-
Notifications
You must be signed in to change notification settings - Fork 0
/
resnet_final_backbone.out
261 lines (257 loc) · 9.61 KB
/
resnet_final_backbone.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
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
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
Downloading: "https://download.pytorch.org/models/resnet50-11ad3fa6.pth" to /home/users/xw214/.cache/torch/hub/checkpoints/resnet50-11ad3fa6.pth
DATASET:
GENETIC:
FIX_PROTOTYPES: True
MODEL_PATH: NA
ORDER_NAME: NA
PROTOTYPE_SHAPE: (0, 0, 0, 0)
SIZE: 0
TAXONOMY_NAME: NA
TRAIN_PATH: NA
TRAIN_PUSH_DIR: NA
TRANSFORM: onehot
VALIDATION_PATH: NA
IMAGE:
MODEL_PATH: NA
PROTOTYPE_SHAPE: (0, 0, 0, 0)
SIZE: 256
TEST_DIR: datasets/bioscan/small_diptera_test
TRAIN_BATCH_SIZE: 80
TRAIN_DIR: datasets/bioscan/small_diptera_train
TRAIN_PUSH_DIR: datasets/bioscan/small_diptera_train
TRANSFORM_MEAN: (0.485, 0.456, 0.406)
TRANSFORM_STD: (0.229, 0.224, 0.225)
NAME: bioscan
NUM_CLASSES: 40
TEST_BATCH_SIZE: 100
TRAIN_BATCH_SIZE: 80
TRAIN_PUSH_BATCH_SIZE: 75
EXPERIMENT_RUN: 1
MODEL:
BACKBONE: resnet50
DEVICE: cuda
GENETIC_MODE: False
PROTOTYPE_ACTIVATION_FUNCTION: linear
PROTOTYPE_DISTANCE_FUNCTION: cosine
PROTOTYPE_SHAPE: (400, 128, 1, 1)
OPTIM:
COEFS:
CLST: 0.8
CRS_ENT: 1
L1: 0.0001
SEP: -0.08
JOINT_OPTIMIZER_LAYERS:
ADD_ON_LAYERS: 0.003
FEATURES: 0.0001
LR_STEP_SIZE: 5
PROTOTYPE_VECTORS: 0.003
WEIGHT_DECAY: 0.001
LAST_LAYER_OPTIMIZER_LAYERS:
LR: 0.0001
NUM_TRAIN_EPOCHS: 100
NUM_WARM_EPOCHS: 5
PUSH_EPOCHS: [0, 10, 20, 30, 40, 50, 60, 70, 80, 90]
PUSH_START: 10
WARM_OPTIMIZER_LAYERS:
ADD_ON_LAYERS: 0.003
PROTOTYPE_VECTORS: 0.003
WEIGHT_DECAY: 0.001
OUTPUT:
IMG_DIR: ./saved_models/bioscan/img
MODEL_DIR: ./saved_models
NO_SAVE: False
PREPROCESS_INPUT_FUNCTION: None
PROTOTYPE_IMG_FILENAME_PREFIX: prototype-img
PROTOTYPE_SELF_ACT_FILENAME_PREFIX: prototype-self-act
PROTO_BOUND_BOXES_FILENAME_PREFIX: bb
WEIGHT_MATRIX_FILENAME: outputL_weights
Loading model...
0%| | 0.00/97.8M [00:00<?, ?B/s] 2%|▏ | 2.01M/97.8M [00:00<00:04, 21.0MB/s] 13%|█▎ | 12.4M/97.8M [00:00<00:01, 72.9MB/s] 27%|██▋ | 26.3M/97.8M [00:00<00:00, 106MB/s] 41%|████ | 40.1M/97.8M [00:00<00:00, 121MB/s] 55%|█████▌ | 54.1M/97.8M [00:00<00:00, 130MB/s] 70%|██████▉ | 68.0M/97.8M [00:00<00:00, 136MB/s] 84%|████████▍ | 82.0M/97.8M [00:00<00:00, 139MB/s]100%|██████████| 97.8M/97.8M [00:00<00:00, 130MB/s]Training
Training Epoch 1/60, Loss: 1.0925797743797303
Testing Epoch 1/60, Loss: 0.9605451139062643
Accuracy on test set: 73.42%
Training Epoch 2/60, Loss: 0.5190008872577122
Testing Epoch 2/60, Loss: 0.9445883677899838
Accuracy on test set: 73.88%
Training Epoch 3/60, Loss: 0.35568031222479685
Testing Epoch 3/60, Loss: 0.7987780440598726
Accuracy on test set: 79.27%
Training Epoch 4/60, Loss: 0.26265997757230486
Testing Epoch 4/60, Loss: 0.7388219505175948
Accuracy on test set: 81.17%
Training Epoch 5/60, Loss: 0.18971972622190203
Testing Epoch 5/60, Loss: 0.6978567501250654
Accuracy on test set: 83.40%
Training Epoch 6/60, Loss: 0.14525631090147154
Testing Epoch 6/60, Loss: 1.3263367290468886
Accuracy on test set: 70.75%
Training Epoch 7/60, Loss: 0.11710710962116719
Testing Epoch 7/60, Loss: 1.0200660972390323
Accuracy on test set: 77.68%
Training Epoch 8/60, Loss: 0.09560253548515694
