuracy = 0.35899
Iteration 44, source accuracy = 0.10255, target accuracy = 0.09326
49 [2.9245906, 0.7043394, 1.1101257, 0.515625, 0.484375]
Iteration 49, source accuracy = 0.52482, target accuracy = 0.44905
Iteration 49, source accuracy = 0.43607, target accuracy = 0.43705
AlexNet -> 5k Dataset (Batch size-32) Trained from scratch -> 3.8M params
4 [3.0169013, 1.0092723, 1.0038145, 0.546875, 0.5]
Iteration 4, source accuracy = 0.50452, target accuracy = 0.35075
Iteration 4, source f1_score = 0.08534, target f1_score = 0.05666
9 [2.8931608, 0.71535146, 1.0889047, 0.53125, 0.453125]
Iteration 9, source accuracy = 0.49809, target accuracy = 0.34372
Iteration 9, source f1_score = 0.00873, target f1_score = 0.01091
14 [2.5568414, 0.7806059, 0.88811773, 0.5625, 0.421875]
Iteration 14, source accuracy = 0.49769, target accuracy = 0.34251
Iteration 14, source f1_score = 0.00160, target f1_score = 0.00426
19 [2.3683963, 0.7297066, 0.8193449, 0.578125, 0.59375]
Iteration 19, source accuracy = 0.49889, target accuracy = 0.34593
Iteration 19, source f1_score = 0.02579, target f1_score = 0.02750
24 [2.3989587, 0.7431066, 0.82792604, 0.578125, 0.5625]
Iteration 24, source accuracy = 0.49467, target accuracy = 0.43015
Iteration 24, source f1_score = 0.41179, target f1_score = 0.41583
29 [2.2093544, 0.71321654, 0.7480689, 0.546875, 0.5625]
Iteration 29, source accuracy = 0.49327, target accuracy = 0.49849
Iteration 29, source f1_score = 0.55857, target f1_score = 0.58187
34 [2.3262138, 0.7180115, 0.8041011, 0.5, 0.59375]
Iteration 34, source accuracy = 0.49990, target accuracy = 0.51839
Iteration 34, source f1_score = 0.55666, target f1_score = 0.61317
39 [2.3656201, 0.76565874, 0.79998064, 0.5, 0.5]
Iteration 39, source accuracy = 0.50372, target accuracy = 0.39940
Iteration 39, source f1_score = 0.32077, target f1_score = 0.30865
44 [2.10317, 0.6621543, 0.72050774, 0.59375, 0.5625]
Iteration 44, source accuracy = 0.50915, target accuracy = 0.40724
Iteration 44, source f1_score = 0.35225, target f1_score = 0.34189
49 [2.493608, 0.7033156, 0.8951462, 0.578125, 0.46875]
Iteration 49, source accuracy = 0.51417, target accuracy = 0.39638
Iteration 49, source f1_score = 0.34231, target f1_score = 0.30373
AlexNet -> 10k Dataset (Batch size-32) Trained from scratch -> 3.8M params
4 [3.1927335, 0.84065795, 1.1760378, 0.59375, 0.5]
Iteration 4, source accuracy = 0.50762, target accuracy = 0.40563
Iteration 4, source f1_score = 0.23351, target f1_score = 0.28592
9 [3.139401, 0.9696721, 1.0848644, 0.53125, 0.53125]
Iteration 9, source accuracy = 0.50650, target accuracy = 0.35216
Iteration 9, source f1_score = 0.05005, target f1_score = 0.05401
14 [2.6046176, 0.8374362, 0.8835907, 0.59375, 0.515625]
Iteration 14, source accuracy = 0.48050, target accuracy = 0.58030
Iteration 14, source f1_score = 0.56436, target f1_score = 0.70060
19 [2.9338522, 0.8687864, 1.0325329, 0.4375, 0.53125]
Iteration 19, source accuracy = 0.48525, target accuracy = 0.54472
Iteration 19, source f1_score = 0.53247, target f1_score = 0.65073
24 [2.4930813, 0.5741277, 0.9594768, 0.625, 0.578125]
Iteration 24, source accuracy = 0.51675, target accuracy = 0.36201
Iteration 24, source f1_score = 0.17216, target f1_score = 0.11980
29 [2.3124352, 0.7415949, 0.78542006, 0.578125, 0.609375]
Iteration 29, source accuracy = 0.52238, target accuracy = 0.36503
