text
stringlengths
5
1.13k
2020-01-31 18:39:17 Iteration 3900 Training Loss: 8.945e-02 Loss in Target Net: 3.936e-02
2020-01-31 18:39:40 Iteration 3950 Training Loss: 8.250e-02 Loss in Target Net: 4.881e-02
2020-01-31 18:40:02 Iteration 3999 Training Loss: 8.367e-02 Loss in Target Net: 2.769e-02
Evaluating against victims networks
DPN92
Using Adam for retraining
Files already downloaded and verified
2020-01-31 18:40:06, Epoch 0, Iteration 7, loss 1.848 (4.338), acc 84.615 (64.400)
2020-01-31 18:40:06, Epoch 30, Iteration 7, loss 0.003 (0.115), acc 100.000 (98.000)
Target Label: 6, Poison label: 8, Prediction:8, Target's Score:[8.965381, -31.588665, -39.19313, -5.531065, -31.50348, -5.137039, 16.251516, -18.656511, 26.428278, -103.50076], Poisons' Predictions:[8, 8, 8, 8, 8]
2020-01-31 18:40:10 Epoch 59, Val iteration 0, acc 90.800 (90.800)
2020-01-31 18:40:17 Epoch 59, Val iteration 19, acc 92.600 (92.160)
* Prec: 92.16000213623047
--------
SENet18
Using Adam for retraining
Files already downloaded and verified
2020-01-31 18:40:19, Epoch 0, Iteration 7, loss 0.460 (0.708), acc 92.308 (87.200)
2020-01-31 18:40:20, Epoch 30, Iteration 7, loss 0.174 (0.141), acc 94.231 (96.800)
Target Label: 6, Poison label: 8, Prediction:8, Target's Score:[1.5276494, 6.638004, -6.1535964, -3.3861344, -0.06253219, -8.8144455, 5.8296404, -2.9020817, 16.462654, -7.039203], Poisons' Predictions:[8, 8, 8, 8, 8]
2020-01-31 18:40:21 Epoch 59, Val iteration 0, acc 92.200 (92.200)
2020-01-31 18:40:22 Epoch 59, Val iteration 19, acc 92.200 (91.410)
* Prec: 91.41000175476074
--------
ResNet50
Using Adam for retraining
Files already downloaded and verified
2020-01-31 18:40:25, Epoch 0, Iteration 7, loss 0.000 (1.375), acc 100.000 (85.600)
2020-01-31 18:40:25, Epoch 30, Iteration 7, loss 0.000 (0.000), acc 100.000 (100.000)
Target Label: 6, Poison label: 8, Prediction:6, Target's Score:[-31.408695, -39.55227, -23.077105, -47.83474, -61.76877, -62.239273, 11.453058, -68.69819, 11.30381, -46.83127], Poisons' Predictions:[8, 8, 8, 8, 8]
2020-01-31 18:40:26 Epoch 59, Val iteration 0, acc 93.000 (93.000)
2020-01-31 18:40:30 Epoch 59, Val iteration 19, acc 93.400 (93.870)
* Prec: 93.87000122070313
--------
ResNeXt29_2x64d
Using Adam for retraining
Files already downloaded and verified
2020-01-31 18:40:32, Epoch 0, Iteration 7, loss 0.951 (1.588), acc 90.385 (76.000)
2020-01-31 18:40:33, Epoch 30, Iteration 7, loss 0.000 (0.019), acc 100.000 (99.200)
Target Label: 6, Poison label: 8, Prediction:6, Target's Score:[-20.879446, -2.4811068, -14.2641535, -0.08066155, -96.65206, -47.325542, 25.544876, -19.458012, 18.20676, -25.832617], Poisons' Predictions:[8, 8, 8, 8, 8]
2020-01-31 18:40:34 Epoch 59, Val iteration 0, acc 92.600 (92.600)
2020-01-31 18:40:38 Epoch 59, Val iteration 19, acc 92.800 (92.900)
* Prec: 92.90000190734864
--------
GoogLeNet
Using Adam for retraining
Files already downloaded and verified
2020-01-31 18:40:41, Epoch 0, Iteration 7, loss 0.365 (0.450), acc 88.462 (89.400)
2020-01-31 18:40:41, Epoch 30, Iteration 7, loss 0.062 (0.025), acc 96.154 (98.800)
Target Label: 6, Poison label: 8, Prediction:6, Target's Score:[-12.229371, -9.712031, -10.603616, -3.0965168, -12.447538, -6.626693, 8.54154, -14.254765, 6.133732, -14.402716], Poisons' Predictions:[8, 8, 8, 8, 8]
2020-01-31 18:40:43 Epoch 59, Val iteration 0, acc 91.200 (91.200)
2020-01-31 18:40:48 Epoch 59, Val iteration 19, acc 91.800 (92.170)
* Prec: 92.17000198364258