Testing Epoch 8/60, Loss: 0.8451841863431037
Accuracy on test set: 80.87%
Training Epoch 9/60, Loss: 0.08512119439989328
Testing Epoch 9/60, Loss: 0.8250278800865636
Accuracy on test set: 82.14%
Training Epoch 10/60, Loss: 0.07444656439551285
Testing Epoch 10/60, Loss: 1.3081400532228873
Accuracy on test set: 74.06%
Training Epoch 11/60, Loss: 0.06746543790187154
Testing Epoch 11/60, Loss: 4.348182715638541
Accuracy on test set: 53.33%
Training Epoch 12/60, Loss: 0.06072451831826142
Testing Epoch 12/60, Loss: 1.0269174939393997
Accuracy on test set: 80.08%
Training Epoch 13/60, Loss: 0.055311782114207744
Testing Epoch 13/60, Loss: 1.4298809592425823
Accuracy on test set: 76.80%
Training Epoch 14/60, Loss: 0.05377948686799833
Testing Epoch 14/60, Loss: 0.8090560851991176
Accuracy on test set: 84.29%
Training Epoch 15/60, Loss: 0.05135117906837591
Testing Epoch 15/60, Loss: 1.227831965108053
Accuracy on test set: 77.15%
Training Epoch 16/60, Loss: 0.04330245834615614
Testing Epoch 16/60, Loss: 0.8630528219975531
Accuracy on test set: 84.31%
Training Epoch 17/60, Loss: 0.04056611268182418
Testing Epoch 17/60, Loss: 0.8127420899271965
Accuracy on test set: 85.10%
Training Epoch 18/60, Loss: 0.04255797993178879
Testing Epoch 18/60, Loss: 0.8854627164825797
Accuracy on test set: 84.07%
Training Epoch 19/60, Loss: 0.03896623794202294
Testing Epoch 19/60, Loss: 0.8591565826162696
Accuracy on test set: 84.91%
Training Epoch 20/60, Loss: 0.03725728525035083
Testing Epoch 20/60, Loss: 0.9370030312938615
Accuracy on test set: 83.23%
Training Epoch 21/60, Loss: 0.038162701729152884
Testing Epoch 21/60, Loss: 0.7420697955414652
Accuracy on test set: 87.52%
Training Epoch 22/60, Loss: 0.025659670997743626
Testing Epoch 22/60, Loss: 0.8416399205476046
Accuracy on test set: 84.38%
Training Epoch 23/60, Loss: 0.037979630995009624
Testing Epoch 23/60, Loss: 0.762481302395463
Accuracy on test set: 86.21%
Training Epoch 24/60, Loss: 0.024698503512356963
Testing Epoch 24/60, Loss: 0.8586861792951822
Accuracy on test set: 85.66%
Training Epoch 25/60, Loss: 0.038715796427269064
Testing Epoch 25/60, Loss: 0.9577198899909853
Accuracy on test set: 82.54%
Training Epoch 26/60, Loss: 0.02134975399848606
Testing Epoch 26/60, Loss: 0.9460262135788798
Accuracy on test set: 83.36%
Training Epoch 27/60, Loss: 0.024934167585801335
Testing Epoch 27/60, Loss: 0.868682396877557
Accuracy on test set: 85.56%
Training Epoch 28/60, Loss: 0.024801321939033057
Testing Epoch 28/60, Loss: 0.835393206551671
Accuracy on test set: 85.08%
Training Epoch 29/60, Loss: 0.008436412986600772
Testing Epoch 29/60, Loss: 0.6721358523424715
Accuracy on test set: 89.45%
Training Epoch 30/60, Loss: 0.0013190824306137593
Testing Epoch 30/60, Loss: 0.6814174854755402
Accuracy on test set: 89.63%
Training Epoch 31/60, Loss: 0.0007565087374283134
Testing Epoch 31/60, Loss: 0.6854107189271599
Accuracy on test set: 89.75%
Training Epoch 32/60, Loss: 0.0005207775808272085
Testing Epoch 32/60, Loss: 0.6901055780611932
Accuracy on test set: 89.95%
Training Epoch 33/60, Loss: 0.0003841524904128164
Testing Epoch 33/60, Loss: 0.7003271477064118
Accuracy on test set: 89.99%
Training Epoch 34/60, Loss: 0.00028560954909439065
Testing Epoch 34/60, Loss: 0.706322756940499