Iteration 29, source f1_score = 0.18267, target f1_score = 0.12420
34 [2.117728, 0.72548985, 0.69611907, 0.625, 0.609375]
Iteration 34, source accuracy = 0.52963, target accuracy = 0.42754
Iteration 34, source f1_score = 0.44897, target f1_score = 0.39456
39 [2.7769504, 0.7057755, 1.0355874, 0.453125, 0.53125]
Iteration 39, source accuracy = 0.52963, target accuracy = 0.40844
Iteration 39, source f1_score = 0.38503, target f1_score = 0.31606
44 [2.3798842, 0.64256966, 0.8686573, 0.515625, 0.59375]
Iteration 44, source accuracy = 0.53163, target accuracy = 0.46050
Iteration 44, source f1_score = 0.42487, target f1_score = 0.46661
49 [2.547595, 0.73210806, 0.90774345, 0.5, 0.53125]
Iteration 49, source accuracy = 0.54300, target accuracy = 0.51518
Iteration 49, source f1_score = 0.55813, target f1_score = 0.59489
AlexNet -> 10k Dataset (Batch size-64) Trained from scratch -> 3.8M params
4 [3.2197187, 0.858701, 1.1805089, 0.5, 0.5234375]
Iteration 4, source accuracy = 0.51825, target accuracy = 0.35397
Iteration 4, source f1_score = 0.10911, target f1_score = 0.05858
9 [3.5522282, 0.73549604, 1.4083661, 0.4765625, 0.390625]
Iteration 9, source accuracy = 0.51050, target accuracy = 0.35779
Iteration 9, source f1_score = 0.15165, target f1_score = 0.09924
14 [2.6106937, 0.6950207, 0.9578365, 0.546875, 0.53125]
Iteration 14, source accuracy = 0.49562, target accuracy = 0.59437
Iteration 14, source f1_score = 0.61339, target f1_score = 0.71665
19 [2.531765, 0.68875736, 0.92150384, 0.53125, 0.5390625]
Iteration 19, source accuracy = 0.50187, target accuracy = 0.65206
Iteration 19, source f1_score = 0.66510, target f1_score = 0.78784
24 [2.5543988, 0.7356261, 0.90938634, 0.5234375, 0.5625]
Iteration 24, source accuracy = 0.49462, target accuracy = 0.64683
Iteration 24, source f1_score = 0.65665, target f1_score = 0.78274
29 [2.716497, 0.65394586, 1.0312755, 0.4921875, 0.5]
Iteration 29, source accuracy = 0.49575, target accuracy = 0.64563
Iteration 29, source f1_score = 0.65738, target f1_score = 0.78183
34 [2.4289496, 0.70255804, 0.8631958, 0.4921875, 0.53125]
Iteration 34, source accuracy = 0.49662, target accuracy = 0.59196
Iteration 34, source f1_score = 0.61615, target f1_score = 0.71985
39 [2.5893495, 0.7101801, 0.93958473, 0.5625, 0.46875]
Iteration 39, source accuracy = 0.49862, target accuracy = 0.61166
Iteration 39, source f1_score = 0.62580, target f1_score = 0.74060
44 [2.4386559, 0.6502918, 0.89418197, 0.5703125, 0.53125]
Iteration 44, source accuracy = 0.49888, target accuracy = 0.63538
Iteration 44, source f1_score = 0.65353, target f1_score = 0.76962
49 [2.1989832, 0.67174494, 0.7636192, 0.5546875, 0.578125]
Iteration 49, source accuracy = 0.50787, target accuracy = 0.58111
Iteration 49, source f1_score = 0.63448, target f1_score = 0.70706
AlexNet -> 5k Dataset (Batch size-64) Trained from scratch -> 3.8M params
4 [2.557248, 0.6361023, 0.96057296, 0.609375, 0.53125]
Iteration 4, source accuracy = 0.50894, target accuracy = 0.41528
Iteration 4, source f1_score = 0.38355, target f1_score = 0.36387
9 [2.2657123, 0.68282217, 0.7914451, 0.4921875, 0.625]
Iteration 9, source accuracy = 0.48482, target accuracy = 0.45166
Iteration 9, source f1_score = 0.42726, target f1_score = 0.45809
14 [2.1814475, 0.769445, 0.7060013, 0.484375, 0.6015625]
Iteration 14, source accuracy = 0.52462, target accuracy = 0.39819