--------
MobileNetV2
Using Adam for retraining
Files already downloaded and verified
2020-01-31 18:40:50, Epoch 0, Iteration 7, loss 0.882 (3.014), acc 86.538 (63.800)
2020-01-31 18:40:50, Epoch 30, Iteration 7, loss 0.264 (0.238), acc 92.308 (93.600)
Target Label: 6, Poison label: 8, Prediction:8, Target's Score:[6.918476, -20.848701, -6.780697, 1.1666086, -48.726692, -10.1453, 7.44598, -27.130976, 10.538126, -23.500862], Poisons' Predictions:[8, 8, 8, 8, 8]
2020-01-31 18:40:51 Epoch 59, Val iteration 0, acc 88.800 (88.800)
2020-01-31 18:40:53 Epoch 59, Val iteration 19, acc 87.400 (87.300)
* Prec: 87.30000152587891
--------
ResNet18
Using Adam for retraining
Files already downloaded and verified
2020-01-31 18:40:55, Epoch 0, Iteration 7, loss 0.542 (0.598), acc 90.385 (88.200)
2020-01-31 18:40:56, Epoch 30, Iteration 7, loss 0.009 (0.022), acc 100.000 (99.400)
Target Label: 6, Poison label: 8, Prediction:8, Target's Score:[-33.957993, -3.2979393, -18.71028, -1.9121279, -39.042255, -7.7329454, 0.5100708, -8.614806, 6.202431, -22.010166], Poisons' Predictions:[8, 8, 8, 8, 8]
2020-01-31 18:40:56 Epoch 59, Val iteration 0, acc 93.400 (93.400)
2020-01-31 18:40:58 Epoch 59, Val iteration 19, acc 93.200 (92.740)
* Prec: 92.7400016784668
--------
DenseNet121
Using Adam for retraining
Files already downloaded and verified
2020-01-31 18:41:01, Epoch 0, Iteration 7, loss 0.215 (0.384), acc 90.385 (93.200)
2020-01-31 18:41:01, Epoch 30, Iteration 7, loss 0.001 (0.002), acc 100.000 (100.000)
Target Label: 6, Poison label: 8, Prediction:8, Target's Score:[-7.0874248, -13.00754, -13.315794, -6.0484023, -7.4859576, -9.703103, 1.6638383, -31.248932, 6.022441, -14.527675], Poisons' Predictions:[8, 8, 8, 8, 8]
2020-01-31 18:41:03 Epoch 59, Val iteration 0, acc 93.200 (93.200)
2020-01-31 18:41:07 Epoch 59, Val iteration 19, acc 93.400 (92.980)
* Prec: 92.98000106811523
--------
------SUMMARY------
TIME ELAPSED (mins): 29
TARGET INDEX: 11
DPN92 1
SENet18 1
ResNet50 0
ResNeXt29_2x64d 0
GoogLeNet 0
MobileNetV2 1
ResNet18 1
DenseNet121 1
Namespace(chk_path='chk-black-ourmean/', chk_subdir='poisons', device='cuda', dset_path='datasets', end2end=False, eval_poison_path='', gpu='0', lr_decay_epoch=[30, 45], mode='mean', model_resume_path='model-chks', nearest=False, net_repeat=1, num_per_class=50, original_grad=True, poison_decay_ites=[], poison_decay_ratio=0.1, poison_epsilon=0.1, poison_ites=4000, poison_label=8, poison_lr=0.04, poison_momentum=0.9, poison_num=5, poison_opt='adam', resume_poison_ite=0, retrain_bsize=64, retrain_epochs=60, retrain_lr=0.1, retrain_momentum=0.9, retrain_opt='adam', retrain_wd=0, subs_chk_name=['ckpt-%s-4800-dp0.200-droplayer0.000-seed1226.t7', 'ckpt-%s-4800-dp0.250-droplayer0.000-seed1226.t7', 'ckpt-%s-4800-dp0.300-droplayer0.000.t7'], subs_dp=[0.2, 0.25, 0.3], subset_group=0, substitute_nets=['DPN92', 'SENet18', 'ResNet50', 'ResNeXt29_2x64d', 'GoogLeNet', 'MobileNetV2'], target_index=12, target_label=6, target_net=['DPN92', 'SENet18', 'ResNet50', 'ResNeXt29_2x64d', 'GoogLeNet', 'MobileNetV2', 'ResNet18', 'DenseNet121'], test_chk_name='ckpt-%s-4800.t7', tol=1e-06, train_data_path='datasets/CIFAR10_TRAIN_Split.pth')
Path: chk-black-ourmean/mean/4000/12
Selected base image indices: [213, 225, 227, 247, 249]
2020-01-31 18:46:46 Iteration 0 Training Loss: 1.084e+00 Loss in Target Net: 4.226e-01
2020-01-31 18:47:08 Iteration 50 Training Loss: 1.217e-01 Loss in Target Net: 7.700e-03