Accuracy on test set: 90.13%
Training Epoch 35/60, Loss: 0.00023303520223791045
Testing Epoch 35/60, Loss: 0.7168659665901214
Accuracy on test set: 90.13%
Training Epoch 36/60, Loss: 0.00017214724021115605
Testing Epoch 36/60, Loss: 0.7204502664040774
Accuracy on test set: 90.29%
Training Epoch 37/60, Loss: 0.00014059847034513952
Testing Epoch 37/60, Loss: 0.7289565366879106
Accuracy on test set: 90.26%
Training Epoch 38/60, Loss: 0.00011259972502427575
Testing Epoch 38/60, Loss: 0.7354567777365446
Accuracy on test set: 90.35%
Training Epoch 39/60, Loss: 9.374311582983604e-05
Testing Epoch 39/60, Loss: 0.7465319980494678
Accuracy on test set: 90.29%
Training Epoch 40/60, Loss: 7.716225013947613e-05
Testing Epoch 40/60, Loss: 0.7522581665962934
Accuracy on test set: 90.45%
Training Epoch 41/60, Loss: 6.34678963355587e-05
Testing Epoch 41/60, Loss: 0.7576911328267306
Accuracy on test set: 90.45%
Training Epoch 42/60, Loss: 5.674715381714382e-05
Testing Epoch 42/60, Loss: 0.7646527234185487
Accuracy on test set: 90.53%
Training Epoch 43/60, Loss: 4.067511531996258e-05
Testing Epoch 43/60, Loss: 0.7762203330593184
Accuracy on test set: 90.50%
Training Epoch 44/60, Loss: 3.284541432679232e-05
Testing Epoch 44/60, Loss: 0.7878314607357606
Accuracy on test set: 90.61%
Training Epoch 45/60, Loss: 2.977662056920443e-05
Testing Epoch 45/60, Loss: 0.800885678208433
Accuracy on test set: 90.47%
Training Epoch 46/60, Loss: 2.326449757622738e-05
Testing Epoch 46/60, Loss: 0.8052733086142689
Accuracy on test set: 90.59%
Training Epoch 47/60, Loss: 1.6578483046292344e-05
Testing Epoch 47/60, Loss: 0.819636245239526
Accuracy on test set: 90.56%
Training Epoch 48/60, Loss: 1.6985726189886918e-05
Testing Epoch 48/60, Loss: 0.8257575530279427
Accuracy on test set: 90.61%
Training Epoch 49/60, Loss: 1.2318456965918553e-05
Testing Epoch 49/60, Loss: 0.8409090590663254
Accuracy on test set: 90.73%
Training Epoch 50/60, Loss: 1.1472151477002106e-05
Testing Epoch 50/60, Loss: 0.8505250875791535
Accuracy on test set: 90.65%
Training Epoch 51/60, Loss: 7.90611877151345e-06
Testing Epoch 51/60, Loss: 0.8569674004986882
Accuracy on test set: 90.71%
Training Epoch 52/60, Loss: 6.569703901342499e-06
Testing Epoch 52/60, Loss: 0.8656564135057851
Accuracy on test set: 90.63%
Training Epoch 53/60, Loss: 5.468437799793589e-06
Testing Epoch 53/60, Loss: 0.8680883020255714
Accuracy on test set: 90.72%
Training Epoch 54/60, Loss: 4.0830505480958635e-06
Testing Epoch 54/60, Loss: 0.8809984231507406
Accuracy on test set: 90.76%
Training Epoch 55/60, Loss: 3.7207747784902625e-06
Testing Epoch 55/60, Loss: 0.8885905058821664
Accuracy on test set: 90.86%
Training Epoch 56/60, Loss: 0.0002462456354279701
Testing Epoch 56/60, Loss: 0.9651767870981712
Accuracy on test set: 89.38%
Training Epoch 57/60, Loss: 0.0004998449203236045
Testing Epoch 57/60, Loss: 0.9037833582569147
Accuracy on test set: 89.82%
Training Epoch 58/60, Loss: 0.00016093621356802552
Testing Epoch 58/60, Loss: 0.89388392494875
Accuracy on test set: 89.83%
Training Epoch 59/60, Loss: 7.161673953383537e-05
Testing Epoch 59/60, Loss: 0.8897182567208074
Accuracy on test set: 89.89%
Training Epoch 60/60, Loss: 9.050435651027199e-05
Testing Epoch 60/60, Loss: 0.8966240226046648
Accuracy on test set: 89.89%