Iteration 14, source f1_score = 0.36236, target f1_score = 0.28476
19 [2.2567937, 0.73718774, 0.759803, 0.484375, 0.5625]
Iteration 19, source accuracy = 0.50593, target accuracy = 0.36643
Iteration 19, source f1_score = 0.14176, target f1_score = 0.13596
24 [2.460589, 0.78210294, 0.83924294, 0.46875, 0.5625]
Iteration 24, source accuracy = 0.50251, target accuracy = 0.35558
Iteration 24, source f1_score = 0.10488, target f1_score = 0.08817
29 [2.4212527, 0.7427023, 0.83927524, 0.4921875, 0.578125]
Iteration 29, source accuracy = 0.51337, target accuracy = 0.38834
Iteration 29, source f1_score = 0.19701, target f1_score = 0.22353
34 [2.3099642, 0.7033905, 0.80328685, 0.5234375, 0.5546875]
Iteration 34, source accuracy = 0.52563, target accuracy = 0.50794
Iteration 34, source f1_score = 0.47509, target f1_score = 0.56886
39 [2.2287946, 0.6802105, 0.7742921, 0.515625, 0.609375]
Iteration 39, source accuracy = 0.51678, target accuracy = 0.54714
Iteration 39, source f1_score = 0.52752, target f1_score = 0.64004
44 [2.4446526, 0.698408, 0.87312233, 0.484375, 0.5078125]
Iteration 44, source accuracy = 0.51437, target accuracy = 0.53387
Iteration 44, source f1_score = 0.50714, target f1_score = 0.61902
49 [1.9782839, 0.6412079, 0.668538, 0.5078125, 0.6171875]
Iteration 49, source accuracy = 0.52462, target accuracy = 0.53447
Iteration 49, source f1_score = 0.50179, target f1_score = 0.61630
AlexNet -> 11k Dataset (Batch size-64) Trained from scratch -> 3.8M params
4 [2.8109837, 0.8267522, 0.9921157, 0.5078125, 0.5546875]
Iteration 4, source accuracy = 0.54191, target accuracy = 0.65528
Iteration 4, source f1_score = 0.70182, target f1_score = 0.79108
9 [2.8295593, 0.8161861, 1.0066867, 0.46875, 0.5390625]
Iteration 9, source accuracy = 0.45582, target accuracy = 0.34372
Iteration 9, source f1_score = 0.00730, target f1_score = 0.00850
14 [2.4213603, 0.7276465, 0.84685683, 0.453125, 0.59375]
Iteration 14, source accuracy = 0.45709, target accuracy = 0.34312
Iteration 14, source f1_score = 0.01485, target f1_score = 0.00910
19 [2.4627109, 0.7339592, 0.8643759, 0.46875, 0.515625]
Iteration 19, source accuracy = 0.50827, target accuracy = 0.56543
Iteration 19, source f1_score = 0.62320, target f1_score = 0.68484
24 [2.71882, 0.7388731, 0.98997355, 0.3671875, 0.5078125]
Iteration 24, source accuracy = 0.49300, target accuracy = 0.51477
Iteration 24, source f1_score = 0.54440, target f1_score = 0.59834
29 [2.3299608, 0.70821404, 0.8108733, 0.4453125, 0.5703125]
Iteration 29, source accuracy = 0.48082, target accuracy = 0.37749
Iteration 29, source f1_score = 0.17673, target f1_score = 0.18350
34 [2.4033937, 0.70498466, 0.8492045, 0.421875, 0.5546875]
Iteration 34, source accuracy = 0.51191, target accuracy = 0.43518
Iteration 34, source f1_score = 0.40916, target f1_score = 0.40792
39 [1.9721886, 0.7557534, 0.6082176, 0.40625, 0.6328125]
Iteration 39, source accuracy = 0.53945, target accuracy = 0.62472
Iteration 39, source f1_score = 0.67953, target f1_score = 0.76098
44 [2.193553, 0.6960177, 0.7487677, 0.4453125, 0.6171875]
Iteration 44, source accuracy = 0.54245, target accuracy = 0.65347
Iteration 44, source f1_score = 0.70129, target f1_score = 0.78909
49 [2.223763, 0.6845157, 0.7696236, 0.390625, 0.5703125]
Iteration 49, source accuracy = 0.53973, target accuracy = 0.64884
Iteration 49, source f1_score = 0.69735, target f1_score = 0.78467